local predictors, which are trained on a small amount of data, as well as a fair and class-balanced predictor at the middle and lower levels. To effectively overcome saddle points for minority classes, our lower-level formulation incorporates sharpness-aware minimization. Meanwhile, at the upper level, the framework dynamically adjusts the loss function based on validation loss, ensuring a close alignment between the *global* predictor and local predictors. Theoretical analysis demonstrates the framework’s ability to enhance classification and fairness generalization, potentially resulting in improvements in the generalization bound. Empirical results validate the superior performance of our tri-level framework compared to existing state-of-the-art approaches. The source code can be found at \url{https://github.com/PennShenLab/FACIMS}.'
volume: 216
URL: https://proceedings.mlr.press/v216/tarzanagh23a.html
PDF: https://proceedings.mlr.press/v216/tarzanagh23a/tarzanagh23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-tarzanagh23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Davoud Ataee
family: Tarzanagh
- given: Bojian
family: Hou
- given: Boning
family: Tong
- given: Qi
family: Long
- given: Li
family: Shen
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2123-2133
id: tarzanagh23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2123
lastpage: 2133
published: 2023-07-02 00:00:00 +0000
- title: 'SPDF: Sparse Pre-training and Dense Fine-tuning for Large Language Models'
abstract: 'The pre-training and fine-tuning paradigm has contributed to a number of breakthroughs in Natural Language Processing (NLP). Instead of directly training on a downstream task, language models are first pre-trained on large datasets with cross-domain knowledge (e.g., Pile, MassiveText, etc.) and then fine-tuned on task-specific data (e.g., natural language generation, text summarization, etc.). Scaling the model and dataset size has helped improve the performance of LLMs, but unfortunately, this also lead to highly prohibitive computational costs. Pre-training LLMs often require orders of magnitude more FLOPs than fine-tuning and the model capacity often remains the same between the two phases. To achieve training efficiency w.r.t training FLOPs, we propose to decouple the model capacity between the two phases and introduce Sparse Pre-training and Dense Fine-tuning (SPDF). In this work, we show the benefits of using unstructured weight sparsity to train only a subset of weights during pre-training (Sparse Pre-training) and then recover the representational capacity by allowing the zeroed weights to learn (Dense Fine-tuning). We demonstrate that we can induce up to 75% sparsity into a 1.3B parameter GPT-3 XL model resulting in a 2.5x reduction in pre-training FLOPs, without a significant loss in accuracy on the downstream tasks relative to the dense baseline. By rigorously evaluating multiple downstream tasks, we also establish a relationship between sparsity, task complexity and dataset size. Our work presents a promising direction to train large GPT models at a fraction of the training FLOPs using weight sparsity, while retaining the benefits of pre-trained textual representations for downstream tasks.'
volume: 216
URL: https://proceedings.mlr.press/v216/thangarasa23a.html
PDF: https://proceedings.mlr.press/v216/thangarasa23a/thangarasa23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-thangarasa23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Vithursan
family: Thangarasa
- given: Abhay
family: Gupta
- given: William
family: Marshall
- given: Tianda
family: Li
- given: Kevin
family: Leong
- given: Dennis
family: DeCoste
- given: Sean
family: Lie
- given: Shreyas
family: Saxena
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2134-2146
id: thangarasa23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2134
lastpage: 2146
published: 2023-07-02 00:00:00 +0000
- title: 'Bandits with costly reward observations'
abstract: 'Many machine learning applications rely on large datasets that are conveniently collected from existing sources or that are labeled automatically as a by-product of user actions. However, in settings such as content moderation, accurately and reliably labeled data comes at substantial cost. If a learning algorithm has to pay for reward information, for example by asking a human for feedback, how does this change the exploration/exploitation tradeoff? We study this question in the context of bandit learning. Specifically, we investigate Bandits with Costly Reward Observations, where a cost needs to be paid in order to observe the reward of the bandit’s action. We show that the observation cost implies an $\Omega(c^{1/3}T^{2/3})$ lower bound on the regret. Furthermore, we develop a general non-adaptive bandit algorithm which matches this lower bound, and we present several competitive adaptive learning algorithms for both k-armed and contextual bandits.'
volume: 216
URL: https://proceedings.mlr.press/v216/tucker23a.html
PDF: https://proceedings.mlr.press/v216/tucker23a/tucker23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-tucker23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Aaron D.
family: Tucker
- given: Caleb
family: Biddulph
- given: Claire
family: Wang
- given: Thorsten
family: Joachims
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2147-2156
id: tucker23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2147
lastpage: 2156
published: 2023-07-02 00:00:00 +0000
- title: 'Probabilistic circuits that know what they don’t know'
abstract: 'Probabilistic circuits (PCs) are models that allow exact and tractable probabilistic inference. In contrast to neural networks, they are often assumed to be well-calibrated and robust to out-of-distribution (OOD) data. In this paper, we show that PCs are in fact not robust to OOD data, i.e., they don’t know what they don’t know. We then show how this challenge can be overcome by model uncertainty quantification. To this end, we propose tractable dropout inference (TDI), an inference procedure to estimate uncertainty by deriving an analytical solution to Monte Carlo dropout (MCD) through variance propagation. Unlike MCD in neural networks, which comes at the cost of multiple network evaluations, TDI provides tractable sampling-free uncertainty estimates in a single forward pass. TDI improves the robustness of PCs to distribution shift and OOD data, demonstrated through a series of experiments evaluating the classification confidence and uncertainty estimates on real-world data.'
volume: 216
URL: https://proceedings.mlr.press/v216/ventola23a.html
PDF: https://proceedings.mlr.press/v216/ventola23a/ventola23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-ventola23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Fabrizio
family: Ventola
- given: Steven
family: Braun
- given: Yu
family: Zhongjie
- given: Martin
family: Mundt
- given: Kristian
family: Kersting
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2157-2167
id: ventola23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2157
lastpage: 2167
published: 2023-07-02 00:00:00 +0000
- title: 'A policy gradient approach for optimization of smooth risk measures'
abstract: 'We propose policy gradient algorithms for solving a risk-sensitive reinforcement learning (RL) problem in on-policy as well as off-policy settings. We consider episodic Markov decision processes, and model the risk using the broad class of smooth risk measures of the cumulative discounted reward. We propose two template policy gradient algorithms that optimize a smooth risk measure in on-policy and off-policy RL settings, respectively. We derive non-asymptotic bounds that quantify the rate of convergence of our proposed algorithms to a stationary point of the smooth risk measure. As special cases, we establish that our algorithms apply to optimization of mean-variance and distortion risk measures, respectively.'
volume: 216
URL: https://proceedings.mlr.press/v216/vijayan23a.html
PDF: https://proceedings.mlr.press/v216/vijayan23a/vijayan23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-vijayan23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Nithia
family: Vijayan
- given: L. A.
family: Prashanth
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2168-2178
id: vijayan23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2168
lastpage: 2178
published: 2023-07-02 00:00:00 +0000
- title: 'Birds of an odd feather: guaranteed out-of-distribution (OOD) novel category detection'
abstract: 'In this work, we solve the problem of novel category detection under distribution shift. This problem is critical to ensuring the safety and efficacy of machine learning models, particularly in domains such as healthcare where timely detection of novel subgroups of patients is crucial. To address this problem, we propose a method based on constrained learning. Our approach is guaranteed to detect a novel category under a relatively weak assumption, namely that rare events in past data have bounded frequency under the shifted distribution. Prior works on the problem do not provide such guarantees, as they either attend to very specific types of distribution shift or make stringent assumptions that limit their guarantees. We demonstrate favorable performance of our method on challenging novel category detection problems over real world datasets.'
