 |
python_feedgen09_jnboehm.com.atom.xml - sfeed_tests - sfeed tests and RSS and A… |
 |
git clone git://git.codemadness.org/sfeed_tests |
 |
Log |
|
Files |
|
Refs |
|
README |
|
LICENSE |
|
--- |
|
python_feedgen09_jnboehm.com.atom.xml (142453B) |
|
--- |
|
1 <?xml version='1.0' encoding='UTF-8'?> |
|
2 <feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en"><id>http://arxiv… |
|
3 networks for time series (TS) forecasting on a large set of time series. |
|
4 Current state-of-the-art deep ensemble models have high memory and |
|
5 computational requirements, hampering their use to forecast millions of … |
|
6 practical scenarios. We propose N-BEATS(P), a global multivariate varian… |
|
7 the N-BEATS model designed to allow simultaneous training of multiple |
|
8 univariate TS forecasting models. Our model addresses the practical limi… |
|
9 of related models, reducing the training time by half and memory require… |
|
10 a factor of 5, while keeping the same level of accuracy. We have perform… |
|
11 multiple experiments detailing the various ways to train our model and h… |
|
12 obtained results that demonstrate its capacity to support zero-shot TS |
|
13 forecasting, i.e., to train a neural network on a source TS dataset and … |
|
14 it on a different target TS dataset without retraining, which provides an |
|
15 efficient and reliable solution to forecast at scale even in difficult |
|
16 forecasting conditions. |
|
17 </summary></entry><entry><id>http://arxiv.org/abs/2109.02624</id><title>… |
|
18 and -- for shapes -- also scale, we extend generalized additive regressi… |
|
19 models for the shape/form of planar curves or landmark configurations. T… |
|
20 model respects the resulting quotient geometry of the response, employin… |
|
21 squared geodesic distance as loss function and a geodesic response funct… |
|
22 mapping the additive predictor to the shape/form space. For fitting the … |
|
23 we propose a Riemannian $L_2$-Boosting algorithm well-suited for a poten… |
|
24 large number of possibly parameter-intensive model terms, which also yie… |
|
25 automated model selection. We provide novel intuitively interpretable |
|
26 visualizations for (even non-linear) covariate effects in the shape/form… |
|
27 via suitable tensor based factorizations. The usefulness of the proposed |
|
28 framework is illustrated in an analysis of 1) astragalus shapes of wild … |
|
29 domesticated sheep and 2) cell forms generated in a biophysical model, a… |
|
30 as 3) in a realistic simulation study with response shapes and forms mot… |
|
31 from a dataset on bottle outlines. |
|
32 </summary></entry><entry><id>http://arxiv.org/abs/2107.04136</id><title>… |
|
33 precision matrices encodes complete information about independence and |
|
34 conditional independence properties. For general distributions, the cova… |
|
35 and precision matrices reveal correlations and so-called partial correla… |
|
36 between variables, but these do not, in general, have any correspondence… |
|
37 respect to independence properties. In this paper, we prove that, for a … |
|
38 class of non-Gaussian distributions, these correspondences still hold, e… |
|
39 for the covariance and approximately for the precision. The distribution… |
|
40 sometimes referred to as "nonparanormal" -- are given by diagonal |
|
41 transformations of multivariate normal random variables. We provide seve… |
|
42 analytic and numerical examples illustrating these results. |
|
43 </summary></entry><entry><id>http://arxiv.org/abs/2106.09370</id><title>… |
|
44 renewables is one of the pillars to power a carbon-neutral society by 20… |
|
45 However, in contrast to conventional power plants, renewable energy is s… |
|
46 to uncertainty raising challenges for their interaction with power syste… |
|
47 Scenario-based probabilistic forecasting models have become a vital tool… |
|
48 equip decision-makers. This paper presents to the power systems forecast… |
|
49 practitioners a recent deep learning technique, the normalizing flows, to |
|
50 produce accurate scenario-based probabilistic forecasts that are crucial… |
|
51 face the new challenges in power systems applications. The strength of t… |
|
52 technique is to directly learn the stochastic multivariate distribution … |
|
53 underlying process by maximizing the likelihood. Through comprehensive |
|
54 empirical evaluations using the open data of the Global Energy Forecasti… |
|
55 Competition 2014, we demonstrate that this methodology is competitive wi… |
|
56 other state-of-the-art deep learning generative models: generative adver… |
|
57 networks and variational autoencoders. The models producing weather-base… |
|
58 solar power, and load scenarios are properly compared in terms of foreca… |
|
59 value by considering the case study of an energy retailer and quality us… |
|
60 several complementary metrics. The numerical experiments are simple and … |
|
61 reproducible. Thus, we hope it will encourage other forecasting practiti… |
|
62 to test and use normalizing flows in power system applications such as b… |
|
63 on electricity markets, scheduling power systems with high renewable ene… |
|
64 sources penetration, energy management of virtual power plan or microgri… |
|
65 unit commitment. |
|
66 </summary></entry><entry><id>http://arxiv.org/abs/2105.14367</id><title>… |
|
67 probability of an event conditioned on some inputs. A neural network (NN… |
|
68 be used to compute the output distribution for continuous-domain, but it… |
|
69 difficult to explicitly approximate a free-form one without knowing the |
|
70 information of its general form a priori. In order to fit an arbitrary |
|
71 conditional distribution, discretizing the continuous domain into bins i… |
|
72 effective strategy, as long as we have sufficiently narrow bins and very… |
|
73 data. However, collecting enough data is often hard to reach and falls f… |
|
74 short of that ideal in many circumstances, especially in multivariate CD… |
|
75 the curse of dimensionality. In this paper, we demonstrate the benefits … |
|
76 modeling free-form conditional distributions using a deconvolution-based… |
|
77 net framework, coping with data deficiency problems in discretization. I… |
|
78 the advantage of being flexible but also takes advantage of the hierarch… |
|
79 smoothness offered by the deconvolution layers. We compare our method to… |
|
80 number of other density-estimation approaches and show that our Deconvol… |
|
81 Density Network (DDN) outperforms the competing methods on many univaria… |
|
82 multivariate tasks. |
|
83 </summary></entry><entry><id>http://arxiv.org/abs/2102.07767</id><title>… |
|
84 agents seeks to agree on a set of hypotheses that best describes a seque… |
|
85 private observations. In the scenario where the set of hypotheses is lar… |
|
86 propose a belief update rule where agents share compressed (either spars… |
|
87 quantized) beliefs with an arbitrary positive compression rate. Our algo… |
|
88 leverages a unified communication rule that enables agents to access |
|
89 wide-ranging compression operators as black-box modules. We prove the al… |
|
90 sure asymptotic exponential convergence of beliefs around the set of opt… |
|
91 hypotheses. Additionally, we show a non-asymptotic, explicit, and linear |
|
92 concentration rate in probability of the beliefs on the optimal hypothes… |
|
93 We provide numerical experiments to illustrate the communication benefit… |
|
94 our method. The simulation results show that the number of transmitted b… |
|
95 be reduced to 5-10% of the non-compressed method in the studied scenario… |
|
96 </summary></entry><entry><id>http://arxiv.org/abs/2012.15059</id><title>… |
|
97 models that are trained across sets of time series, known as Global Fore… |
|
98 Models (GFM), are regularly outperforming traditional univariate forecas… |
|
99 models that work on isolated series. As GFMs usually share the same set … |
|
100 parameters across all time series, they often have the problem of not be… |
|
101 localised enough to a particular series, especially in situations where |
|
102 datasets are heterogeneous. We study how ensembling techniques can be us… |
|
103 generic GFMs and univariate models to solve this issue. Our work systema… |
|
104 and compares relevant current approaches, namely clustering series and t… |
|
105 separate submodels per cluster, the so-called ensemble of specialists ap… |
|
106 and building heterogeneous ensembles of global and local models. We fill… |
|
107 gaps in the existing GFM localisation approaches, in particular by |
|
108 incorporating varied clustering techniques such as feature-based cluster… |
|
109 distance-based clustering and random clustering, and generalise them to … |
|
110 different underlying GFM model types. We then propose a new methodology … |
|
111 clustered ensembles where we train multiple GFMs on different clusters of |
|
112 series, obtained by changing the number of clusters and cluster seeds. U… |
|
113 Feed-forward Neural Networks, Recurrent Neural Networks, and Pooled Regr… |
|
114 models as the underlying GFMs, in our evaluation on eight publicly avail… |
|
115 datasets, the proposed models are able to achieve significantly higher a… |
|
116 than baseline GFM models and univariate forecasting methods. |
|
117 </summary></entry><entry><id>http://arxiv.org/abs/2009.13267</id><title>… |
|
118 such as BLEU score has been studied before for autoregressive neural mac… |
|
119 translation (NMT) and resulted in alternative training algorithms (Ranza… |
|
120 al., 2016; Norouzi et al., 2016; Shen et al., 2016; Wu et al., 2018). Ho… |
|
121 MLE training remains the de facto approach for autoregressive NMT becaus… |
|
122 its computational efficiency and stability. Despite this mismatch betwee… |
|
123 training objective and task measure, we notice that the samples drawn fr… |
|
124 MLE-based trained NMT support the desired distribution -- there are samp… |
|
125 with much higher BLEU score comparing to the beam decoding output. To be… |
|
126 from this observation, we train an energy-based model to mimic the behav… |
|
127 the task measure (i.e., the energy-based model assigns lower energy to s… |
|
128 with higher BLEU score), which is resulted in a re-ranking algorithm bas… |
|
129 the samples drawn from NMT: energy-based re-ranking (EBR). We use both m… |
|
130 energy models (over target sentence) and joint energy models (over both … |
|
131 and target sentences). Our EBR with the joint energy model consistently |
|
132 improves the performance of the Transformer-based NMT: +4 BLEU points on |
|
133 IWSLT'14 German-English, +3.0 BELU points on Sinhala-English, +1.2 BLEU … |
|
134 WMT'16 English-German tasks. |
|
135 </summary></entry><entry><id>http://arxiv.org/abs/2005.11079</id><title>… |
|
136 neural networks (GNNs) have been extensively explored. However, most exi… |
|
137 GNNs inherently suffer from the limitations of over-smoothing, non-robus… |
|
138 and weak-generalization when labeled nodes are scarce. In this paper, we |
|
139 propose a simple yet effective framework -- GRAPH RANDOM NEURAL NETWORKS |
|
140 (GRAND) -- to address these issues. In GRAND, we first design a random |
|
141 propagation strategy to perform graph data augmentation. Then we leverage |
|
142 consistency regularization to optimize the prediction consistency of unl… |
|
143 nodes across different data augmentations. Extensive experiments on graph |
|
144 benchmark datasets suggest that GRAND significantly outperforms |
|
145 state-of-the-art GNN baselines on semi-supervised node classification. F… |
|
146 we show that GRAND mitigates the issues of over-smoothing and non-robust… |
|
147 exhibiting better generalization behavior than existing GNNs. The source… |
|
148 of GRAND is publicly available at https://github.com/Grand20/grand. |
|
149 </summary></entry><entry><id>http://arxiv.org/abs/2004.14427</id><title>… |
|
150 restless bandits with average reward, using the paradigms of Q-learning … |
|
151 Whittle index. Specifically, we leverage the structure of the Whittle in… |
|
152 policy to reduce the search space of Q-learning, resulting in major |
|
153 computational gains. Rigorous convergence analysis is provided, supporte… |
|
154 numerical experiments. The numerical experiments show excellent empirical |
|
155 performance of the proposed scheme. |
|
156 </summary></entry><entry><id>http://arxiv.org/abs/2003.05738</id><title>… |
|
157 state and action spaces. Multi-agent reinforcement learning attempts to … |
|
158 this challenge by distributing control to specialized agents. However, |
|
159 specialization hinders generalization and transferability, and the |
|
160 computational graphs underlying neural-networks architectures -- dominat… |
|
161 the multi-agent setting -- do not offer the flexibility to handle an arb… |
|
162 number of entities which changes both between road networks, and over ti… |
|
163 vehicles traverse the network. We introduce Inductive Graph Reinforcement |
|
164 Learning (IG-RL) based on graph-convolutional networks which adapts to t… |
|
165 structure of any road network, to learn detailed representations of |
|
166 traffic-controllers and their surroundings. Our decentralized approach e… |
|
167 learning of a transferable-adaptive-traffic-signal-control policy. After… |
|
168 trained on an arbitrary set of road networks, our model can generalize t… |
|
169 road networks, traffic distributions, and traffic regimes, with no addit… |