volume: 216
URL: https://proceedings.mlr.press/v216/wald23a.html
PDF: https://proceedings.mlr.press/v216/wald23a/wald23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wald23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yoav
family: Wald
- given: Suchi
family: Saria
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2179-2191
id: wald23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2179
lastpage: 2191
published: 2023-07-02 00:00:00 +0000
- title: 'Exploration for Free: How Does Reward Heterogeneity Improve Regret in Cooperative Multi-agent Bandits?'
abstract: 'This paper studies a cooperative multi-agent bandit scenario in which the rewards observed by agents are heterogeneous—one agent’s meat can be another agent’s poison. Specifically, the total reward observed by each agent is the sum of two values: an arm-specific reward, capturing the intrinsic value of the arm, and a privately-known agent-specific reward, which captures the personal preference/limitations of the agent. This heterogeneity in total reward leads to different local optimal arms for agents but creates an opportunity for \textit{free exploration} in a cooperative setting—an agent can freely explore its local optimal arm with no regret and share this free observation with some other agents who would suffer regrets if they pull this arm since the arm is not optimal for them. We first characterize a regret lower bound that captures free exploration, i.e., arms that can be freely explored have no contribution to the regret lower bound. Then, we present a cooperative bandit algorithm that takes advantage of free exploration and achieves a near-optimal regret upper bound which tightly matches the regret lower bound up to a constant factor. Lastly, we run numerical simulations to compare our algorithm with various baselines without free exploration.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23a.html
PDF: https://proceedings.mlr.press/v216/wang23a/wang23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Xuchuang
family: Wang
- given: Lin
family: Yang
- given: Yu-zhen Janice
family: Chen
- given: Xutong
family: Liu
- given: Mohammad
family: Hajiesmaili
- given: Don
family: Towsley
- given: John C.S.
family: Lui
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2192-2202
id: wang23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2192
lastpage: 2202
published: 2023-07-02 00:00:00 +0000
- title: 'Efficient Privacy-Preserving Stochastic Nonconvex Optimization'
abstract: 'While many solutions for privacy-preserving convex empirical risk minimization (ERM) have been developed, privacy-preserving nonconvex ERM remains a challenge. We study nonconvex ERM, which takes the form of minimizing a finite-sum of nonconvex loss functions over a training set. We propose a new differentially private stochastic gradient descent algorithm for nonconvex ERM that achieves strong privacy guarantees efficiently, and provide a tight analysis of its privacy and utility guarantees, as well as its gradient complexity. Our algorithm reduces gradient complexity while matching the best-known utility guarantee. Our experiments on benchmark nonconvex ERM problems demonstrate superior performance in terms of both training cost and utility gains compared with previous differentially private methods using the same privacy budgets.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23b.html
PDF: https://proceedings.mlr.press/v216/wang23b/wang23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Lingxiao
family: Wang
- given: Bargav
family: Jayaraman
- given: David
family: Evans
- given: Quanquan
family: Gu
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2203-2213
id: wang23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2203
lastpage: 2213
published: 2023-07-02 00:00:00 +0000
- title: 'Diversity-enhanced probabilistic ensemble for uncertainty estimation'
abstract: 'Ensemble methods combine multiple individual models for prediction, which have demonstrated their effectiveness in accurate uncertainty quantification (UQ) and strong robustness. Obtaining a diverse ensemble set of model parameters results in better model averaging performance and better approximation of the true posterior distribution of these parameters. In this paper, we propose the diversity-enhanced probabilistic ensemble method with the adaptive uncertainty-guided ensemble learning strategy for better quantifying uncertainty and further improving the model robustness. Specifically, we construct the probabilistic ensemble model by building a Gaussian distribution of the model parameters for each ensemble component using Laplacian approximation in a post-processing manner. Then a mixture of Gaussian model is established with learnable and refinable parameters in an EM-like algorithm. During ensemble training, we leverage the uncertainty estimated from previous models as guidance when training the next one such that the new model will focus more on the less explored regions by previous models. Various experiments including out-of-distribution detection and image classification under distributional shifts have demonstrated better uncertainty estimation and improved model generalization ability of our proposed method.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23c.html
PDF: https://proceedings.mlr.press/v216/wang23c/wang23c.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Hanjing
family: Wang
- given: Qiang
family: Ji
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2214-2225
id: wang23c
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2214
lastpage: 2225
published: 2023-07-02 00:00:00 +0000
- title: 'A trajectory is worth three sentences: multimodal transformer for offline reinforcement learning'
abstract: 'Transformers hold tremendous promise in solving offline reinforcement learning (RL) by formulating it as a sequence modeling problem inspired by language modeling (LM). Prior works using transformers model a sample (trajectory) of RL as one sequence analogous to a sequence of words (one sentence) in LM, despite the fact that each trajectory includes tokens from three diverse modalities: state, action, and reward, while a sentence contains words only. Rather than taking a modality-agnostic approach which uniformly models the tokens from different modalities as one sequence, we propose a multimodal sequence modeling approach in which a trajectory (one “sentence”) of three modalities (state, action, reward) is disentangled into three unimodal ones (three “sentences”). We investigate the correlation of different modalities during sequential decision-making and use the insights to design a multimodal transformer, named Decision Transducer (DTd). DTd outperforms prior art in offline RL on the conducted D4RL benchmarks and enjoys better sample efficiency and algorithm flexibility. Our code is made publicly here.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23d.html
PDF: https://proceedings.mlr.press/v216/wang23d/wang23d.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yiqi
family: Wang
- given: Mengdi
family: Xu
- given: Laixi
family: Shi
- given: Yuejie
family: Chi
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2226-2236
id: wang23d
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2226
lastpage: 2236
published: 2023-07-02 00:00:00 +0000
- title: 'Robust distillation for worst-class performance: on the interplay between teacher and student objectives'
abstract: 'Knowledge distillation is a popular technique that has been shown to produce remarkable gains in average accuracy. However, recent work has shown that these gains are not uniform across subgroups in the data, and can often come at the cost of accuracy on rare subgroups and classes. Robust optimization is a common remedy to improve worst-class accuracy in standard learning settings, but in distillation it is unknown whether it is best to apply robust objectives when training the teacher, the student, or both. This work studies the interplay between robust objectives for the teacher and student. Empirically, we show that that jointly modifying the teacher and student objectives can lead to better worst-class student performance and even Pareto improvement in the trade-off between worst-class and overall performance. Theoretically, we show that the *per-class calibration* of teacher scores is key when training a robust student. Both the theory and experiments support the surprising finding that applying a robust teacher training objective does not always yield a more robust student.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23e.html
PDF: https://proceedings.mlr.press/v216/wang23e/wang23e.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Serena
family: Wang
- given: Harikrishna
family: Narasimhan
- given: Yichen
family: Zhou
- given: Sara
family: Hooker
- given: Michal
family: Lukasik
- given: Aditya Krishna
family: Menon
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2237-2247
id: wang23e
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2237
lastpage: 2247
published: 2023-07-02 00:00:00 +0000
- title: 'A constrained Bayesian approach to out-of-distribution prediction'
abstract: 'Consider the problem of out-of-distribution prediction given data from multiple environments. While a sufficiently diverse collection of training environments will facilitate the identification of an invariant predictor, with an optimal generalization performance, many applications only provide us with a limited number of environments. It is thus necessary to consider adapting to distribution shift using a handful of labeled test samples. We propose a constrained Bayesian approach for this task, which restricts to models with a worst-group training loss above a prespecified threshold. Our method avoids a pathology of the standard Bayesian posterior, which occurs when spurious correlations improve in-distribution prediction. We also show that on certain high-dimensional linear problems, constrained modeling improves the sample efficiency of adaptation. Synthetic and real-world experiments demonstrate the robust performance of our approach.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23f.html
PDF: https://proceedings.mlr.press/v216/wang23f/wang23f.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Ziyu
family: Wang
- given: Binjie
family: Yuan
- given: Jiaxun
family: Lu
- given: Bowen
family: Ding
- given: Yunfeng
family: Shao
- given: Qibin
family: Wu
- given: Jun
family: Zhu
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2248-2258
id: wang23f
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2248
lastpage: 2258
published: 2023-07-02 00:00:00 +0000
- title: 'On the Role of Generalization in Transferability of Adversarial Examples'
abstract: 'Black-box adversarial attacks designing adversarial examples for unseen deep neural networks (DNNs) have received great attention over the past years. However, the underlying factors driving the transferability of black-box adversarial examples still lack a thorough understanding. In this paper, we aim to demonstrate the role of the generalization behavior of the substitute classifier used for generating adversarial examples in the transferability of the attack scheme to unobserved DNN classifiers. To do this, we apply the max-min adversarial example game framework and show the importance of the generalization properties of the substitute DNN from training to test data in the success of the black-box attack scheme in application to different DNN classifiers. We prove theoretical generalization bounds on the difference between the attack transferability rates on training and test samples. Our bounds suggest that operator norm-based regularization methods could improve the transferability of the designed adversarial examples. We support our theoretical results by performing several numerical experiments showing the role of the substitute network’s generalization in generating transferable adversarial examples. Our empirical results indicate the power of Lipschitz regularization and early stopping methods in improving the transferability of designed adversarial examples.'