|
170 training and a constant number of parameters, enabling greater scalabili… |
|
171 compared to prior methods. Furthermore, our approach can exploit the |
|
172 granularity of available data by capturing the (dynamic) demand at both … |
|
173 lane and the vehicle levels. The proposed method is tested on both road |
|
174 networks and traffic settings never experienced during training. We comp… |
|
175 IG-RL to multi-agent reinforcement learning and domain-specific baseline… |
|
176 both synthetic road networks and in a larger experiment involving the co… |
|
177 of the 3,971 traffic signals of Manhattan, we show that different |
|
178 instantiations of IG-RL outperform baselines. |
|
179 </summary></entry><entry><id>http://arxiv.org/abs/1905.10029</id><title>… |
|
180 data. However, they have been recently shown to be vulnerable to topolog… |
|
181 attacks. To enhance adversarial robustness, we go beyond spectral graph … |
|
182 to robust graph theory. By challenging the classical graph Laplacian, we |
|
183 propose a new convolution operator that is provably robust in the spectr… |
|
184 domain and is incorporated in the GCN architecture to improve expressivi… |
|
185 interpretability. By extending the original graph to a sequence of graph… |
|
186 also propose a robust training paradigm that encourages transferability … |
|
187 graphs that span a range of spatial and spectral characteristics. The pr… |
|
188 approaches are demonstrated in extensive experiments to simultaneously i… |
|
189 performance in both benign and adversarial situations. |
|
190 </summary></entry><entry><id>http://arxiv.org/abs/2109.10319</id><title>… |
|
191 physiology and computer science. However, at present, most network analy… |
|
192 ignores the direction. In this paper, we construct a spectral clustering… |
|
193 based on the singular decomposition of the adjacency matrix to detect co… |
|
194 in directed stochastic block model (DiSBM). By considering a sparsity |
|
195 parameter, under some mild conditions, we show the proposed approach can |
|
196 consistently recover hidden row and column communities for different sca… |
|
197 degrees. |
|
198 </summary></entry><entry><id>http://arxiv.org/abs/2109.10298</id><title>… |
|
199 Linear Unit (ReLU) Neural Network (NN) architecture (number of layers and |
|
200 number of neurons per layer) with the assurance that it is sufficiently |
|
201 parametrized to control a nonlinear system; i.e. control the system to s… |
|
202 a given formal specification. This is unlike current techniques, which p… |
|
203 no assurances on the resultant architecture. Moreover, our approach requ… |
|
204 only limited knowledge of the underlying nonlinear system and specificat… |
|
205 assume only that the specification can be satisfied by a Lipschitz-conti… |
|
206 controller with a known bound on its Lipschitz constant; the specific |
|
207 controller need not be known. From this assumption, we bound the number … |
|
208 affine functions needed to construct a Continuous Piecewise Affine (CPWA) |
|
209 function that can approximate any Lipschitz-continuous controller that |
|
210 satisfies the specification. Then we connect this CPWA to a NN architect… |
|
211 using the authors' recent results on the Two-Level Lattice (TLL) NN |
|
212 architecture; the TLL architecture was shown to be parameterized by the … |
|
213 of affine functions present in the CPWA function it realizes. |
|
214 </summary></entry><entry><id>http://arxiv.org/abs/2109.10279</id><title>… |
|
215 challenging setup in applied machine learning. Even though model interpr… |
|
216 has attracted more attention in recent years, many modeling approaches s… |
|
217 focus mainly on performance. To further improve the interpretability of … |
|
218 learning models, we suggest the adoption of concepts and tools from the |
|
219 well-established framework of component based multiblock analysis, also … |
|
220 as chemometrics. Nevertheless, artificial neural networks provide greater |
|
221 flexibility in model architecture and thus, often deliver superior predi… |
|
222 performance. In this study, we propose a setup to transfer the concepts … |
|
223 component based statistical models, including multiblock variants of pri… |
|
224 component regression and partial least squares regression, to neural net… |
|
225 architectures. Thereby, we combine the flexibility of neural networks wi… |
|
226 concepts for interpreting block relevance in multiblock methods. In two … |
|
227 cases we demonstrate how the concept can be implemented in practice, and |
|
228 compare it to both common feed-forward neural networks without blocks, a… |
|
229 as statistical component based multiblock methods. Our results underline… |
|
230 multiblock networks allow for basic model interpretation while matching … |
|
231 performance of ordinary feed-forward neural networks. |
|
232 </summary></entry><entry><id>http://arxiv.org/abs/2109.10262</id><title>… |
|
233 reverse-mode automatic differentiation. We use this operator to generali… |
|
234 several optimization algorithms, including a straightforward generalizat… |
|
235 gradient descent and a novel generalization of Newton's method. We then … |
|
236 which properties of these algorithms are preserved in this generalized s… |
|
237 First, we show that the transformation invariances of these algorithms a… |
|
238 preserved: while generalized Newton's method is invariant to all inverti… |
|
239 linear transformations, generalized gradient descent is invariant only to |
|
240 orthogonal linear transformations. Next, we show that we can express the… |
|
241 in loss of generalized gradient descent with an inner product-like expre… |
|
242 thereby generalizing the non-increasing and convergence properties of the |
|
243 gradient descent optimization flow. Finally, we include several numerical |
|
244 experiments to illustrate the ideas in the paper and demonstrate how we … |
|
245 them to optimize polynomial functions over an ordered ring. |
|
246 </summary></entry><entry><id>http://arxiv.org/abs/2109.10254</id><title>… |
|
247 tasks, there is a greater need for accurate quantification of predictive |
|
248 uncertainty. While the common goal in uncertainty quantification (UQ) in |
|
249 machine learning is to approximate the true distribution of the target d… |
|
250 many works in UQ tend to be disjoint in the evaluation metrics utilized,… |
|
251 disparate implementations for each metric lead to numerical results that… |
|
252 not directly comparable across different works. To address this, we intr… |
|
253 Uncertainty Toolbox, an open-source python library that helps to assess, |
|
254 visualize, and improve UQ. Uncertainty Toolbox additionally provides |
|
255 pedagogical resources, such as a glossary of key terms and an organized |
|
256 collection of key paper references. We hope that this toolbox is useful … |
|
257 accelerating and uniting research efforts in uncertainty in machine lear… |
|
258 </summary></entry><entry><id>http://arxiv.org/abs/2109.10219</id><title>… |
|
259 are available. Physical experiments or detailed simulations that accurat… |
|
260 capture the behavior of the system are regarded as high-fidelity models … |
|
261 low model uncertainty, however, they are expensive to run. On the other … |
|
262 simplified physical experiments or numerical models are seen as low-fide… |
|
263 models that are cheaper to evaluate. Although low-fidelity models are of… |
|
264 suitable for direct use in reliability analysis due to their low accurac… |
|
265 can offer information about the trend of the high-fidelity model thus pr… |
|
266 the opportunity to explore the design space at a low cost. This study pr… |
|
267 a new approach called adaptive multi-fidelity Gaussian process for relia… |
|
268 analysis (AMGPRA). Contrary to selecting training points and information |
|
269 sources in two separate stages as done in state-of-the-art mfEGRA method… |
|
270 proposed approach finds the optimal training point and information source |
|
271 simultaneously using the novel collective learning function (CLF). CLF i… |
|
272 to assess the global impact of a candidate training point from an inform… |
|
273 source and it accommodates any learning function that satisfies a certain |
|
274 profile. In this context, CLF provides a new direction for quantifying t… |
|
275 impact of new training points and can be easily extended with new learni… |
|
276 functions to adapt to different reliability problems. The performance of… |
|
277 proposed method is demonstrated by three mathematical examples and one |
|
278 engineering problem concerning the wind reliability of transmission towe… |
|
279 is shown that the proposed method achieves similar or higher accuracy wi… |
|
280 reduced computational costs compared to state-of-the-art single and |
|
281 multi-fidelity methods. A key application of AMGPRA is high-fidelity fra… |
|
282 modeling using complex and costly physics-based computational models. |
|
283 </summary></entry><entry><id>http://arxiv.org/abs/2109.10162</id><title>… |
|
284 $\varepsilon,\delta\in(0,1)$, a bounded function $f:\{-1,1\}^n\to[-1,1]$… |
|
285 degree at most $d$ can be learned with probability at least $1-\delta$ a… |
|
286 $L_2$-error $\varepsilon$ using $\log(\tfrac{n}{\delta})\,\varepsilon^{-… |
|
287 C^{d^{3/2}\sqrt{\log d}}$ random queries for a universal finite constant… |
|
288 </summary></entry><entry><id>http://arxiv.org/abs/2109.09988</id><title>… |
|
289 has many practical applications. With existing classifiers we may be abl… |
|
290 accurately classify signals, however that accuracy may decline if using a |
|
291 reduced number of attributes. Transforming the data then undertaking red… |
|
292 in dimensionality may improve the quality of the data analysis, decrease… |
|
293 required for classification and simplify models. We propose an approach,… |
|
294 chooses suitable wavelets to transform the data, then combines the outpu… |
|
295 these transforms to construct a dataset to then apply ensemble classifie… |
|
296 We demonstrate this on different data sets, across different classifiers… |
|
297 use differing evaluation methods. Our experimental results demonstrate t… |
|
298 effectiveness of the proposed technique, compared to the approaches that… |
|
299 either raw signal data or a single wavelet transform. |
|
300 </summary></entry><entry><id>http://arxiv.org/abs/2109.09859</id><title>… |
|
301 covariates, and the associated nonconvex problem of fitting these models… |
|
302 data. We develop a general recipe for analyzing the convergence of itera… |
|
303 algorithms for this task from a random initialization. In particular, pr… |
|
304 each iteration can be written as the solution to a convex optimization p… |
|
305 satisfying some natural conditions, we leverage Gaussian comparison theo… |
|
306 derive a deterministic sequence that provides sharp upper and lower boun… |
|
307 the error of the algorithm with sample-splitting. Crucially, this determ… |
|
308 sequence accurately captures both the convergence rate of the algorithm … |
|
309 eventual error floor in the finite-sample regime, and is distinct from t… |
|
310 commonly used "population" sequence that results from taking the |
|
311 infinite-sample limit. We apply our general framework to derive several |
|
312 concrete consequences for parameter estimation in popular statistical mo… |
|
313 including phase retrieval and mixtures of regressions. Provided the samp… |
|
314 scales near-linearly in the dimension, we show sharp global convergence … |
|
315 for both higher-order algorithms based on alternating updates and first-… |
|
316 algorithms based on subgradient descent. These corollaries, in turn, yie… |
|
317 multiple consequences, including: (a) Proof that higher-order algorithms… |
|
318 converge significantly faster than their first-order counterparts (and |
|
319 sometimes super-linearly), even if the two share the same population upd… |
|
320 (b) Intricacies in super-linear convergence behavior for higher-order |
|
321 algorithms, which can be nonstandard (e.g., with exponent 3/2) and sensi… |
|
322 the noise level in the problem. We complement these results with extensi… |
|
323 numerical experiments, which show excellent agreement with our theoretic… |
|
324 predictions. |
|
325 </summary></entry><entry><id>http://arxiv.org/abs/2109.09856</id><title>… |
|
326 proposed for predicting failure of a system or device with multivariate … |
|
327 series sensor data. We treat the multivariate time series sensor data as… |
|
328 for both visualization and computation. Failure follows various patterns… |
|
329 are closely related to the root causes. Different predefined transformat… |
|
330 are applied on the original sensors data to better characterize the fail… |
|
331 patterns. In addition to feature derivation, ensemble method is used to … |
|
332 improve the performance. In addition, a general algorithm architecture o… |
|
333 neural network is proposed to handle multiple types of data with less ma… |
|
334 feature engineering. We apply the proposed method on the early predict f… |
|
335 of computer disk drive in order to improve storage systems availability … |
|
336 avoid data loss. The classification accuracy is largely improved with the |
|
337 enriched features, named smart features. |
|
338 </summary></entry><entry><id>http://arxiv.org/abs/2109.09855</id><title>… |
|
339 actions, dubbed R(MA)^2B. The state of each arm evolves according to a |