volume: 216
URL: https://proceedings.mlr.press/v216/wang23g.html
PDF: https://proceedings.mlr.press/v216/wang23g/wang23g.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wang23g.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yilin
family: Wang
- given: Farzan
family: Farnia
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2259-2270
id: wang23g
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2259
lastpage: 2270
published: 2023-07-02 00:00:00 +0000
- title: 'Bidirectional Attention as a Mixture of Continuous Word Experts'
abstract: 'Bidirectional attention - composed of the neural network architecture of self-attention with positional encodings, together with the masked language model (MLM) objective - has emerged as a key component of modern large language models (LLMs). Despite its empirical success, few studies have examined its statistical underpinnings: What statistical model is bidirectional attention implicitly fitting? What sets it apart from its non-attention predecessors? We explore these questions in this paper. The key observation is that fitting a single-layer single-head bidirectional attention, upon reparameterization, is equivalent to fitting a continuous bag of words (CBOW) model with mixture-of-experts (MoE) weights. Further, bidirectional attention with multiple heads and multiple layers is equivalent to stacked MoEs and a mixture of MoEs, respectively. This statistical viewpoint reveals the distinct use of MoE in bidirectional attention, which aligns with its practical effectiveness in handling heterogeneous data. It also suggests an immediate extension to categorical tabular data, if we view each word location in a sentence as a tabular feature. Across empirical studies, we find that this extension outperforms existing tabular extensions of transformers in out-of-distribution (OOD) generalization. Finally, this statistical perspective of bidirectional attention enables us to theoretically characterize when linear word analogies are present in its word embeddings. These analyses show that bidirectional attention can require much stronger assumptions to exhibit linear word analogies than its non-attention predecessors.'
volume: 216
URL: https://proceedings.mlr.press/v216/wibisono23a.html
PDF: https://proceedings.mlr.press/v216/wibisono23a/wibisono23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wibisono23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Kevin C.
family: Wibisono
- given: Yixin
family: Wang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2271-2281
id: wibisono23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2271
lastpage: 2281
published: 2023-07-02 00:00:00 +0000
- title: 'Quantifying aleatoric and epistemic uncertainty in machine learning: Are conditional entropy and mutual information appropriate measures?'
abstract: 'The quantification of aleatoric and epistemic uncertainty in terms of conditional entropy and mutual information, respectively, has recently become quite common in machine learning. While the properties of these measures, which are rooted in information theory, seem appealing at first glance, we identify various incoherencies that call their appropriateness into question. In addition to the measures themselves, we critically discuss the idea of an additive decomposition of total uncertainty into its aleatoric and epistemic constituents. Experiments across different computer vision tasks support our theoretical findings and raise concerns about current practice in uncertainty quantification.'
volume: 216
URL: https://proceedings.mlr.press/v216/wimmer23a.html
PDF: https://proceedings.mlr.press/v216/wimmer23a/wimmer23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wimmer23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Lisa
family: Wimmer
- given: Yusuf
family: Sale
- given: Paul
family: Hofman
- given: Bernd
family: Bischl
- given: Eyke
family: Hüllermeier
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2282-2292
id: wimmer23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2282
lastpage: 2292
published: 2023-07-02 00:00:00 +0000
- title: 'Learning To Invert: Simple Adaptive Attacks for Gradient Inversion in Federated Learning'
abstract: 'Gradient inversion attack enables recovery of training samples from model gradients in federated learning (FL), and constitutes a serious threat to data privacy. To mitigate this vulnerability, prior work proposed both principled defenses based on differential privacy, as well as heuristic defenses based on gradient compression as countermeasures. These defenses have so far been very effective, in particular those based on gradient compression that allow the model to maintain high accuracy while greatly reducing the effectiveness of attacks. In this work, we argue that such findings underestimate the privacy risk in FL. As a counterexample, we show that existing defenses can be broken by a simple adaptive attack, where a model trained on auxiliary data is able to invert gradients on both vision and language tasks.'
volume: 216
URL: https://proceedings.mlr.press/v216/wu23a.html
PDF: https://proceedings.mlr.press/v216/wu23a/wu23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wu23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Ruihan
family: Wu
- given: Xiangyu
family: Chen
- given: Chuan
family: Guo
- given: Kilian Q.
family: Weinberger
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2293-2303
id: wu23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2293
lastpage: 2303
published: 2023-07-02 00:00:00 +0000
- title: 'Uniform-PAC Guarantees for Model-Based RL with Bounded Eluder Dimension'
abstract: 'Recently, there has been remarkable progress in reinforcement learning (RL) with general function approximation. However, all these works only provide regret or sample complexity guarantees. It is still an open question if one can achieve stronger performance guarantees, i.e., the uniform probably approximate correctness (Uniform-PAC) guarantee that can imply both a sub-linear regret bound and a polynomial sample complexity for any target learning accuracy. We study this problem by proposing algorithms for both nonlinear bandits and model-based episodic RL using the general function class with a bounded eluder dimension. The key idea of the proposed algorithms is to assign each action to different levels according to its width with respect to the confidence set. The achieved Uniform-PAC sample complexity is tight in the sense that it matches the state-of-the-art regret bounds or sample complexity guarantees when reduced to the linear case. To the best of our knowledge, this is the first work for Uniform-PAC guarantees on bandit and RL that goes beyond linear cases.'
volume: 216
URL: https://proceedings.mlr.press/v216/wu23b.html
PDF: https://proceedings.mlr.press/v216/wu23b/wu23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wu23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yue
family: Wu
- given: Jiafan
family: He
- given: Quanquan
family: Gu
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2304-2313
id: wu23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2304
lastpage: 2313
published: 2023-07-02 00:00:00 +0000
- title: 'Robust Quickest Change Detection for Unnormalized Models'
abstract: 'Detecting an abrupt and persistent change in the underlying distribution of online data streams is an important problem in many applications. This paper proposes a new robust score-based algorithm called RSCUSUM, which can be applied to unnormalized models and addresses the issue of unknown post-change distributions. RSCUSUM replaces the Kullback-Leibler divergence with the Fisher divergence between pre- and post-change distributions for computational efficiency in unnormalized statistical models and introduces a notion of the “least favorable” distribution for robust change detection. The algorithm and its theoretical analysis are demonstrated through simulation studies.'
volume: 216
URL: https://proceedings.mlr.press/v216/wu23c.html
PDF: https://proceedings.mlr.press/v216/wu23c/wu23c.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-wu23c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Suya
family: Wu
- given: Enmao
family: Diao
- given: Jie
family: Ding
- given: Taposh
family: Banerjee
- given: Vahid
family: Tarokh
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2314-2323
id: wu23c
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2314
lastpage: 2323
published: 2023-07-02 00:00:00 +0000
- title: 'A one-sample decentralized proximal algorithm for non-convex stochastic composite optimization'
abstract: 'We focus on decentralized stochastic non-convex optimization, where $n$ agents work together to optimize a composite objective function which is a sum of a smooth term and a non-smooth convex term. To solve this problem, we propose two single-time scale algorithms: \texttt{Prox-DASA} and \texttt{Prox-DASA-GT}. These algorithms can find $\epsilon$-stationary points in $\mathcal{O}(n^{-1}\epsilon^{-2})$ iterations using constant batch sizes (i.e., $\mathcal{O}(1)$). Unlike prior work, our algorithms achieve comparable complexity without requiring large batch sizes, more complex per-iteration operations (such as double loops), or stronger assumptions. Our theoretical findings are supported by extensive numerical experiments, which demonstrate the superiority of our algorithms over previous approaches. Our code is available at \url{https://github.com/xuxingc/ProxDASA}.'
volume: 216
URL: https://proceedings.mlr.press/v216/xiao23a.html
PDF: https://proceedings.mlr.press/v216/xiao23a/xiao23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-xiao23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Tesi
family: Xiao
- given: Xuxing
family: Chen
- given: Krishnakumar
family: Balasubramanian
- given: Saeed
family: Ghadimi
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2324-2334
id: xiao23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2324
lastpage: 2334
published: 2023-07-02 00:00:00 +0000
- title: 'Two-stage holistic and contrastive explanation of image classification'
abstract: 'The need to explain the output of a deep neural network classifier is now widely recognized. While previous methods typically explain a single class in the output, we advocate explaining the whole output, which is a probability distribution over multiple classes. A whole-output explanation can help a human user gain an overall understanding of model behaviour instead of only one aspect of it. It can also provide a natural framework where one can examine the evidence used to discriminate between competing classes, and thereby obtain contrastive explanations. In this paper, we propose a contrastive whole-output explanation (CWOX) method for image classification, and evaluate it using quantitative metrics and through human subject studies. The source code of CWOX is available at https://github.com/vaynexie/CWOX.'
volume: 216
URL: https://proceedings.mlr.press/v216/xie23a.html
PDF: https://proceedings.mlr.press/v216/xie23a/xie23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-xie23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Weiyan
family: Xie
- given: Xiao-Hui
family: Li
- given: Zhi
family: Lin
- given: Leonard K. M.
family: Poon
- given: Caleb Chen
family: Cao
- given: Nevin L.
family: Zhang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2335-2345
id: xie23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2335
lastpage: 2345
published: 2023-07-02 00:00:00 +0000
- title: 'Conformal Risk Control for Ordinal Classification'
abstract: 'As a natural extension to the standard conformal prediction method, several conformal risk control methods have been recently developed and applied to various learning problems. In this work, we seek to control the conformal risk in expectation for ordinal classification tasks, which have broad applications to many real problems. For this purpose, we firstly formulated the ordinal classification task in the conformal risk control framework, and provided theoretic risk bounds of the risk control method. Then we proposed two types of loss functions specially designed for ordinal classification tasks, and developed corresponding algorithms to determine the prediction set for each case to control their risks at a desired level. We demonstrated the effectiveness of our proposed methods, and analyzed the difference between the two types of risks on three different datasets, including a simulated dataset, the UTKFace dataset and the diabetic retinopathy detection dataset.'
volume: 216
URL: https://proceedings.mlr.press/v216/xu23a.html
PDF: https://proceedings.mlr.press/v216/xu23a/xu23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-xu23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yunpeng
family: Xu
- given: Wenge
family: Guo
- given: Zhi
family: Wei
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2346-2355
id: xu23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2346
lastpage: 2355
published: 2023-07-02 00:00:00 +0000
- title: '$E(2)$-Equivariant Vision Transformer'
abstract: 'Vision Transformer (ViT) has achieved remarkable performance in computer vision. However, positional encoding in ViT makes it substantially difficult to learn the intrinsic equivariance in data. Ini- tial attempts have been made on designing equiv- ariant ViT but are proved defective in some cases in this paper. To address this issue, we design a Group Equivariant Vision Transformer (GE-ViT) via a novel, effective positional encoding opera- tor. We prove that GE-ViT meets all the theoreti- cal requirements of an equivariant neural network. Comprehensive experiments are conducted on standard benchmark datasets, demonstrating that GE-ViT significantly outperforms non-equivariant self-attention networks. The code is available at https://github.com/ZJUCDSYangKaifan/GEVit.'
volume: 216
URL: https://proceedings.mlr.press/v216/xu23b.html
PDF: https://proceedings.mlr.press/v216/xu23b/xu23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-xu23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Renjun
family: Xu
- given: Kaifan
family: Yang
- given: Ke
family: Liu
- given: Fengxiang
family: He
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2356-2366
id: xu23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2356
lastpage: 2366
published: 2023-07-02 00:00:00 +0000
- title: 'Provably Efficient Adversarial Imitation Learning with Unknown Transitions'
abstract: 'Imitation learning (IL) has proven to be an effective method for learning good policies from expert demonstrations. Adversarial imitation learning (AIL), a subset of IL methods, is particularly promising, but its theoretical foundation in the presence of unknown transitions has yet to be fully developed. This paper explores the theoretical underpinnings of AIL in this context, where the stochastic and uncertain nature of environment transitions presents a challenge. We examine the expert sample complexity and interaction complexity required to recover good policies. To this end, we establish a framework connecting reward-free exploration and AIL, and propose an algorithm, MB-TAIL, that achieves the minimax optimal expert sample complexity of $\widetilde{\mathcal{O}} (H^{3/2} |\mathcal{S}|/\varepsilon)$ and interaction complexity of $\widetilde{\mathcal{O}} (H^{3} |\mathcal{S}|^2 |\mathcal{A}|/\varepsilon^2)$. Here, $H$ represents the planning horizon, $|\mathcal{S}|$ is the state space size, $|\mathcal{A}|$ is the action space size, and $\varepsilon$ is the desired imitation gap. MB-TAIL is the first algorithm to achieve this level of expert sample complexity in the unknown transition setting and improves upon the interaction complexity of the best-known algorithm, OAL, by $\mathcal{O} (H)$. Additionally, we demonstrate the generalization ability of MB-TAIL by extending it to the function approximation setting and proving that it can achieve expert sample and interaction complexity independent of $|\mathcal{S}|$.'
volume: 216
URL: https://proceedings.mlr.press/v216/xu23c.html
PDF: https://proceedings.mlr.press/v216/xu23c/xu23c.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-xu23c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Tian
family: Xu
- given: Ziniu
family: Li
- given: Yang
family: Yu
- given: Zhi-Quan
family: Luo
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2367-2378
id: xu23c
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2367
lastpage: 2378
published: 2023-07-02 00:00:00 +0000
- title: 'Pessimistic Model Selection for Offline Deep Reinforcement Learning'
abstract: 'Deep Reinforcement Learning (DRL) has demonstrated great potentials in solving sequential decision making problems in many applications. Despite its promising performance, practical gaps exist when deploying DRL in real-world scenarios. One main barrier is the over-fitting issue that leads to poor generalizability of the policy learned by DRL. In particular, for offline DRL with observational data, model selection is a challenging task as there is no ground truth available for performance demonstration, in contrast with the online setting with simulated environments. In this work, we propose a pessimistic model selection (PMS) approach for offline DRL with a theoretical guarantee, which features a provably effective framework for finding the best policy among a set of candidate models. Two refined approaches are also proposed to address the potential bias of DRL model in identifying the optimal policy. Numerical studies demonstrated the superior performance of our approach over existing methods.'