|
340 controlled Markov decision process (MDP), and the reward of pulling an a… |
|
341 depends on both the current state of the corresponding MDP and the action |
|
342 taken. The goal is to sequentially choose actions for arms so as to maxi… |
|
343 the expected value of the cumulative rewards collected. Since finding the |
|
344 optimal policy is typically intractable, we propose a computationally ap… |
|
345 index policy which we call Occupancy-Measured-Reward Index Policy. Our p… |
|
346 is well-defined even if the underlying MDPs are not indexable. We prove … |
|
347 is asymptotically optimal when the activation budget and number of arms … |
|
348 scaled up, while keeping their ratio as a constant. For the case when the |
|
349 system parameters are unknown, we develop a learning algorithm. Our lear… |
|
350 algorithm uses the principle of optimism in the face of uncertainty and … |
|
351 uses a generative model in order to fully exploit the structure of |
|
352 Occupancy-Measured-Reward Index Policy. We call it the R(MA)^2B-UCB algo… |
|
353 As compared with the existing algorithms, R(MA)^2B-UCB performs close to… |
|
354 offline optimum policy, and also achieves a sub-linear regret with a low |
|
355 computational complexity. Experimental results show that R(MA)^2B-UCB |
|
356 outperforms the existing algorithms in both regret and run time. |
|
357 </summary></entry><entry><id>http://arxiv.org/abs/2109.09847</id><title>… |
|
358 interpreting machine learning models, with strong theoretical guarantees |
|
359 (consistency, local accuracy) and a wide availability of implementations… |
|
360 use cases. Even though computing SHAP values takes exponential time in g… |
|
361 TreeSHAP takes polynomial time on tree-based models. While the speedup is |
|
362 significant, TreeSHAP can still dominate the computation time of industr… |
|
363 machine learning solutions on datasets with millions or more entries, ca… |
|
364 delays in post-hoc model diagnosis and interpretation service. In this p… |
|
365 present two new algorithms, Fast TreeSHAP v1 and v2, designed to improve… |
|
366 computational efficiency of TreeSHAP for large datasets. We empirically … |
|
367 that Fast TreeSHAP v1 is 1.5x faster than TreeSHAP while keeping the mem… |
|
368 cost unchanged. Similarly, Fast TreeSHAP v2 is 2.5x faster than TreeSHAP… |
|
369 the cost of a slightly higher memory usage, thanks to the pre-computatio… |
|
370 expensive TreeSHAP steps. We also show that Fast TreeSHAP v2 is well-sui… |
|
371 multi-time model interpretations, resulting in as high as 3x faster expl… |
|
372 of newly incoming samples. |
|
373 </summary></entry><entry><id>http://arxiv.org/abs/2109.09831</id><title>… |
|
374 algorithms, can substantially impact their performance. To support users… |
|
375 determining well-performing hyperparameter configurations for their algo… |
|
376 datasets and applications at hand, SMAC3 offers a robust and flexible fr… |
|
377 for Bayesian Optimization, which can improve performance within a few |
|
378 evaluations. It offers several facades and pre-sets for typical use case… |
|
379 as optimizing hyperparameters, solving low dimensional continuous (artif… |
|
380 global optimization problems and configuring algorithms to perform well … |
|
381 multiple problem instances. The SMAC3 package is available under a permi… |
|
382 BSD-license at https://github.com/automl/SMAC3. |
|
383 </summary></entry><entry><id>http://arxiv.org/abs/2109.09816</id><title>… |
|
384 systems. In the beginning, the recommender and rational users have diffe… |
|
385 pieces of knowledge, and the recommender needs to learn the users' knowl… |
|
386 make better recommendations. The recommender learns users' knowledge by |
|
387 observing whether each user followed or deviated from her recommendation… |
|
388 show that learning frequently stalls if the recommender always recommend… |
|
389 choice: users tend to follow the recommendation blindly, and their choic… |
|
390 not reflect their knowledge. Social welfare and the learning rate are im… |
|
391 drastically if the recommender abstains from recommending a choice when … |
|
392 predicts that multiple arms will produce a similar payoff. |
|
393 </summary></entry><entry><id>http://arxiv.org/abs/2011.02602</id><title>… |
|
394 of small businesses and online shops. When processing these digital |
|
395 transactions, recognizing each merchant's real identity (i.e., business … |
|
396 is vital to ensure the integrity of payment processing systems. Conventi… |
|
397 this problem is formulated as a time series classification problem solel… |
|
398 the merchant transaction history. However, with the large scale of the d… |
|
399 and changing behaviors of merchants and consumers over time, it is extre… |
|
400 challenging to achieve satisfying performance from off-the-shelf classif… |
|
401 methods. In this work, we approach this problem from a multi-modal learn… |
|
402 perspective, where we use not only the merchant time series data but als… |
|
403 information of merchant-merchant relationship (i.e., affinity) to verify… |
|
404 self-reported business type (i.e., merchant category) of a given merchan… |
|
405 Specifically, we design two individual encoders, where one is responsibl… |
|
406 encoding temporal information and the other is responsible for affinity |
|
407 information, and a mechanism to fuse the outputs of the two encoders to |
|
408 accomplish the identification task. Our experiments on real-world credit… |
|
409 transaction data between 71,668 merchants and 433,772,755 customers have |
|
410 demonstrated the effectiveness and efficiency of the proposed model. |
|
411 </summary></entry><entry><id>http://arxiv.org/abs/2007.05303</id><title>… |
|
412 provide critical insights for payment processing companies. The capabili… |
|
413 predicting merchants' future is crucial for fraud detection and recommen… |
|
414 systems. Conventionally, this problem is formulated to predict one multi… |
|
415 time series under the multi-horizon setting. However, real-world applica… |
|
416 often require more than one future trend prediction considering the |
|
417 uncertainties, where more than one multivariate time series needs to be |
|
418 predicted. This problem is called multi-future prediction. In this work,… |
|
419 combine the two research directions and propose to study this new proble… |
|
420 multi-future, multi-horizon and multivariate time series prediction. This |
|
421 problem is crucial as it has broad use cases in the financial industry to |
|
422 reduce the risk while improving user experience by providing alternative |
|
423 futures. This problem is also challenging as now we not only need to cap… |
|
424 the patterns and insights from the past but also train a model that has a |
|
425 strong inference capability to project multiple possible outcomes. To so… |
|
426 this problem, we propose a new model using convolutional neural networks… |
|
427 simple yet effective encoder-decoder structure to learn the time series … |
|
428 from multiple perspectives. We use experiments on real-world merchant |
|
429 transaction data to demonstrate the effectiveness of our proposed model.… |
|
430 also provide extensive discussions on different model design choices in … |
|
431 experimental section. |
|
432 </summary></entry><entry><id>http://arxiv.org/abs/2109.09690</id><title>… |
|
433 fast uncertainty estimates for predictions with Deep Neural Networks (DN… |
|
434 Our main contribution is a practical and principled combination of DNNs … |
|
435 sparse Gaussian Processes (GPs). We prove theoretically that DNNs can be… |
|
436 as a special case of sparse GPs, namely mixtures of GP experts (MoE-GP),… |
|
437 devise a learning algorithm that brings the derived theory into practice… |
|
438 experiments from two different robotic tasks -- inverse dynamics of a |
|
439 manipulator and object detection on a micro-aerial vehicle (MAV) -- we s… |
|
440 effectiveness of our approach in terms of predictive uncertainty, improv… |
|
441 scalability, and run-time efficiency on a Jetson TX2. We thus argue that… |
|
442 approach can pave the way towards reliable and fast robot learning syste… |
|
443 uncertainty awareness. |
|
444 </summary></entry><entry><id>http://arxiv.org/abs/2109.09658</id><title>… |
|
445 extensive amount of data generated by today's clinical systems, has led … |
|
446 development of imaging AI solutions across the whole value chain of medi… |
|
447 imaging, including image reconstruction, medical image segmentation, |
|
448 image-based diagnosis and treatment planning. Notwithstanding the succes… |
|
449 future potential of AI in medical imaging, many stakeholders are concern… |
|
450 the potential risks and ethical implications of imaging AI solutions, wh… |
|
451 perceived as complex, opaque, and difficult to comprehend, utilise, and … |
|
452 in critical clinical applications. Despite these concerns and risks, the… |
|
453 currently no concrete guidelines and best practices for guiding future AI |
|
454 developments in medical imaging towards increased trust, safety and adop… |
|
455 To bridge this gap, this paper introduces a careful selection of guiding |
|
456 principles drawn from the accumulated experiences, consensus, and best |
|
457 practices from five large European projects on AI in Health Imaging. The… |
|
458 guiding principles are named FUTURE-AI and its building blocks consist o… |
|
459 Fairness, (ii) Universality, (iii) Traceability, (iv) Usability, (v) Rob… |
|
460 and (vi) Explainability. In a step-by-step approach, these guidelines are |
|
461 further translated into a framework of concrete recommendations for spec… |
|
462 developing, evaluating, and deploying technically, clinically and ethica… |
|
463 trustworthy AI solutions into clinical practice. |
|
464 </summary></entry><entry><id>http://arxiv.org/abs/2109.09105</id><title>… |
|
465 including spoken language understanding (SLU). Spoken language requires … |
|
466 understanding of speaker interactions, dialog states and speech induced |
|
467 multimodal behaviors to generate a meaningful representation of the |
|
468 conversation. In this work, we propose to dissect SLU into three represe… |
|
469 properties:conversational (disfluency, pause, overtalk), channel (speake… |
|
470 turn-tasks) and ASR (insertion, deletion,substitution). We probe BERT ba… |
|
471 language models (BERT, RoBERTa) trained on spoken transcripts to investi… |
|
472 its ability to understand multifarious properties in absence of any spee… |
|
473 cues. Empirical results indicate that LM is surprisingly good at capturi… |
|
474 conversational properties such as pause prediction and overtalk detectio… |
|
475 lexical tokens. On the downsides, the LM scores low on turn-tasks and ASR |
|
476 errors predictions. Additionally, pre-training the LM on spoken transcri… |
|
477 restrain its linguistic understanding. Finally, we establish the efficac… |
|
478 transferability of the mentioned properties on two benchmark datasets: |
|
479 Switchboard Dialog Act and Disfluency datasets. |
|
480 </summary></entry><entry><id>http://arxiv.org/abs/2109.07436</id><title>… |
|
481 errors and deviations in execution if there is uncertainty in identifyin… |
|
482 state. So an algorithm that computes a policy for a human to execute oug… |
|
483 consider these effects in its computations. An optimal MDP policy that is |
|
484 poorly executed (because of a human agent) maybe much worse than another… |
|
485 that is executed with fewer errors. In this paper, we consider the probl… |
|
486 erroneous execution and execution delay when computing policies for a hu… |
|
487 agent that would act in a setting modeled by a Markov Decision Process. … |
|
488 present a framework to model the likelihood of policy execution errors a… |
|
489 likelihood of non-policy actions like inaction (delays) due to state |
|
490 uncertainty. This is followed by a hill climbing algorithm to search for… |
|
491 policies that account for these errors. We then use the best policy foun… |
|
492 hill climbing with a branch and bound algorithm to find the optimal poli… |
|
493 show experimental results in a Gridworld domain and analyze the performa… |
|
494 the two algorithms. We also present human studies that verify if our |
|
495 assumptions on policy execution by humans under state-aliasing are reaso… |
|
496 </summary></entry><entry><id>http://arxiv.org/abs/2109.01134</id><title>… |
|
497 for representation learning. It shifts from the tradition of using image… |
|
498 discrete labels for learning a fixed set of weights, seen as visual conc… |
|
499 to aligning images and raw text for two separate encoders. Such a paradi… |
|
500 benefits from a broader source of supervision and allows zero-shot trans… |
|
501 downstream tasks since visual concepts can be diametrically generated fr… |
|
502 natural language, known as prompt. In this paper, we identify that a maj… |
|
503 challenge of deploying such models in practice is prompt engineering. Th… |
|
504 because designing a proper prompt, especially for context words surround… |
|
505 class name, requires domain expertise and typically takes a significant … |
|
506 of time for words tuning since a slight change in wording could have a h… |
|
507 impact on performance. Moreover, different downstream tasks require spec… |
|
508 designs, further hampering the efficiency of deployment. To overcome this |
|