volume: 216
URL: https://proceedings.mlr.press/v216/yang23a.html
PDF: https://proceedings.mlr.press/v216/yang23a/yang23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yang23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Chao-Han Huck
family: Yang
- given: Zhengling
family: Qi
- given: Yifan
family: Cui
- given: Pin-Yu
family: Chen
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2379-2389
id: yang23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2379
lastpage: 2389
published: 2023-07-02 00:00:00 +0000
- title: 'Mixture of Normalizing Flows for European Option Pricing'
abstract: 'We present a mixture of normalizing flows (MoNF) approach to European option pricing with guarantees that its estimations are free from static arbitrage. In contrast to many existing methods that meet economic rationality constraints (e.g., non-arbitrage) by introducing auxiliary losses, our solution meets those constraints exactly by design. To achieve this, we propose to build a model for risk neutral density using normalizing flows, which results in a pricing model, instead of modelling the option pricing function directly. First, we convert the constraints for direct pricing models to the constraints for models backed by risk neutral density estimation, then we design a specific NF architecture that meets these constraints. Furthermore, we find that employing a mixture of such normalizing flows improves the performance significantly, compared to using a deeper single NF. Finally, we present a mechanism to regularise the proposed model, and this regularisation can serve as a bridge between our method and any sample-based mathematical finance method. The evaluations on five option datasets show superiority of our method compared to mathematical finance solutions and some other neural networks based methods. The code is available at \url{https://github.com/qmfin/MoNF}.'
volume: 216
URL: https://proceedings.mlr.press/v216/yang23b.html
PDF: https://proceedings.mlr.press/v216/yang23b/yang23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yang23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yongxin
family: Yang
- given: Timothy M.
family: Hospedales
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2390-2399
id: yang23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2390
lastpage: 2399
published: 2023-07-02 00:00:00 +0000
- title: 'Multi-modal differentiable unsupervised feature selection'
abstract: 'Multi-modal high throughput biological data presents a great scientific opportunity and a significant computational challenge. In multi-modal measurements, every sample is observed simultaneously by two or more sets of sensors. In such settings, many observed variables in both modalities are often nuisance and do not carry information about the phenomenon of interest. Here, we propose a multi-modal unsupervised feature selection framework: identifying informative variables based on coupled high-dimensional measurements. Our method is designed to identify features associated with two types of latent low-dimensional structures: (i) shared structures that govern the observations in both modalities, and (ii) differential structures that appear in only one modality. To that end, we propose two Laplacian-based scoring operators. We incorporate the scores with differentiable gates that mask nuisance features and enhance the accuracy of the structure captured by the graph Laplacian. The performance of the new scheme is illustrated using synthetic and real datasets, including an extended biological application to single-cell multi-omics.'
volume: 216
URL: https://proceedings.mlr.press/v216/yang23c.html
PDF: https://proceedings.mlr.press/v216/yang23c/yang23c.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yang23c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Junchen
family: Yang
- given: Ofir
family: Lindenbaum
- given: Yuval
family: Kluger
- given: Ariel
family: Jaffe
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2400-2410
id: yang23c
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2400
lastpage: 2410
published: 2023-07-02 00:00:00 +0000
- title: 'MMEL: A Joint Learning Framework for Multi-Mention Entity Linking'
abstract: 'Entity linking, bridging mentions in the contexts with their corresponding entities in the knowledge bases, has attracted wide attention due to many potential applications. Recently, plenty of multimodal entity linking approaches have been proposed to take full advantage of the visual information rather than solely the textual modality. Although feasible, these methods mainly focus on the single-mention scenarios and neglect the scenarios where multiple mentions exist simultaneously in the same context, which limits the performance. In fact, such multi-mention scenarios are pretty common in public datasets and real-world applications. To solve this challenge, we first propose a joint feature extraction module to learn the representations of context and entity candidates, from both the visual and textual perspectives. Then, we design a pairwise training scheme (for training) and a multi-mention collaborative ranking method (for testing) to model the potential connections between different mentions. We evaluate our method on a public dataset and a self-constructed dataset, NYTimes-MEL, under both text-only and multimodal scenarios. The experimental results demonstrate that our method can largely outperform the state-of-the-art methods, especially in multi-mention scenarios. Our dataset and source code are publicly available at https://github.com/ycm094/MMEL-main.'
volume: 216
URL: https://proceedings.mlr.press/v216/yang23d.html
PDF: https://proceedings.mlr.press/v216/yang23d/yang23d.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yang23d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Chengmei
family: Yang
- given: Bowei
family: He
- given: Yimeng
family: Wu
- given: Chao
family: Xing
- given: Lianghua
family: He
- given: Chen
family: Ma
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2411-2421
id: yang23d
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2411
lastpage: 2421
published: 2023-07-02 00:00:00 +0000
- title: 'Mitigating Transformer Overconfidence via Lipschitz Regularization'
abstract: 'Though Transformers have achieved promising results in many computer vision tasks, they tend to be over-confident in predictions, as the standard Dot Product Self-Attention (DPSA) can barely preserve distance for the unbounded input domain. In this work, we fill this gap by proposing a novel Lipschitz Regularized Transformer (LRFormer). Specifically, we present a new similarity function with the distance within Banach Space to ensure the Lipschitzness and also regularize the term by a contractive Lipschitz Bound. The proposed method is analyzed with a theoretical guarantee, providing a rigorous basis for its effectiveness and reliability. Extensive experiments conducted on standard vision benchmarks demonstrate that our method outperforms the state-of-the-art single forward pass approaches in prediction, calibration, and uncertainty estimation.'
volume: 216
URL: https://proceedings.mlr.press/v216/ye23a.html
PDF: https://proceedings.mlr.press/v216/ye23a/ye23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-ye23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Wenqian
family: Ye
- given: Yunsheng
family: Ma
- given: Xu
family: Cao
- given: Kun
family: Tang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2422-2432
id: ye23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2422
lastpage: 2432
published: 2023-07-02 00:00:00 +0000
- title: 'Towards Physically Reliable Molecular Representation Learning'
abstract: 'Estimating the energetic properties of molecular systems is a critical task in material design. Machine learning has shown remarkable promise on this task over classical force fields, but a fully data-driven approach suffers from limited labeled data; not just the amount of available data lacks, but the distribution of labeled examples is highly skewed to stable states. In this work, we propose a molecular representation learning method that extrapolates well beyond the training distribution, powered by physics-driven parameter estimation from classical energy equations and self-supervised learning inspired from masked language modeling. To ensure reliability of the proposed model, we introduce a series of novel evaluation schemes in multifaceted ways, beyond the energy or force accuracy that has been dominantly used. From extensive experiments, we demonstrate that the proposed method is effective in discovering molecular structures, outperforming other baselines. Furthermore, we extrapolate it to the chemical reaction pathways beyond stable states, taking a step towards physically reliable molecular representation learning.'
volume: 216
URL: https://proceedings.mlr.press/v216/yi23a.html
PDF: https://proceedings.mlr.press/v216/yi23a/yi23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yi23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Seunghoon
family: Yi
- given: Youngwoo
family: Cho
- given: Jinhwan
family: Sul
- given: Seung Woo
family: Ko
- given: Soo Kyung
family: Kim
- given: Jaegul
family: Choo
- given: Hongkee
family: Yoon
- given: Joonseok
family: Lee
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2433-2443
id: yi23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2433
lastpage: 2443
published: 2023-07-02 00:00:00 +0000
- title: 'Monte-Carlo Search for an Equilibrium in Dec-POMDPs'
abstract: 'Decentralized partially observable Markov decision processes (Dec-POMDPs) formalize the problem of designing individual controllers for a group of collaborative agents under stochastic dynamics and partial observability. Seeking a global optimum is difficult (NEXP complete), but seeking a Nash equilibrium - each agent policy being a best response to the other agents - is more accessible, and allowed addressing infinite-horizon problems with solutions in the form of finite state controllers. In this paper, we show that this approach can be adapted to cases where only a generative model (a simulator) of the Dec-POMDP is available. This requires relying on a simulation-based POMDP solver to construct an agent’s FSC node by node. A related process is used to heuristically derive initial FSCs. Experiment with benchmarks shows that MC-JESP is competitive with existing Dec-POMDP solvers, even better than many offline methods using explicit models.'
volume: 216
URL: https://proceedings.mlr.press/v216/you23a.html
PDF: https://proceedings.mlr.press/v216/you23a/you23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-you23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yang
family: You
- given: Vincent
family: Thomas
- given: Francis
family: Colas
- given: Olivier
family: Buffet
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2444-2453
id: you23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2444
lastpage: 2453
published: 2023-07-02 00:00:00 +0000
- title: 'Online estimation of similarity matrices with incomplete data'
abstract: 'The similarity matrix measures pairwise similarities between a set of data points and is an essential concept in data processing, routinely used in practical applications. Obtaining a similarity matrix is typically straightforward when data points are completely observed. However, incomplete observations can make it challenging to obtain a high-quality similarity matrix, which becomes even more complex in online data. To address this challenge, we propose matrix correction algorithms that leverage the positive semi-definiteness (PSD) of the similarity matrix to improve similarity estimation in both offline and online scenarios. Our approaches have a solid theoretical guarantee of performance and excellent potential for parallel execution on large-scale data. Empirical evaluations demonstrate their high effectiveness and efficiency with significantly improved results over classical imputation-based methods, benefiting downstream applications with superior performance. Our code is available at \url{https://github.com/CUHKSZ-Yu/OnMC}.'
volume: 216
URL: https://proceedings.mlr.press/v216/yu23a.html
PDF: https://proceedings.mlr.press/v216/yu23a/yu23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-yu23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Fangchen
family: Yu
- given: Yicheng
family: Zeng
- given: Jianfeng
family: Mao
- given: Wenye
family: Li
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2454-2464
id: yu23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2454
lastpage: 2464
published: 2023-07-02 00:00:00 +0000
- title: 'Fast Teammate Adaptation in the Presence of Sudden Policy Change'
abstract: ' Cooperative multi-agent reinforcement learning (MARL), where agents coordinates with teammate(s) for a shared goal, may sustain non-stationary caused by the policy change of teammates. Prior works mainly concentrate on the policy change cross episodes, ignoring the fact that teammates may suffer from sudden policy change within an episode, which might lead to miscoordination and poor performance. We formulate the problem as an open Dec-POMDP, where we control some agents to coordinate with uncontrolled teammates, whose policies could be changed within one episode. Then we develop a new framework \textit{\textbf{Fas}t \textbf{t}eammates \textbf{a}da\textbf{p}tation (\textbf{Fastap})} to address the problem. Concretely, we first train versatile teammates’ policies and assign them to different clusters via the Chinese Restaurant Process (CRP). Then, we train the controlled agent(s) to coordinate with the sampled uncontrolled teammates by capturing their identifications as context for fast adaptation. Finally, each agent applies its local information to anticipate the teammates’ context for decision-making accordingly. This process proceeds alternately, leading to a robust policy that can adapt to any teammates during the decentralized execution phase. We show in multiple multi-agent benchmarks that Fastap can achieve superior performance than multiple baselines in stationary and non-stationary scenarios. '
volume: 216
URL: https://proceedings.mlr.press/v216/zhang23a.html
PDF: https://proceedings.mlr.press/v216/zhang23a/zhang23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhang23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Ziqian
family: Zhang
- given: Lei
family: Yuan
- given: Lihe
family: Li
- given: Ke
family: Xue
- given: Chengxing
family: Jia
- given: Cong
family: Guan
- given: Chao
family: Qian
- given: Yang
family: Yu
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2465-2476
id: zhang23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2465
lastpage: 2476
published: 2023-07-02 00:00:00 +0000
- title: 'Energy-based Predictive Representations for Partially Observed Reinforcement Learning'
abstract: 'In real-world applications, handling partial observability is a common requirement for reinforcement learning algorithms, which is not captured by a Markov decision process (MDP). Although partially observable Markov decision processes (POMDPs) have been specifically designed to address this requirement, they present significant computational and statistical challenges in learning and planning. In this work, we introduce the *Energy-based Predictive Representation (EPR)* to provide a unified approach for designing practical reinforcement learning algorithms in both the MDP and POMDP settings. This framework enables coherent handling of *learning, exploration, and planning* tasks. The proposed framework leverages a powerful neural energy-based model to extract an adequate representation, allowing for efficient approximation of Q-functions. This representation facilitates the efficient computation of confidence, enabling the implementation of optimism or pessimism in planning when faced with uncertainty. Consequently, it effectively manages the trade-off between exploration and exploitation. Experimental investigations demonstrate that the proposed algorithm achieves state-of-the-art performance in both MDP and POMDP settings.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhang23b.html
PDF: https://proceedings.mlr.press/v216/zhang23b/zhang23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhang23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Tianjun
family: Zhang
- given: Tongzheng
family: Ren
- given: Chenjun
family: Xiao
- given: Wenli
family: Xiao
- given: Joseph E.
family: Gonzalez
- given: Dale
family: Schuurmans
- given: Bo
family: Dai
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2477-2487
id: zhang23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2477
lastpage: 2487
published: 2023-07-02 00:00:00 +0000
- title: 'Provably efficient representation selection in Low-rank Markov Decision Processes: from online to offline RL'
abstract: 'The success of deep reinforcement learning (DRL) lies in its ability to learn a representation that is well-suited for the exploration and exploitation task. To understand how the choice of representation can improve the efficiency of reinforcement learning (RL), we study representation selection for a class of low-rank Markov Decision Processes (MDPs) where the transition kernel can be represented in a bilinear form. We propose an efficient algorithm, called ReLEX, for representation learning in both online and offline RL. Specifically, we show that the online version of ReLEX, called ReLEX-UCB, always performs no worse than the state-of-the-art algorithm without representation selection, and achieves a strictly better constant regret if the representation function class has a "coverage" property over the entire state-action space. For the offline counterpart, ReLEX-LCB, we show that the algorithm can find the optimal policy if the representation class can cover the state-action space and achieves gap-dependent sample complexity. This is the first result with constant sample complexity for representation learning in offline RL.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhang23c.html
PDF: https://proceedings.mlr.press/v216/zhang23c/zhang23c.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhang23c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: W.
family: Zhang
- given: J.
family: He
- given: D.
family: Zhou
- given: Q.
family: Gu
- given: A.