509 challenge, we propose a novel approach named context optimization (CoOp)… |
|
510 main idea is to model context in prompts using continuous representation… |
|
511 perform end-to-end learning from data while keeping the pre-trained para… |
|
512 fixed. In this way, the design of task-relevant prompts can be fully aut… |
|
513 Experiments on 11 datasets show that CoOp effectively turns pre-trained |
|
514 vision-language models into data-efficient visual learners, requiring as… |
|
515 one or two shots to beat hand-crafted prompts with a decent margin and a… |
|
516 gain significant improvements when using more shots (e.g., at 16 shots t… |
|
517 average gain is around 17% with the highest reaching over 50%). CoOp also |
|
518 exhibits strong robustness to distribution shift. |
|
519 </summary></entry><entry><id>http://arxiv.org/abs/2108.09432</id><title>… |
|
520 deformation shape generators. The key idea is to enforce the preservatio… |
|
521 local rigidity among the generated shapes. Our approach builds on an |
|
522 approximation of the as-rigid-as possible (or ARAP) deformation energy. … |
|
523 how to develop the unsupervised loss via a spectral decomposition of the |
|
524 Hessian of the ARAP energy. Our loss nicely decouples pose and shape var… |
|
525 through a robust norm. The loss admits simple closed-form expressions. I… |
|
526 easy to train and can be plugged into any standard generation models, e.… |
|
527 variational auto-encoder (VAE) and auto-decoder (AD). Experimental resul… |
|
528 that our approach outperforms existing shape generation approaches consi… |
|
529 on public benchmark datasets of various shape categories such as human, … |
|
530 and bone. |
|
531 </summary></entry><entry><id>http://arxiv.org/abs/2107.11913</id><title>… |
|
532 has become the subject of interest of academia, government, and industry. |
|
533 Efforts towards measuring different phenomena have gained traction in th… |
|
534 community, as illustrated by the publication of several influential field |
|
535 reports and policy documents. These metrics are designed to help decision |
|
536 takers to inform themselves about the fast-moving and impacting influenc… |
|
537 key advances in Artificial Intelligence in general and Machine Learning … |
|
538 particular. In this paper we propose to use such newfound capabilities o… |
|
539 technologies to augment our AI measuring capabilities. We do so by train… |
|
540 model to classify publications related to ethical issues and concerns. I… |
|
541 methodology we use an expert, manually curated dataset as the training s… |
|
542 then evaluate a large set of research papers. Finally, we highlight the |
|
543 implications of AI metrics, in particular their contribution towards dev… |
|
544 trustful and fair AI-based tools and technologies. Keywords: AI Ethics; … |
|
545 Fairness; AI Measurement. Ethics in Computer Science. |
|
546 </summary></entry><entry><id>http://arxiv.org/abs/2107.04775</id><title>… |
|
547 high-dimensional environments to learn complex tasks, but can often exhi… |
|
548 unsafe behaviors and require extensive environment interaction when expl… |
|
549 is unconstrained. A promising strategy for learning in dynamically uncer… |
|
550 environments is requiring that the agent can robustly return to learned … |
|
551 sets, where task success (and therefore safety) can be guaranteed. While… |
|
552 approach has been successful in low-dimensions, enforcing this constrain… |
|
553 environments with visual observations is exceedingly challenging. We pre… |
|
554 novel continuous representation for safe sets by framing it as a binary |
|
555 classification problem in a learned latent space, which flexibly scales … |
|
556 image observations. We then present a new algorithm, Latent Space Safe S… |
|
557 (LS3), which uses this representation for long-horizon tasks with sparse |
|
558 rewards. We evaluate LS3 on 4 domains, including a challenging sequential |
|
559 pushing task in simulation and a physical cable routing task. We find th… |
|
560 can use prior task successes to restrict exploration and learn more effi… |
|
561 than prior algorithms while satisfying constraints. See |
|
562 https://tinyurl.com/latent-ss for code and supplementary material. |
|
563 </summary></entry><entry><id>http://arxiv.org/abs/2106.07857</id><title>… |
|
564 human-robot interaction. Current researches in this field mainly focus on |
|
565 generating responses consistent with the robot's pre-assigned persona, w… |
|
566 ignoring the user's persona. Such responses may be inappropriate or even |
|
567 offensive, which may lead to the bad user experience. Therefore, we prop… |
|
568 Bilateral Personalized Dialogue Generation (BPDG) method for dyadic |
|
569 conversation, which integrates user and robot personas into dialogue gen… |
|
570 via designing a dynamic persona-aware fusion method. To bridge the gap b… |
|
571 the learning objective function and evaluation metrics, the Conditional … |
|
572 Information Maximum (CMIM) criterion is adopted with contrastive learnin… |
|
573 select the proper response from the generated candidates. Moreover, a bi… |
|
574 persona accuracy metric is designed to measure the degree of bilateral |
|
575 personalization. Experimental results demonstrate that, compared with se… |
|
576 state-of-the-art methods, the final results of the proposed method are m… |
|
577 personalized and consistent with bilateral personas in terms of both aut… |
|
578 and manual evaluations. |
|
579 </summary></entry><entry><id>http://arxiv.org/abs/2105.15033</id><title>… |
|
580 and conceptual knowledge, especially in the medical domain. However, the… |
|
581 of high-quality annotated corpora remains a crucial problem for advancin… |
|
582 research and applications on this task. In order to accelerate the resea… |
|
583 domain-specific knowledge graphs in the medical domain, we introduce Dia… |
|
584 high-quality Chinese dataset for Diabetes knowledge graph, which contains |
|
585 22,050 entities and 6,890 relations in total. We implement recent typical |
|
586 methods for Named Entity Recognition and Relation Extraction as a benchm… |
|
587 evaluate the proposed dataset thoroughly. Empirical results show that th… |
|
588 is challenging for most existing methods and further analysis is conduct… |
|
589 discuss future research direction for improvements. We hope the release … |
|
590 dataset can assist the construction of diabetes knowledge graphs and fac… |
|
591 AI-based applications. |
|
592 </summary></entry><entry><id>http://arxiv.org/abs/2105.11844</id><title>… |
|
593 aerial and satellite images is of high importance in several fields such… |
|
594 security, anomaly detection, land use planning and land use change detec… |
|
595 However, the detection of such infrastructures is complex as they have h… |
|
596 variable shapes and sizes, i.e., some infrastructures, such as electrical |
|
597 substations, are too small while others, such as airports, are too large. |
|
598 Besides, airports can have a surface area either small or too large with |
|
599 completely different shapes, which makes its correct detection challengi… |
|
600 far as we know, these limitations have not been tackled yet in previous … |
|
601 This paper presents (1) a smart Critical Infrastructure dataset, named |
|
602 CI-dataset, organised into two scales, small and large scales critical |
|
603 infrastructures and (2) a two-level resolution-independent critical |
|
604 infrastructure detection (DetDSCI) methodology that first determines the |
|
605 spatial resolution of the input image using a classification model, then |
|
606 analyses the image using the appropriate detector for that spatial resol… |
|
607 The present study targets two representative classes, airports and elect… |
|
608 substations. Our experiments show that DetDSCI methodology achieves up to |
|
609 37,53% F1 improvement with respect to Faster R-CNN, one of the most infl… |
|
610 detection models. |
|
611 </summary></entry><entry><id>http://arxiv.org/abs/2103.13460</id><title>… |
|
612 widely deployed in industrial robotics settings. Part of the challenge l… |
|
613 identifying slip and other key events from the tactile data stream. In t… |
|
614 paper, we present a learning-based method to detect slip using barometric |
|
615 tactile sensors. Although these sensors have a low resolution, they have… |
|
616 other desirable properties including high reliability and durability, a … |
|
617 slim profile, and a low cost. We are able to achieve slip detection accu… |
|
618 of greater than 91% while being robust to the speed and direction of the… |
|
619 motion. Further, we test our detector on two robot manipulation tasks in… |
|
620 common household objects and demonstrate successful generalization to |
|
621 real-world scenarios not seen during training. We show that barometric t… |
|
622 sensing technology, combined with data-driven learning, is potentially s… |
|
623 for complex manipulation tasks such as slip compensation. |
|
624 </summary></entry><entry><id>http://arxiv.org/abs/2102.08633</id><title>… |
|
625 rules, answer high-level questions such as "May I qualify for VA health … |
|
626 benefits?", and ask follow-up clarification questions whose answer is ne… |
|
627 to answer the original question. However, existing works assume the rule… |
|
628 is provided for each user question, which neglects the essential retriev… |
|
629 in real scenarios. In this work, we propose and investigate an open-retr… |
|
630 setting of conversational machine reading. In the open-retrieval setting… |
|
631 relevant rule texts are unknown so that a system needs to retrieve |
|
632 question-relevant evidence from a collection of rule texts, and answer u… |
|
633 high-level questions according to multiple retrieved rule texts in a |
|
634 conversational manner. We propose MUDERN, a Multi-passage Discourse-aware |
|
635 Entailment Reasoning Network which extracts conditions in the rule texts |
|
636 through discourse segmentation, conducts multi-passage entailment reason… |
|
637 answer user questions directly, or asks clarification follow-up question… |
|
638 inquiry more information. On our created OR-ShARC dataset, MUDERN achiev… |
|
639 state-of-the-art performance, outperforming existing single-passage |
|
640 conversational machine reading models as well as a new multi-passage |
|
641 conversational machine reading baseline by a large margin. In addition, … |
|
642 conduct in-depth analyses to provide new insights into this new setting … |
|
643 model. |
|
644 </summary></entry><entry><id>http://arxiv.org/abs/2102.07358</id><title>… |
|
645 methods. In some target problem domains, there are not many data samples |
|
646 available, which could significantly hinder the learning process. While … |
|
647 from similar domains may be leveraged to help through domain adaptation, |
|
648 obtaining high-quality labeled data for those source domains themselves … |
|
649 be difficult or costly. To address such challenges on data insufficiency… |
|
650 classification problem in a target domain, we propose a weak adaptation |
|
651 learning (WAL) approach that leverages unlabeled data from a similar sou… |
|
652 domain, a low-cost weak annotator that produces labels based on task-spe… |
|
653 heuristics, labeling rules, or other methods (albeit with inaccuracy), a… |
|
654 small amount of labeled data in the target domain. Our approach first co… |
|
655 a theoretical analysis on the error bound of the trained classifier with |
|
656 respect to the data quantity and the performance of the weak annotator, … |
|
657 then introduces a multi-stage weak adaptation learning method to learn an |
|
658 accurate classifier by lowering the error bound. Our experiments demonst… |
|
659 the effectiveness of our approach in learning an accurate classifier with |
|
660 limited labeled data in the target domain and unlabeled data in the sour… |
|
661 domain. |
|
662 </summary></entry><entry><id>http://arxiv.org/abs/2102.04394</id><title>… |
|
663 powerful formalism to represent both the quantum and classical uncertain… |
|
664 quantum systems and to express different statistical operations such as |
|
665 measurement, system combination and expectations as linear algebra opera… |
|
666 This paper explores how density matrices can be used as a building block… |
|
667 build machine learning models exploiting their ability to straightforwar… |
|
668 combine linear algebra and probability. One of the main results of the p… |
|
669 to show that density matrices coupled with random Fourier features could |
|
670 approximate arbitrary probability distributions over $\mathbb{R}^n$. Bas… |
|
671 this finding the paper builds different models for density estimation, |
|
672 classification and regression. These models are differentiable, so it is |
|
673 possible to integrate them with other differentiable components, such as… |
|
674 learning architectures and to learn their parameters using gradient-based |
|
675 optimization. In addition, the paper presents optimization-less training |
|
676 strategies based on estimation and model averaging. The models are evalu… |
|
677 benchmark tasks and the results are reported and discussed. |
|
678 </summary></entry><entry><id>http://arxiv.org/abs/2011.11152</id><title>… |
|
679 training deep neural networks that generalize well. Previous work usually |
|
680 interpreted weight decay as a Gaussian prior from the Bayesian perspecti… |