family: Zhang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2488-2497
id: zhang23c
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2488
lastpage: 2497
published: 2023-07-02 00:00:00 +0000
- title: 'Graph Self-supervised Learning via Proximity Distribution Minimization'
abstract: 'Self-supervised learning (SSL) for graphs is an essential problem since graph data are ubiquitous and labeling can be costly. We argue that existing SSL approaches for graphs have two limitations. First, they rely on corruption techniques such as node attribute perturbation and edge dropping to generate graph views for contrastive learning. These unnatural corruption techniques require extensive tuning efforts and provide marginal improvements. Second, the current approaches require the computation of multiple graph views, which is memory and computationally inefficient. These shortcomings of graph SSL call for a corruption-free single-view learning approach, but the strawman approach of using neighboring nodes as positive examples suffers two problems: it ignores the strength of connections between nodes implied by the graph structure on a macro level, and cannot deal with the high noise in real-world graphs. We propose Proximity Divergence Minimization (PDM), a corruption-free single-view graph SSL approach that overcomes these problems by leveraging node proximity to measure connection strength and denoise the graph structure. Through extensive experiments, we show that PDM achieves up to 4.55% absolute improvement in ROC-AUC on graph SSL tasks over state-of-the-art approaches while being more memory efficient. Moreover, PDM even outperforms supervised training on node classification tasks of ogbn-proteins dataset. Our code is publicly available.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhang23d.html
PDF: https://proceedings.mlr.press/v216/zhang23d/zhang23d.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhang23d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Tianyi
family: Zhang
- given: Zhenwei
family: Dai
- given: Zhaozhuo
family: Xu
- given: Anshumali
family: Shrivastava
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2498-2508
id: zhang23d
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2498
lastpage: 2508
published: 2023-07-02 00:00:00 +0000
- title: 'Greed is good: correspondence recovery for unlabeled linear regression'
abstract: 'We consider the unlabeled linear regression reading as $\mathbf{Y} = \mathbf{\Pi}^{*}\mathbf{X}\mathbf{B}^* + \mathbf{W}$, where $\mathbf{\Pi}^{*}, \mathbf{B}^*$ and $\mathbf{W}$ represents missing (or incomplete) correspondence information, signals, and additive noise, respectively. Our goal is to perform data alignment between $\mathbf{Y}$ and $\mathbf{X}$, or equivalently, reconstruct the correspondence information encoded by $\mathbf{\Pi}^*$. Based on whether signal $\mathbf{B}^*$ is given a prior, we separately propose two greedy-selection-based estimators, which both reach the mini-max optimality. Compared with previous works, our work $(i)$ supports partial recovery of the correspondence information; and $(ii)$ applies to a general matrix family rather than the permutation matrices, to put more specifically, selection matrices, where multiple rows of $\mathbf{X}$ can correspond to the same row in $\mathbf{Y}$. Moreover, numerical experiments are provided to corroborate our claims.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhang23e.html
PDF: https://proceedings.mlr.press/v216/zhang23e/zhang23e.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhang23e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Hang
family: Zhang
- given: Ping
family: Li
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2509-2518
id: zhang23e
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2509
lastpage: 2518
published: 2023-07-02 00:00:00 +0000
- title: 'Conditional counterfactual causal effect for individual attribution'
abstract: 'Identifying the causes of an event, also termed as causal attribution, is a commonly encountered task in many application problems. Available methods, mostly in Bayesian or causal inference literature, suffer from two main drawbacks: 1) cannot attribute for individuals, and 2) attributing one single cause at a time and cannot deal with the interaction effect among multiple causes. In this paper, based on our proposed new measurement, called conditional counterfactual causal effect (CCCE), we introduce an individual causal attribution method, which is able to utilize the individual observation as the evidence and consider common influence and interaction effect of multiple causes simultaneously. We discuss the identifiability of CCCE and also give the identification formulas under proper assumptions. Finally, we conduct experiments on simulated and real data to illustrate the effectiveness of CCCE and the results show that our proposed method outperforms significantly state-of-the-art methods.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhao23a.html
PDF: https://proceedings.mlr.press/v216/zhao23a/zhao23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhao23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Ruiqi
family: Zhao
- given: Lei
family: Zhang
- given: Shengyu
family: Zhu
- given: Zitong
family: Lu
- given: Zhenhua
family: Dong
- given: Chaoliang
family: Zhang
- given: Jun
family: Xu
- given: Zhi
family: Geng
- given: Yangbo
family: He
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2519-2528
id: zhao23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2519
lastpage: 2528
published: 2023-07-02 00:00:00 +0000
- title: 'Conditionally optimistic exploration for cooperative deep multi-agent reinforcement learning'
abstract: 'Efficient exploration is critical in cooperative deep Multi-Agent Reinforcement Learning (MARL). In this work, we propose an exploration method that effectively encourages cooperative exploration based on the idea of sequential action-computation scheme. The high-level intuition is that to perform optimism-based exploration, agents would explore cooperative strategies if each agent’s optimism estimate captures a structured dependency relationship with other agents. Assuming agents compute actions following a sequential order at *each environment timestep*, we provide a perspective to view MARL as tree search iterations by considering agents as nodes at different depths of the search tree. Inspired by the theoretically justified tree search algorithm UCT (Upper Confidence bounds applied to Trees), we develop a method called Conditionally Optimistic Exploration (COE). COE augments each agent’s state-action value estimate with an action-conditioned optimistic bonus derived from the visitation count of the global state and joint actions of preceding agents. COE is performed during training and disabled at deployment, making it compatible with any value decomposition method for centralized training with decentralized execution. Experiments across various cooperative MARL benchmarks show that COE outperforms current state-of-the-art exploration methods on hard-exploration tasks.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhao23b.html
PDF: https://proceedings.mlr.press/v216/zhao23b/zhao23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhao23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Xutong
family: Zhao
- given: Yangchen
family: Pan
- given: Chenjun
family: Xiao
- given: Sarath
family: Chandar
- given: Janarthanan
family: Rajendran
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2529-2540
id: zhao23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2529
lastpage: 2540
published: 2023-07-02 00:00:00 +0000
- title: 'RDM-DC: Poisoning Resilient Dataset Condensation with Robust Distribution Matching'
abstract: 'Dataset condensation aims to condense the original training dataset into a small synthetic dataset for data-efficient learning. The recently proposed dataset condensation techniques allow the model trainers with limited resources to learn acceptable deep learning models on a small amount of synthetic data. However, in an adversarial environment, given the original dataset as a poisoned dataset, dataset condensation may encode the poisoning information into the condensed synthetic dataset. To explore the vulnerability of dataset condensation to data poisoning, we revisit the state-of-the-art targeted data poisoning method and customize a targeted data poisoning algorithm for dataset condensation. By executing the two poisoning methods, we demonstrate that, when the synthetic dataset is condensed from a poisoned dataset, the models trained on the synthetic dataset may predict the targeted sample as the attack-targeted label. To defend against data poisoning, we introduce the concept of poisoned deviation to quantify the poisoning effect. We further propose a poisoning-resilient dataset condensation algorithm with a calibration method to reduce poisoned deviation. Extensive evaluations demonstrate that our proposed algorithm can protect the synthetic dataset from data poisoning with minor performance drop.'