|
681 However, weight decay sometimes shows mysterious behaviors beyond the |
|
682 conventional understanding. For example, the optimal weight decay value … |
|
683 to be zero given long enough training time. Moreover, existing work typi… |
|
684 failed to recognize the importance of scheduling weight decay during tra… |
|
685 Our work aims at theoretically understanding novel behaviors of weight d… |
|
686 and designing schedulers for weight decay in deep learning. This paper m… |
|
687 has three contributions. First, we propose a novel theoretical interpret… |
|
688 of weight decay from the perspective of learning dynamics. Second, we pr… |
|
689 novel weight-decay linear scaling rule for large-batch training that |
|
690 proportionally increases weight decay rather than the learning rate as t… |
|
691 batch size increases. Third, we provide an effective learning-rate-aware |
|
692 scheduler for weight decay, called the Stable Weight Decay (SWD) method,… |
|
693 to the best of our knowledge, is the first practical design for weight d… |
|
694 scheduling. In our various experiments, the SWD method often makes impro… |
|
695 over $L_{2}$ Regularization and Decoupled Weight Decay. |
|
696 </summary></entry><entry><id>http://arxiv.org/abs/2011.02073</id><title>… |
|
697 policies for high-dimensional, complex robotic tasks, but tends to be |
|
698 data-inefficient. Model-based RL tends to be more data-efficient but oft… |
|
699 suffers from learning a high-dimensional model that is good enough for p… |
|
700 improvement. This limits its use to learning simple models for restricti… |
|
701 domains. Optimal control generates solutions without collecting any data, |
|
702 assuming an accurate model of the system and environment is known, which… |
|
703 often true in many control theory applications. However, optimal control… |
|
704 be scaled to problems with a high-dimensional state space. In this paper… |
|
705 propose a novel approach to alleviate data inefficiency of model-free RL… |
|
706 high-dimensional problems by warm-starting the learning process using a |
|
707 lower-dimensional model-based solution. Particularly, we initialize a ba… |
|
708 function for the high-dimensional RL problem via supervision from a |
|
709 lower-dimensional value function, which can be obtained by solving a |
|
710 lower-dimensional problem with a known, approximate model using "classic… |
|
711 techniques such as value iteration or optimal control. Therefore, our ap… |
|
712 implicitly exploits the model priors from simplified problem space to |
|
713 facilitate the policy learning in high-dimensional RL tasks. We demonstr… |
|
714 approach on two representative robotic learning tasks and observe signif… |
|
715 improvement in policy performance and learning efficiency. We also evalu… |
|
716 method empirically with a third task. |
|
717 </summary></entry><entry><id>http://arxiv.org/abs/2004.12908</id><title>… |
|
718 current task, but also on previously encountered and as yet unencountered |
|
719 tasks. In contrast, classical machine learning starts from a blank slate… |
|
720 tabula rasa, using data only for the single task at hand. While typical |
|
721 transfer learning algorithms can improve performance on future tasks, th… |
|
722 performance on prior tasks degrades upon learning new tasks (called |
|
723 catastrophic forgetting). Many recent approaches for continual or lifelo… |
|
724 learning have attempted to maintain performance given new tasks. But str… |
|
725 to avoid forgetting sets the goal unnecessarily low: the goal of lifelong |
|
726 learning, whether biological or artificial, should be to improve perform… |
|
727 all tasks (including past and future) with any new data. We propose |
|
728 omnidirectional transfer learning algorithms, which includes two special… |
|
729 of interest: decision forests and deep networks. Our key insight is the |
|
730 development of the omni-voter layer, which ensembles representations lea… |
|
731 independently on all tasks to jointly decide how to proceed on any given… |
|
732 data point, thereby improving performance on both past and future tasks.… |
|
733 algorithms demonstrate omnidirectional transfer in a variety of simulate… |
|
734 real data scenarios, including tabular data, image data, spoken data, and |
|
735 adversarial tasks. Moreover, they do so with quasilinear space and time |
|
736 complexity. |
|
737 </summary></entry><entry><id>http://arxiv.org/abs/2109.10322</id><title>… |
|
738 dense visual recognition tasks, such as scene segmentation. The last lay… |
|
739 FCN is typically a global classifier (1x1 convolution) to recognize each… |
|
740 to a semantic label. We empirically show that this global classifier, ig… |
|
741 the intra-class distinction, may lead to sub-optimal results. |
|
742 </summary></entry><entry><id>http://arxiv.org/abs/2109.10317</id><title>… |
|
743 do. But deep neural networks are fragile and their behaviors are often |
|
744 surprising. In many settings, we need to provide formal guarantees on the |
|
745 safety, security, correctness, or robustness of neural networks. This bo… |
|
746 covers foundational ideas from formal verification and their adaptation … |
|
747 reasoning about neural networks and deep learning. |
|
748 </summary></entry><entry><id>http://arxiv.org/abs/2109.10312</id><title>… |
|
749 skills from raw images that can be sequenced to complete long-horizon |
|
750 visuomotor tasks. Reinforcement learning (RL) is a promising approach for |
|
751 acquiring short-horizon skills autonomously. However, the focus of RL |
|
752 algorithms has largely been on the success of those individual skills, m… |
|
753 than learning and grounding a large repertoire of skills that can be seq… |
|
754 to complete extended multi-stage tasks. The latter demands robustness and |
|
755 persistence, as errors in skills can compound over time, and may require… |
|
756 robot to have a number of primitive skills in its repertoire, rather tha… |
|
757 one. To this end, we introduce EMBR, a model-based RL method for learning |
|
758 primitive skills that are suitable for completing long-horizon visuomotor |
|
759 tasks. EMBR learns and plans using a learned model, critic, and success |
|
760 classifier, where the success classifier serves both as a reward functio… |
|
761 RL and as a grounding mechanism to continuously detect if the robot shou… |
|
762 retry a skill when unsuccessful or under perturbations. Further, the lea… |
|
763 model is task-agnostic and trained using data from all skills, enabling … |
|
764 robot to efficiently learn a number of distinct primitives. These visuom… |
|
765 primitive skills and their associated pre- and post-conditions can then … |
|
766 directly combined with off-the-shelf symbolic planners to complete long-… |
|
767 tasks. On a Franka Emika robot arm, we find that EMBR enables the robot … |
|
768 complete three long-horizon visuomotor tasks at 85% success rate, such as |
|
769 organizing an office desk, a file cabinet, and drawers, which require |
|
770 sequencing up to 12 skills, involve 14 unique learned primitives, and de… |
|
771 generalization to novel objects. |
|
772 </summary></entry><entry><id>http://arxiv.org/abs/2109.10303</id><title>… |
|
773 deterministic finite automata with rewards as outputs, based on Kolmogor… |
|
774 complexity. Kolmogorov complexity is considered since it can detect |
|
775 computational regularities of deterministic optimal policies. We present… |
|
776 planning objective yielding an explicit trade-off between a policy's |
|
777 performance and complexity. It is proven that maximising this objective … |
|
778 non-trivial in the sense that dynamic programming is infeasible. We pres… |
|
779 algorithms obtaining low-complexity policies, where the first algorithm … |
|
780 a low-complexity optimal policy, and the second algorithm finds a policy |
|
781 maximising performance while maintaining local (stage-wise) complexity |
|
782 constraints. We evaluate the algorithms on a simple navigation task for a |
|
783 mobile robot, where our algorithms yield low-complexity policies that co… |
|
784 with intuition. |
|
785 </summary></entry><entry><id>http://arxiv.org/abs/2109.10285</id><title>… |
|
786 light of itssignificance in a wide range of applications including healt… |
|
787 transportation and fi-nance. Until now, the early classification problem… |
|
788 been dealt with by considering onlyirrevocable decisions. This paper int… |
|
789 a new problem calledearly and revocabletimeseries classification, where … |
|
790 decision maker can revoke its earlier decisions based on thenew available |
|
791 measurements. In order to formalize and tackle this problem, we propose … |
|
792 cost-based framework and derive two new approaches from it. The first ap… |
|
793 doesnot consider explicitly the cost of changing decision, while the sec… |
|
794 does. Exten-sive experiments are conducted to evaluate these approaches … |
|
795 large benchmark of realdatasets. The empirical results obtained convinci… |
|
796 show (i) that the ability of revok-ing decisions significantly improves |
|
797 performance over the irrevocable regime, and (ii) thattaking into accoun… |
|
798 cost of changing decision brings even better results in |
|
799 general.Keywords:revocable decisions, cost estimation, online decision m… |
|
800 </summary></entry><entry><id>http://arxiv.org/abs/2109.10246</id><title>… |
|
801 their lack of grounding, i.e., connecting words to their meanings in the |
|
802 physical world. Vision-and-Language (VL) models, trained jointly on text… |
|
803 image or video data, have been offered as a response to such criticisms. |
|
804 However, while VL pretraining has shown success on multimodal tasks such… |
|
805 visual question answering, it is not yet known how the internal linguist… |
|
806 representations themselves compare to their text-only counterparts. This… |
|
807 compares the semantic representations learned via VL vs. text-only pretr… |
|
808 for two recent VL models using a suite of analyses (clustering, probing,… |
|
809 performance on a commonsense question answering task) in a language-only |
|
810 setting. We find that the multimodal models fail to significantly outper… |
|
811 the text-only variants, suggesting that future work is required if multi… |
|
812 pretraining is to be pursued as a means of improving NLP in general. |
|
813 </summary></entry><entry><id>http://arxiv.org/abs/2109.10231</id><title>… |
|
814 provide insights towards behavior change. Prior work has explored how |
|
815 self-trackers reflect on their logged data, but it remains unclear how m… |
|
816 they learn from the tracking feedback, and which information is more use… |
|
817 Indeed, the feedback can still be overwhelming, and making it concise can |
|
818 improve learning by increasing focus and reducing interpretation burden.… |
|
819 conducted a field study of mobile food logging with two feedback modes (… |
|
820 journaling and automatic annotation of food images) and identified learn… |
|
821 differences regarding nutrition, assessment, behavioral, and contextual |
|
822 information. We propose a Self-Tracking Feedback Saliency Framework to d… |
|
823 when to provide feedback, on which specific information, why those detai… |
|
824 how to present them (as manual inquiry or automatic feedback). We propose |
|
825 SalienTrack to implement these requirements. Using the data collected fr… |
|
826 user study, we trained a machine learning model to predict whether a use… |
|
827 learn from each tracked event. Using explainable AI (XAI) techniques, we |
|
828 identified the most salient features per instance and why they lead to p… |
|
829 learning outcomes. We discuss implications for learnability in self-trac… |
|
830 and how adding model explainability expands opportunities for improving |
|
831 feedback experience. |
|
832 </summary></entry><entry><id>http://arxiv.org/abs/2109.10217</id><title>… |
|
833 of content in various industries. These techniques require extensive kno… |
|
834 of the desired content, and about how to actually implement such procedu… |
|
835 methods. Algorithms for learning interpretable generative models from ex… |
|
836 content could alleviate both difficulties. We propose SIGI, a novel meth… |
|
837 inferring shapes and inducing a shape grammar from grid-based 3D building |
|
838 examples. This interpretable grammar is well-suited for co-creative desi… |
|
839 Applied to Minecraft buildings, we show how the shape grammar can be use… |
|
840 automatically generate new buildings in a similar style. |
|
841 </summary></entry><entry><id>http://arxiv.org/abs/2109.10200</id><title>… |
|
842 where both customer locations and demands are uncertain. In particular, |
|
843 potential customers are not restricted to a predefined customer set but … |
|
844 continuously spatially distributed in a given service area. The objectiv… |
|
845 maximize the served demands while fulfilling vehicle capacities and time |
|
846 restrictions. We call this problem the VRP with stochastic customers and |
|
847 demands (VRPSCD). For this problem, we first propose a Markov Decision P… |
|
848 (MDP) formulation representing the classical centralized decision-making |
|
849 perspective where one decision-maker establishes the routes of all vehic… |
|
850 While the resulting formulation turns out to be intractable, it provides… |
|
851 with the ground to develop a new MDP formulation of the VRPSCD represent… |
|
852 decentralized decision-making framework, where vehicles autonomously est… |
|
853 their own routes. This new formulation allows us to develop several stra… |
|
854 to reduce the dimension of the state and action spaces, resulting in a |
|
855 considerably more tractable problem. We solve the decentralized problem … |
|
856 Reinforcement Learning, and in particular, we develop a Q-learning algor… |
|
857 featuring state-of-the-art acceleration techniques such as Replay Memory… |
|
858 Double Q Network. Computational results show that our method considerably |
|
859 outperforms two commonly adopted benchmark policies (random and heuristi… |
|
860 Moreover, when comparing with existing literature, we show that our appr… |
|
861 can compete with specialized methods developed for the particular case o… |
|
862 VRPSCD where customer locations and expected demands are known in advanc… |
|
863 Finally, we show that the value functions and policies obtained by our |
|
864 algorithm can be easily embedded in Rollout algorithms, thus further imp… |
|
865 their performances. |
|
866 </summary></entry><entry><id>http://arxiv.org/abs/2109.10199</id><title>… |
|
867 led researchers and engineers to investigate novel models for robust and |
|
868 reliable control of autonomous robots (navigation, obstacle detection and |
|
869 avoidance, etc.), especially for quadrotors in challenging contexts such… |
|
870 drone racing and aggressive maneuvers. Using spiking neural networks, th… |
|
871 models can be run on neuromorphic hardware to benefit from outstanding u… |
|
872 rates and high energy efficiency. Yet, low-level controllers are often |
|
873 neglected and remain outside of the neuromorphic loop. Designing low-lev… |
|
874 neuromorphic controllers is crucial to remove the standard PID, and ther… |
|
875 benefit from all the advantages of closing the neuromorphic loop. In this |
|
876 paper, we propose a parsimonious and adjustable neuromorphic PID control… |
|
877 endowed with a minimal number of 93 neurons sparsely connected to achieve |
|
878 autonomous, onboard altitude control of a quadrotor equipped with Intel'… |
|
879 neuromorphic chip. We successfully demonstrate the robustness of our pro… |
|
880 network in a set of experiments where the quadrotor is requested to reac… |
|
881 target altitude from take-off. Our results confirm the suitability of su… |
|
882 low-level neuromorphic controllers, ultimately with a very high update |
|
883 frequency. |
|
884 </summary></entry><entry><id>http://arxiv.org/abs/2109.10187</id><title>… |
|
885 in aerial images are displayed in arbitrary directions and usually dense… |
|
886 packed. Although considerable progress has been made, there are still |
|
887 challenges that existing regression-based rotation detectors suffer the … |
|
888 of discontinuous boundaries, which is directly caused by angular periodi… |
|
889 corner ordering. In this paper, we propose a simple effective framework … |
|
890 address the above challenges. Instead of directly regressing the five |
|
891 parameters (coordinates of the central point, width, height, and rotation |
|
892 angle) or the four vertices, we use the area ratio of parallelogram (ARP… |
|
893 accurately describe a multi-oriented object. Specifically, we regress |
|
894 coordinates of center point, height and width of minimum circumscribed |
|
895 rectangle of oriented object and three area ratios {\lambda}_1, {\lambda… |
|
896 {\lambda}_3. This may facilitate the offset learning and avoid the issue… |
|
897 angular periodicity or label points sequence for oriented objects. To fu… |
|
898 remedy the confusion issue nearly horizontal objects, we employ the area… |
|
899 between the object and its horizontal bounding box (minimum circumscribed |
|
900 rectangle) to guide the selection of horizontal or oriented detection fo… |
|
901 object. We also propose a rotation efficient IoU loss (R-EIoU) to connec… |
|
902 horizontal bounding box with the three area ratios and improve the accur… |
|
903 the rotating bounding box. Experimental results on three remote sensing |
|
904 datasets including HRSC2016, DOTA and UCAS-AOD and scene text including |
|
905 ICDAR2015 show that our method achieves superior detection performance c… |
|
906 with many state-of-the-art approaches. The code and model will be coming… |
|
907 paper published. |
|
908 </summary></entry><entry><id>http://arxiv.org/abs/2109.10173</id><title>… |
|
909 the quality of learned policy. Hard-exploration environments are defined… |
|
910 huge state space and sparse rewards. In such conditions, an exhaustive |
|
911 exploration of the environment is often impossible, and the successful t… |
|
912 of an agent requires a lot of interaction steps. In this paper, we propo… |
|
913 exploration method called Rollback-Explore (RbExplore), which utilizes t… |
|
914 concept of the persistent Markov decision process, in which agents during |
|
915 training can roll back to visited states. We test our algorithm in the |
|
916 hard-exploration Prince of Persia game, without rewards and domain knowl… |
|
917 At all used levels of the game, our agent outperforms or shows comparable |
|
918 results with state-of-the-art curiosity methods with knowledge-based int… |
|
919 motivation: ICM and RND. An implementation of RbExplore can be found at |
|
920 https://github.com/cds-mipt/RbExplore. |
|
921 </summary></entry><entry><id>http://arxiv.org/abs/2109.10149</id><title>… |
|
922 feedback methods require human assessment from facilitators or peers. Th… |
|
923 not scalable to large crowds. We propose Interpretable Directed Diversit… |
|
924 automatically predict ideation quality and diversity scores, and provide… |
|
925 explanations - Attribution, Contrastive Attribution, and Counterfactual |
|
926 Suggestions - for deeper feedback on why ideations were scored (low), an… |
|
927 to get higher scores. These explanations provide multi-faceted feedback … |
|
928 users iteratively improve their ideation. We conducted think aloud and |
|
929 controlled user studies to understand how various explanations are used,… |
|
930 evaluated whether explanations improve ideation diversity and quality. U… |
|
931 appreciated that explanation feedback helped focus their efforts and pro… |
|
932 directions for improvement. This resulted in explanations improving dive… |
|
933 compared to no feedback or feedback with predictions only. Hence, our ap… |
|
934 opens opportunities for explainable AI towards scalable and rich feedbac… |
|
935 iterative crowd ideation. |
|
936 </summary></entry><entry><id>http://arxiv.org/abs/2109.10129</id><title>… |
|
937 domains can be expressed and learned in terms of a pool of features defi… |
|
938 from the domain predicates using a description logic grammar. At the sam… |
|
939 most description logics correspond to a fragment of $k$-variable countin… |
|
940 ($C_k$) for $k=2$, that has been shown to provide a tight characterizati… |
|
941 the expressive power of graph neural networks. In this work, we make use… |
|
942 these results to understand the power and limits of using graph neural n… |
|
943 (GNNs) for learning optimal general policies over a number of tractable |
|
944 planning domains where such policies are known to exist. For this, we tr… |
|
945 simple GNN in a supervised manner to approximate the optimal value funct… |
|
946 $V^{*}(s)$ of a number of sample states $s$. As predicted by the theory,… |
|
947 observed that general optimal policies are obtained in domains where gen… |
|
948 optimal value functions can be defined with $C_2$ features but not in th… |
|
949 requiring more expressive $C_3$ features. In addition, it is observed th… |
|
950 features learned are in close correspondence with the features needed to |
|
951 express $V^{*}$ in closed form. The theory and the analysis of the domai… |
|
952 us understand the features that are actually learned as well as those th… |
|
953 cannot be learned in this way, and let us move in a principled manner fr… |
|
954 combinatorial optimization approach to learning general policies to a |
|
955 potentially, more robust and scalable approach based on deep learning. |
|
956 </summary></entry><entry><id>http://arxiv.org/abs/2109.10106</id><title>… |
|
957 planning complex missions for heterogeneous multi-robot teams. This clas… |
|
958 problems involves tasks that can be executed in different ways and are |
|
959 associated with cross-schedule dependencies that constrain the schedules… |
|
960 different robots in the system. The proposed approach involves a |
|
961 multi-objective heuristic search of the mission, represented as a hierar… |
|
962 tree that defines the mission goal. This procedure outputs several favor… |
|
963 ways to fulfill the mission, which directly feed into the next stage of … |
|
964 method. We propose a distributed metaheuristic based on evolutionary |
|
965 computation to allocate tasks and generate schedules for the set of chos… |
|
966 decompositions. The method is evaluated in a simulation setup of an auto… |
|
967 greenhouse use case, where we demonstrate the method's ability to adapt … |
|
968 planning strategy depending on the available robots and the given optimi… |
|
969 criteria. |
|
970 </summary></entry><entry><id>http://arxiv.org/abs/2109.10100</id><title>… |
|
971 deep neural networks. However the computation of Fisher information matr… |
|
972 becomes more and more difficult as the network structure turns large and |
|
973 complex. This paper proposes a new optimization method whose main idea i… |
|
974 accurately replace the natural gradient optimization by reconstructing t… |
|
975 network. More specifically, we reconstruct the structure of the deep neu… |
|
976 network, and optimize the new network using traditional gradient descent… |
|
977 The reconstructed network achieves the effect of the optimization way wi… |
|
978 natural gradient descent. Experimental results show that our optimization |
|
979 method can accelerate the convergence of deep network models and achieve… |
|
980 performance than GD while sharing its computational simplicity. |
|
981 </summary></entry><entry><id>http://arxiv.org/abs/2109.10086</id><title>… |
|
982 improving the first retriever in ranking pipelines. Learning dense embed… |
|
983 to conduct retrieval using efficient approximate nearest neighbors metho… |
|
984 proven to work well. Meanwhile, there has been a growing interest in lea… |
|
985 \emph{sparse} representations for documents and queries, that could inhe… |
|
986 from the desirable properties of bag-of-words models such as the exact m… |
|
987 of terms and the efficiency of inverted indexes. Introduced recently, the |
|
988 SPLADE model provides highly sparse representations and competitive resu… |
|
989 with respect to state-of-the-art dense and sparse approaches. In this pa… |
|
990 build on SPLADE and propose several significant improvements in terms of |
|
991 effectiveness and/or efficiency. More specifically, we modify the pooling |
|
992 mechanism, benchmark a model solely based on document expansion, and int… |
|
993 models trained with distillation. We also report results on the BEIR ben… |
|
994 Overall, SPLADE is considerably improved with more than $9$\% gains on N… |
|
995 on TREC DL 2019, leading to state-of-the-art results on the BEIR benchma… |
|
996 </summary></entry><entry><id>http://arxiv.org/abs/2109.10085</id><title>… |
|
997 seen increasing importance for investment decisions. Hence, investors (a… |
|
998 managers and asset owners) who wanted to incorporate these issues starte… |
|
999 assess companies based on how they handle such topics. For this assessme… |
|
1000 investors rely on specialized rating agencies that issue ratings along t… |
|
1001 environmental, social and governance (ESG) dimensions. Such ratings allo… |
|
1002 to make investment decisions in favor of sustainability. However, rating |
|
1003 agencies base their analysis on subjective assessment of sustainability |
|
1004 reports, not provided by every company. Furthermore, due to human labor |
|
1005 involved, rating agencies are currently facing the challenge to scale up… |
|
1006 coverage in a timely manner. |
|
1007 </summary></entry><entry><id>http://arxiv.org/abs/2109.10065</id><title>… |
|
1008 based on accuracy, quickness, and consistency for antenna modelling. Usi… |
|
1009 Nntool by MATLAB, 22 different combinations of networks and training alg… |
|
1010 are used to predict the dimensions of a rectangular microstrip antenna u… |
|
1011 dielectric constant, height of substrate, and frequency of operation as … |
|
1012 Comparison and characterization of networks is done based on accuracy, m… |
|
1013 square error, and training time. Algorithms, on the other hand, are anal… |
|
1014 their accuracy, speed, reliability, and smoothness in the training proce… |
|
1015 Finally, these results are analyzed, and recommendations are made for ea… |
|
1016 neural network and algorithm based on uses, advantages, and disadvantage… |
|
1017 example, it is observed that Reduced Radial Bias network is the most acc… |
|
1018 network and Scaled Conjugate Gradient is the most reliable algorithm for |
|
1019 electromagnetic modelling. This paper will help a researcher find the op… |
|
1020 network and algorithm directly without doing time-taking experimentation. |
|
1021 </summary></entry><entry><id>http://arxiv.org/abs/2109.10057</id><title>… |
|
1022 network named Localization Transformer (LOTR). The proposed framework is… |
|
1023 direct coordinate regression approach leveraging a Transformer network to |
|
1024 better utilize the spatial information in the feature map. An LOTR model |
|
1025 consists of three main modules: 1) a visual backbone that converts an in… |
|
1026 image into a feature map, 2) a Transformer module that improves the feat… |
|
1027 representation from the visual backbone, and 3) a landmark prediction he… |
|
1028 directly predicts the landmark coordinates from the Transformer's |
|
1029 representation. Given cropped-and-aligned face images, the proposed LOTR… |
|
1030 trained end-to-end without requiring any post-processing steps. This pap… |
|
1031 introduces the smooth-Wing loss function, which addresses the gradient |
|
1032 discontinuity of the Wing loss, leading to better convergence than stand… |
|
1033 loss functions such as L1, L2, and Wing loss. Experimental results on th… |
|
1034 landmark dataset provided by the First Grand Challenge of 106-Point Faci… |
|
1035 Landmark Localization indicate the superiority of LOTR over the existing |
|
1036 methods on the leaderboard and two recent heatmap-based approaches. |
|
1037 </summary></entry><entry><id>http://arxiv.org/abs/2109.10047</id><title>… |
|
1038 aggregate components with shallow and simple architectures, which are li… |
|
1039 by the 'over-smooth' problem. To further explore the benefits from struc… |
|
1040 diversity and depth of GNN architectures, we propose a GNN generation pi… |
|
1041 with a novel two-stage search space, which aims at automatically generat… |
|
1042 high-performance while transferable deep GNN models in a block-wise mann… |
|
1043 Meanwhile, to alleviate the 'over-smooth' problem, we incorporate multip… |
|
1044 flexible residual connection in our search space and apply identity mapp… |
|
1045 the basic GNN layers. For the search algorithm, we use deep-q-learning w… |
|
1046 epsilon-greedy exploration strategy and reward reshaping. Extensive expe… |
|
1047 on real-world datasets show that our generated GNN models outperforms ex… |
|
1048 manually designed and NAS-based ones. |
|
1049 </summary></entry><entry><id>http://arxiv.org/abs/2109.10034</id><title>… |
|
1050 key functions. This process has often been conceptualised within the fra… |
|
1051 of reinforcement learning, which has also gained prominence in machine l… |
|
1052 and artificial intelligence (AI) as a way to optimise decision-making. A… |
|
1053 aspect of both biological and machine reinforcement learning is the |
|
1054 reactivation of previously experienced episodes, referred to as replay. … |
|
1055 is important for memory consolidation in biological neural networks, and… |
|
1056 to stabilising learning in deep neural networks. Here, we review recent |
|
1057 developments concerning the functional roles of replay in the fields of |
|
1058 neuroscience and AI. Complementary progress suggests how replay might su… |
|
1059 learning processes, including generalisation and continual learning, aff… |
|
1060 opportunities to transfer knowledge across the two fields to advance the |
|
1061 understanding of biological and artificial learning and memory. |
|
1062 </summary></entry><entry><id>http://arxiv.org/abs/2109.10020</id><title>… |
|
1063 payment processing networks is essential for system monitoring. Multivar… |
|
1064 time series, aggregated from the past transaction history, can provide v… |
|
1065 insights for such prediction. The general multivariate time series predi… |
|
1066 problem has been well studied and applied across several domains, includ… |
|
1067 manufacturing, medical, and entomology. However, new domain-related chal… |
|
1068 associated with the data such as concept drift and multi-modality have s… |
|
1069 in addition to the real-time requirements of handling the payment transa… |
|
1070 data at scale. In this work, we study the problem of multivariate time s… |
|
1071 prediction for estimating transaction metrics associated with entities i… |
|
1072 payment transaction database. We propose a model with five unique compon… |
|
1073 estimate the transaction metrics from multi-modality data. Four of these |
|
1074 components capture interaction, temporal, scale, and shape perspectives,… |
|
1075 the fifth component fuses these perspectives together. We also propose a… |
|
1076 offline/online training scheme to address concept drift in the data and … |
|
1077 the real-time requirements. Combining the estimation model with a graphi… |
|
1078 user interface, the prototype transaction metric estimation system has |
|
1079 demonstrated its potential benefit as a tool for improving a payment pro… |
|
1080 company's system monitoring capability. |
|
1081 </summary></entry><entry><id>http://arxiv.org/abs/2109.10016</id><title>… |
|
1082 This task is essential because advanced video retrieval applications sho… |
|
1083 enable users to retrieve a precise moment from a large video corpus. We … |
|
1084 a novel CONtextual QUery-awarE Ranking~(CONQUER) model for effective mom… |
|
1085 localization and ranking. CONQUER explores query context for multi-modal… |
|
1086 and representation learning in two different steps. The first step deriv… |
|
1087 fusion weights for the adaptive combination of multi-modal video content… |
|
1088 second step performs bi-directional attention to tightly couple video an… |
|
1089 as a single joint representation for moment localization. As query conte… |
|
1090 fully engaged in video representation learning, from feature fusion to |
|
1091 transformation, the resulting feature is user-centered and has a larger |
|
1092 capacity in capturing multi-modal signals specific to query. We conduct … |
|
1093 on two datasets, TVR for closed-world TV episodes and DiDeMo for open-wo… |
|
1094 user-generated videos, to investigate the potential advantages of fusing… |
|
1095 and query online as a joint representation for moment retrieval. |
|
1096 </summary></entry><entry><id>http://arxiv.org/abs/2109.10011</id><title>… |
|
1097 intelligence, and it has been widely used to measure the abstract reason… |
|
1098 ability of humans. In this paper, to study the abstract reasoning capabi… |
|
1099 deep neural networks, we propose the first unsupervised learning method … |
|
1100 solving RPM problems. Since the ground truth labels are not allowed, we … |
|
1101 a pseudo target based on the prior constraints of the RPM formulation to |
|
1102 approximate the ground truth label, which effectively converts the unsup… |
|
1103 learning strategy into a supervised one. However, the correct answer is … |
|
1104 labelled by the pseudo target, and thus the noisy contrast will lead to |
|
1105 inaccurate model training. To alleviate this issue, we propose to improv… |
|
1106 model performance with negative answers. Moreover, we develop a |
|
1107 decentralization method to adapt the feature representation to different… |
|
1108 problems. Extensive experiments on three datasets demonstrate that our m… |
|
1109 even outperforms some of the supervised approaches. Our code is availabl… |
|
1110 https://github.com/visiontao/ncd. |
|
1111 </summary></entry><entry><id>http://arxiv.org/abs/2109.10007</id><title>… |
|
1112 to measure the degree of similarity between scientific papers. These app… |
|
1113 are intuitive, easy to put into practice, and computationally cheap. Mor… |
|
1114 they have been used to generate a map of science, allowing visualizing r… |
|
1115 field interactions. Nonetheless, these methods do not work unless two pa… |
|
1116 share a standard reference, limiting the two papers usability with no di… |
|
1117 connection. In this work, we propose to extend bibliographic coupling to… |
|
1118 deep neighborhood, by using graph diffusion methods. This method allows |
|
1119 defining similarity between any two papers, making it possible to genera… |
|
1120 local map of science, highlighting field organization. |
|
1121 </summary></entry><entry><id>http://arxiv.org/abs/2109.09975</id><title>… |
|
1122 trajectories of autonomous vehicles when probabilistic predictions of ot… |
|
1123 agents' futures are generated by deep neural networks (DNNs). The presen… |
|
1124 methods address a wide range of representations for uncertain predictions |
|
1125 including both Gaussian and non-Gaussian mixture models to predict both … |
|
1126 positions and control inputs conditioned on the scene contexts. We show … |
|
1127 the problem of risk assessment when Gaussian mixture models (GMMs) of ag… |
|
1128 positions are learned can be solved rapidly to arbitrary levels of accur… |
|
1129 with existing numerical methods. To address the problem of risk assessme… |
|
1130 non-Gaussian mixture models of agent position, we propose finding upper … |
|
1131 on risk using nonlinear Chebyshev's Inequality and sums-of-squares (SOS) |
|
1132 programming; they are both of interest as the former is much faster whil… |
|
1133 latter can be arbitrarily tight. These approaches only require higher or… |
|
1134 statistical moments of agent positions to determine upper bounds on risk… |
|
1135 perform risk assessment when models are learned for agent control inputs… |
|
1136 opposed to positions, we propagate the moments of uncertain control inpu… |
|
1137 through the nonlinear motion dynamics to obtain the exact moments of unc… |
|
1138 position over the planning horizon. To this end, we construct determinis… |
|
1139 linear dynamical systems that govern the exact time evolution of the mom… |
|
1140 uncertain position in the presence of uncertain control inputs. The pres… |
|
1141 methods are demonstrated on realistic predictions from DNNs trained on t… |
|
1142 Argoverse and CARLA datasets and are shown to be effective for rapidly |
|
1143 assessing the probability of low probability events. |
|
1144 </summary></entry><entry><id>http://arxiv.org/abs/2109.09968</id><title>… |
|
1145 games in studying natural language communication between humans and arti… |
|
1146 agents. However, the generalization still remains a big challenge as the… |
|
1147 depend critically on the complexity and variety of training tasks. In th… |
|
1148 paper, we address this problem by introducing a hierarchical framework b… |
|
1149 upon the knowledge graph-based RL agent. In the high level, a meta-polic… |
|
1150 executed to decompose the whole game into a set of subtasks specified by |
|
1151 textual goals, and select one of them based on the KG. Then a sub-policy… |
|
1152 low level is executed to conduct goal-conditioned reinforcement learning… |
|
1153 carry out experiments on games with various difficulty levels and show t… |
|
1154 proposed method enjoys favorable generalizability. |
|
1155 </summary></entry><entry><id>http://arxiv.org/abs/2109.09960</id><title>… |
|
1156 effectively exploit the unlabeled hard regions for semi-supervised medic… |
|
1157 image segmentation. The MC-Net+ model is motivated by the observation th… |
|
1158 models trained with limited annotations are prone to output highly uncer… |
|
1159 and easily mis-classified predictions in the ambiguous regions (e.g. adh… |
|
1160 edges or thin branches) for the image segmentation task. Leveraging these |
|
1161 region-level challenging samples can make the semi-supervised segmentati… |
|
1162 model training more effective. Therefore, our proposed MC-Net+ model con… |
|
1163 of two new designs. First, the model contains one shared encoder and mul… |
|
1164 sightly different decoders (i.e. using different up-sampling strategies)… |
|
1165 statistical discrepancy of multiple decoders' outputs is computed to den… |
|
1166 model's uncertainty, which indicates the unlabeled hard regions. Second,… |
|
1167 mutual consistency constraint is enforced between one decoder's probabil… |
|
1168 output and other decoders' soft pseudo labels. In this way, we minimize … |
|
1169 model's uncertainty during training and force the model to generate inva… |
|
1170 and low-entropy results in such challenging areas of unlabeled data, in … |
|
1171 to learn a generalized feature representation. We compared the segmentat… |
|
1172 results of the MC-Net+ with five state-of-the-art semi-supervised approa… |
|
1173 three public medical datasets. Extension experiments with two common |
|
1174 semi-supervised settings demonstrate the superior performance of our mod… |
|
1175 other existing methods, which sets a new state of the art for semi-super… |
|
1176 medical image segmentation. |
|
1177 </summary></entry><entry><id>http://arxiv.org/abs/2109.09946</id><title>… |
|
1178 case data has long been recognized. Here, we study the problem of identi… |
|
1179 and measuring biases in large-scale legal case data from an algorithmic |
|
1180 fairness perspective. Our approach utilizes two regression models: A bas… |
|
1181 that represents the decisions of a "typical" judge as given by the data … |
|
1182 "fair" judge that applies one of three fairness concepts. Comparing the |
|
1183 decisions of the "typical" judge and the "fair" judge allows for quantif… |
|
1184 biases across demographic groups, as we demonstrate in four case studies… |
|
1185 criminal data from Cook County (Illinois). |
|
1186 </summary></entry><entry><id>http://arxiv.org/abs/2109.09906</id><title>… |
|
1187 visual or audio content. This typically augments the use of technologies… |
|
1188 as AI and ML by allowing to use natural speech for searching by keywords… |
|
1189 video descriptions. Prior research has successfully provided a number of |
|
1190 solutions for speech to text, in the case of a human speech, but this ar… |
|
1191 aims to investigate possible solutions to retrieve sound events based on… |
|
1192 natural language query, and estimate how effective and accurate they are… |
|
1193 this study, we specifically focus on the YamNet, AlexNet, and ResNet-50 |
|
1194 pre-trained models to automatically classify audio samples using their |
|
1195 respective melspectrograms into a number of predefined classes. The pred… |
|
1196 classes can represent sounds associated with actions within a video frag… |
|
1197 Two tests are conducted to evaluate the performance of the models on two |
|
1198 separate problems: audio classification and intervals retrieval based on… |
|
1199 natural language query. Results show that the benchmarked models are com… |
|
1200 in terms of performance, with YamNet slightly outperforming the other two |
|
1201 models. YamNet was able to classify single fixed-size audio samples with… |
|
1202 accuracy and 68.75% precision while its average accuracy on intervals re… |
|
1203 was 71.62% and precision was 41.95%. The investigated method may be embe… |
|
1204 into an automated event marking architecture for streaming services. |
|
1205 </summary></entry><entry><id>http://arxiv.org/abs/2109.09904</id><title>… |
|
1206 own representations, there is significant discontent about their inscrut… |
|
1207 and the attendant problems in their ability to interact with humans. Whi… |
|
1208 alternatives such as neuro-symbolic approaches have been proposed, there… |
|
1209 lack of consensus on what they are about. There are often two independent |
|
1210 motivations (i) symbols as a lingua franca for human-AI interaction and … |
|
1211 symbols as (system-produced) abstractions use in its internal reasoning.… |
|
1212 jury is still out on whether AI systems will need to use symbols in their |
|
1213 internal reasoning to achieve general intelligence capabilities. Whateve… |
|
1214 answer there is, the need for (human-understandable) symbols in human-AI |
|
1215 interaction seems quite compelling. Symbols, like emotions, may well not… |
|
1216 sine qua non for intelligence per se, but they will be crucial for AI sy… |
|
1217 to interact with us humans--as we can neither turn off our emotions nor … |
|
1218 without our symbols. In particular, in many human-designed domains, huma… |
|
1219 would be interested in providing explicit (symbolic) knowledge and advic… |
|
1220 expect machine explanations in kind. This alone requires AI systems to a… |
|
1221 do their I/O in symbolic terms. In this blue sky paper, we argue this po… |
|
1222 view, and discuss research directions that need to be pursued to allow f… |
|
1223 type of human-AI interaction. |
|
1224 </summary></entry><entry><id>http://arxiv.org/abs/2109.09889</id><title>… |
|
1225 beyond the scope of an RL policy. Such states may make the RL system uns… |
|
1226 impede its deployment in real scenarios. In this paper, we propose a sim… |
|
1227 effective anomaly detection framework for deep RL algorithms that |
|
1228 simultaneously considers random, adversarial and out-of-distribution~(OO… |
|
1229 state outliers. In particular, we attain the class-conditional distribut… |
|
1230 for each action class under the Gaussian assumption, and rely on these |
|
1231 distributions to discriminate between inliers and outliers based on Maha… |
|
1232 Distance~(MD) and Robust Mahalanobis Distance. We conduct extensive expe… |
|
1233 on Atari games that verify the effectiveness of our detection strategies… |
|
1234 the best of our knowledge, we present the first in-detail study of stati… |
|
1235 and adversarial anomaly detection in deep RL algorithms. This simple uni… |
|
1236 anomaly detection paves the way towards deploying safe RL systems in rea… |
|
1237 applications. |
|
1238 </summary></entry><entry><id>http://arxiv.org/abs/2109.09876</id><title>… |
|
1239 extended actions, such as options, that can provide benefits in problems |
|
1240 requiring extensive exploration. One promising approach that learns these |
|
1241 options end-to-end is the option-critic (OC) framework. We examine and s… |
|
1242 this paper that OC does not decompose a problem into simpler sub-problem… |
|
1243 instead increases the size of the search over policy space with each opt… |
|
1244 considering the entire state space during learning. This issue can resul… |
|
1245 practical limitations of this method, including sample inefficient learn… |
|
1246 address this problem, we introduce Context-Specific Representation Abstr… |
|
1247 for Deep Option Learning (CRADOL), a new framework that considers both t… |
|
1248 abstraction and context-specific representation abstraction to effective… |
|
1249 reduce the size of the search over policy space. Specifically, our method |
|
1250 learns a factored belief state representation that enables each option t… |
|
1251 a policy over only a subsection of the state space. We test our method a… |
|
1252 hierarchical, non-hierarchical, and modular recurrent neural network bas… |
|
1253 demonstrating significant sample efficiency improvements in challenging |
|
1254 partially observable environments. |
|
1255 </summary></entry><entry><id>http://arxiv.org/abs/2109.09862</id><title>… |
|
1256 pipelines (Jauhiainen et al.,2019) and is not a solved problem in real-w… |
|
1257 settings. We present a lightweight and effective language identifier tha… |
|
1258 robust to changes of domain and to the absence of copious training data. |
|
1259 </summary></entry><entry><id>http://arxiv.org/abs/2109.09861</id><title>… |
|
1260 for autonomous driving, empirical evidence shows that there are still op… |
|
1261 questions around dealing with the challenges of common knowledge assumpt… |
|
1262 well as modeling bounded rationality. To address some of these practical |
|
1263 challenges, we develop a framework of generalized dynamic cognitive hier… |
|
1264 for both modelling naturalistic human driving behavior as well as behavi… |
|
1265 planning for autonomous vehicles (AV). This framework is built upon a ri… |
|
1266 model of level-0 behavior through the use of automata strategies, an |
|
1267 interpretable notion of bounded rationality through safety and maneuver |
|
1268 satisficing, and a robust response for planning. Based on evaluation on … |
|
1269 large naturalistic datasets as well as simulation of critical traffic |
|
1270 scenarios, we show that i) automata strategies are well suited for level… |
|
1271 behavior in a dynamic level-k framework, and ii) the proposed robust res… |
|
1272 to a heterogeneous population of strategic and non-strategic reasoners c… |
|
1273 an effective approach for game theoretic planning in AV. |
|
1274 </summary></entry><entry><id>http://arxiv.org/abs/2109.09844</id><title>… |
|
1275 monitoring of multiple sclerosis is an important component of successful |
|
1276 disease management. Prior studies have established that multiple scleros… |
|
1277 correlated with speech discrepancies. Early research using objective aco… |
|
1278 measurements has discovered measurable dysarthria. |
|
1279 </summary></entry><entry><id>http://arxiv.org/abs/2109.09833</id><title>… |
|
1280 the noise-induced dynamics during training deep neural networks by |
|
1281 gradient-based optimizers. Specifically, we firstly show that the stocha… |
|
1282 gradient noise possesses finite variance, and therefore the classical Ce… |
|
1283 Limit Theorem (CLT) applies; this indicates that the gradient noise is |
|
1284 asymptotically Gaussian. Such an asymptotic result validates the wide-ac… |
|
1285 assumption of Gaussian noise. We clarify that the recently observed phen… |
|
1286 of heavy tails within gradient noise may not be intrinsic properties, bu… |
|
1287 consequence of insufficient mini-batch size; the gradient noise, which i… |
|
1288 of limited i.i.d. random variables, has not reached the asymptotic regim… |
|
1289 CLT, thus deviates from Gaussian. We quantitatively measure the goodness… |
|
1290 Gaussian approximation of the noise, which supports our conclusion. Seco… |
|
1291 we analyze the noise-induced dynamics of stochastic gradient descent usi… |
|
1292 Langevin equation, granting for momentum hyperparameter in the optimizer… |
|
1293 physical interpretation. We then proceed to demonstrate the existence of… |
|
1294 steady-state distribution of stochastic gradient descent and approximate… |
|
1295 distribution at a small learning rate. |
|
1296 </summary></entry><entry><id>http://arxiv.org/abs/2109.09829</id><title>… |
|
1297 required to be processed on regular basis has pushed processing to the e… |
|
1298 the computing systems. Deploying advanced Neural Networks (NN), such as … |
|
1299 neural networks (DNNs) and spiking neural networks (SNNs), that offer |
|
1300 state-of-the-art results on resource-constrained edge devices is challen… |
|
1301 due to the stringent memory and power/energy constraints. Moreover, these |
|
1302 systems are required to maintain correct functionality under diverse sec… |
|
1303 and reliability threats. This paper first discusses existing approaches … |
|
1304 address energy efficiency, reliability, and security issues at different… |
|
1305 layers, i.e., hardware (HW) and software (SW). Afterward, we discuss how… |
|
1306 further improve the performance (latency) and the energy efficiency of E… |
|
1307 systems through HW/SW-level optimizations, such as pruning, quantization… |
|
1308 approximation. To address reliability threats (like permanent and transi… |
|
1309 faults), we highlight cost-effective mitigation techniques, like fault-a… |
|
1310 training and mapping. Moreover, we briefly discuss effective detection a… |
|
1311 protection techniques to address security threats (like model and data |
|
1312 corruption). Towards the end, we discuss how these techniques can be com… |
|
1313 in an integrated cross-layer framework for realizing robust and |
|
1314 energy-efficient Edge AI systems. |
|
1315 </summary></entry><entry><id>http://arxiv.org/abs/2109.09825</id><title>… |
|
1316 many others, unrealized (null) arguments in certain syntactic positions … |
|
1317 refer to a previously introduced entity, and are thus called anaphoric z… |
|
1318 pronouns. The existing resources for studying anaphoric zero pronoun |
|
1319 interpretation are however still limited. In this paper, we use five data |
|
1320 augmentation methods to generate and detect anaphoric zero pronouns |
|
1321 automatically. We use the augmented data as additional training material… |
|
1322 two anaphoric zero pronoun systems for Arabic. Our experimental results … |
|
1323 that data augmentation improves the performance of the two systems, surp… |
|
1324 the state-of-the-art results. |
|
1325 </summary></entry><entry><id>http://arxiv.org/abs/2109.09809</id><title>… |
|
1326 machine learning systems. An increasingly popular approach has been to s… |
|
1327 provide \emph{counterfactual instance explanations}. These specify close |
|
1328 possible worlds in which, contrary to the facts, a person receives their |
|
1329 desired decision from the machine learning system. This paper will draw … |
|
1330 literature from the philosophy of science to argue that a satisfactory |
|
1331 explanation must consist of both counterfactual instances and a causal e… |
|
1332 (or system of equations) that support the counterfactual instances. We w… |
|
1333 show that counterfactual instances by themselves explain little. We will |
|
1334 further illustrate how explainable AI methods that provide both causal |
|
1335 equations and counterfactual instances can successfully explain machine |
|
1336 learning predictions. |
|
1337 </summary></entry><entry><id>http://arxiv.org/abs/2109.09807</id><title>… |
|
1338 risk associated with dynamic occlusion, i.e., occlusion caused by other |
|
1339 vehicles in traffic. Based on the theory of hypergames, we develop a nov… |
|
1340 multi-agent dynamic occlusion risk (DOR) measure for assessing situation… |
|
1341 in dynamic occlusion scenarios. Furthermore, we present a white-box, |
|
1342 scenario-based, accelerated safety validation framework for assessing sa… |
|
1343 strategic planners in AV. Based on evaluation over a large naturalistic |
|
1344 database, our proposed validation method achieves a 4000% speedup compar… |
|
1345 direct validation on naturalistic data, a more diverse coverage, and abi… |
|
1346 generalize beyond the dataset and generate commonly observed dynamic occ… |
|
1347 crashes in traffic in an automated manner. |
|
1348 </summary></entry><entry><id>http://arxiv.org/abs/2109.09791</id><title>… |
|
1349 either numerical methods for the solution of dynamic model equations or |
|
1350 data-driven artificial intelligence algorithms. Within this latter frame… |
|
1351 the present paper illustrates how a deep learning method, exploiting vid… |
|
1352 radar reflectivity frames as input, can be used to realize a warning mac… |
|
1353 able to sound timely alarms of possible severe thunderstorm events. From… |
|
1354 technical viewpoint, the computational core of this approach is the use … |
|
1355 value-weighted skill score for both transforming the probabilistic outco… |
|
1356 the deep neural network into binary classification and assessing the |
|
1357 forecasting performances. The warning machine has been validated against |
|
1358 weather radar data recorded in the Liguria region, in Italy, |
|
1359 |