volume: 216
URL: https://proceedings.mlr.press/v216/zheng23a.html
PDF: https://proceedings.mlr.press/v216/zheng23a/zheng23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zheng23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Tianhang
family: Zheng
- given: Baochun
family: Li
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2541-2550
id: zheng23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2541
lastpage: 2550
published: 2023-07-02 00:00:00 +0000
- title: 'Learning robust representation for reinforcement learning with distractions by reward sequence prediction'
abstract: 'Reinforcement learning algorithms have achieved remarkable success in acquiring behavioral skills directly from pixel inputs. However, their application in real-world scenarios presents challenges due to their sensitivity to visual distractions (e.g., changes in viewpoint and light). A key factor contributing to this challenge is that the learned representations often suffer from overfitting task-irrelevant information. By comparing several representation learning methods, we find that the key to alleviating overfitting in representation learning is to choose proper prediction targets. Motivated by our comparison, we propose a novel representation learning approach—namely, reward sequence prediction (RSP)—that uses reward sequences or their transforms (e.g., discrete time Fourier transform) as prediction targets. RSP can efficiently learn robust representations as reward sequences rarely contain task-irrelevant information while providing a large number of supervised signals to accelerate representation learning. An appealing feature is that RSP makes no assumption about the type of distractions and thus can improve performance even when multiple types of distractions exist. We evaluate our approach in Distracting Control Suite. Experiments show that our method achieves state-of-the-art sample efficiency and generalization ability in tasks with distractions.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhou23a.html
PDF: https://proceedings.mlr.press/v216/zhou23a/zhou23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhou23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Qi
family: Zhou
- given: Jie
family: Wang
- given: Qiyuan
family: Liu
- given: Yufei
family: Kuang
- given: Wengang
family: Zhou
- given: Houqiang
family: Li
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2551-2562
id: zhou23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2551
lastpage: 2562
published: 2023-07-02 00:00:00 +0000
- title: 'Convergence rates for localized actor-critic in networked Markov potential games'
abstract: 'We introduce a class of networked Markov potential games where agents are associated with nodes in a network. Each agent has its own local potential function, and the reward of each agent depends only on the states and actions of agents within a neighborhood. In this context, we propose a localized actor-critic algorithm. The algorithm is scalable since each agent uses only local information and does not need access to the global state. Further, the algorithm overcomes the curse of dimensionality through the use of function approximation. Our main results provide finite-sample guarantees up to a localization error and a function approximation error. Specifically, we achieve an $\tilde{\mathcal{O}}(\tilde{\epsilon}^{-4})$ sample complexity measured by the averaged Nash regret. This is the first finite-sample bound for multi-agent competitive games that does not depend on the number of agents.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhou23b.html
PDF: https://proceedings.mlr.press/v216/zhou23b/zhou23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhou23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Zhaoyi
family: Zhou
- given: Zaiwei
family: Chen
- given: Yiheng
family: Lin
- given: Adam
family: Wierman
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2563-2573
id: zhou23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2563
lastpage: 2573
published: 2023-07-02 00:00:00 +0000
- title: 'AUC Maximization in Imbalanced Lifelong Learning'
abstract: 'Imbalanced data is ubiquitous in machine learning, such as medical or fine-grained image datasets. The existing continual learning methods employ various techniques such as balanced sampling to improve classification accuracy in this setting. However, classification accuracy is not a suitable metric for imbalanced data, and hence these methods may not obtain a good classifier as measured by other metrics (e.g., Area under the ROC Curve). In this paper, we propose a solution to enable efficient imbalanced continual learning by designing an algorithm to effectively maximize one widely used metric in an imbalanced data setting: Area Under the ROC Curve (AUC). We find that simply replacing accuracy with AUC will cause *gradient interference problem* due to the imbalanced data distribution. To address this issue, we propose a new algorithm, namely DIANA, which performs a novel synthesis of model **D**ecoupl**I**ng **AN**d **A**lignment. In particular, the algorithm updates two models simultaneously: one focuses on learning the current knowledge while the other concentrates on reviewing previously-learned knowledge, and the two models gradually align during training. The results show that the proposed DIANA achieves state-of-the-art performance on all the imbalanced datasets compared with several competitive baselines.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhu23a.html
PDF: https://proceedings.mlr.press/v216/zhu23a/zhu23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhu23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Xiangyu
family: Zhu
- given: Jie
family: Hao
- given: Yunhui
family: Guo
- given: Mingrui
family: Liu
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2574-2585
id: zhu23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2574
lastpage: 2585
published: 2023-07-02 00:00:00 +0000
- title: 'Private Prediction Strikes Back! Private Kernelized Nearest Neighbors with Individual Rényi Filter'
abstract: 'Most existing approaches of differentially private (DP) machine learning focus on *private training*. Despite its many advantages, *private training* lacks the flexibility in adapting to incremental changes to the training dataset such as deletion requests from exercising GDPR’s right to be forgotten. We revisit a long-forgotten alternative, known as *private prediction*, and propose a new algorithm named *Individual Kernelized Nearest Neighbor* (Ind-KNN). Ind-KNN is easily updatable over dataset changes and it allows precise control of the Rényi DP at an individual user level — a user’s privacy loss is measured by the exact amount of her contribution to predictions; and a user is removed if her prescribed privacy budget runs out. Our results show that Ind-KNN consistently improves the accuracy over existing private prediction methods for a wide range of epsilon on four vision and language tasks. We also illustrate several cases under which Ind-KNN is preferable over private training with NoisySGD.'
volume: 216
URL: https://proceedings.mlr.press/v216/zhu23b.html
PDF: https://proceedings.mlr.press/v216/zhu23b/zhu23b.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zhu23b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yuqing
family: Zhu
- given: Xuandong
family: Zhao
- given: Chuan
family: Guo
- given: Yu-Xiang
family: Wang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2586-2596
id: zhu23b
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2586
lastpage: 2596
published: 2023-07-02 00:00:00 +0000
- title: 'MixupE: Understanding and improving Mixup from directional derivative perspective'
abstract: 'Mixup is a popular data augmentation technique for training deep neural networks where additional samples are generated by linearly interpolating pairs of inputs and their labels. This technique is known to improve the generalization performance in many learning paradigms and applications. In this work, we first analyze Mixup and show that it implicitly regularizes infinitely many directional derivatives of all orders. Based on this new insight, we propose an improved version of Mixup, theoretically justified to deliver better generalization performance than the vanilla Mixup. To demonstrate the effectiveness of the proposed method, we conduct experiments across various domains such as images, tabular data, speech, and graphs. Our results show that the proposed method improves Mixup across multiple datasets using a variety of architectures, for instance, exhibiting an improvement over Mixup by 0.8% in ImageNet top-1 accuracy.'
volume: 216
URL: https://proceedings.mlr.press/v216/zou23a.html
PDF: https://proceedings.mlr.press/v216/zou23a/zou23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zou23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Yingtian
family: Zou
- given: Vikas
family: Verma
- given: Sarthak
family: Mittal
- given: Wai Hoh
family: Tang
- given: Hieu
family: Pham
- given: Juho
family: Kannala
- given: Yoshua
family: Bengio
- given: Arno
family: Solin
- given: Kenji
family: Kawaguchi
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2597-2607
id: zou23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2597
lastpage: 2607
published: 2023-07-02 00:00:00 +0000
- title: 'Regularized online DR-submodular optimization'
abstract: 'The utilization of online optimization techniques is prevalent in many fields of artificial intelligence, enabling systems to continuously learn and adjust to their surroundings. This paper outlines a regularized online optimization problem, where the regularizer is defined on the average of the actions taken. The objective is to maximize the sum of rewards and the regularizer value while adhering to resource constraints, where the reward function is assumed to be DR-submodular. Both concave and DR-submodular regularizers are analyzed. Concave functions are useful in describing the impartiality of decisions, while DR-submodular functions can be employed to represent the overall effect of decisions on all relevant parties. We have developed two algorithms for each of the concave and DR-submodular regularizers. These algorithms are easy to implement, efficient, and produce sublinear regret in both cases. The performance of the proposed algorithms and regularizers has been verified through numerical experiments in the context of internet advertising.'
volume: 216
URL: https://proceedings.mlr.press/v216/zuo23a.html
PDF: https://proceedings.mlr.press/v216/zuo23a/zuo23a.pdf
edit: https://github.com/mlresearch//v216/edit/gh-pages/_posts/2023-07-02-zuo23a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence'
publisher: 'PMLR'
author:
- given: Pengyu
family: Zuo
- given: Yao
family: Wang
- given: Shaojie
family: Tang
editor:
- given: Robin J.
family: Evans
- given: Ilya
family: Shpitser
page: 2608-2617
id: zuo23a
issued:
date-parts:
- 2023
- 7
- 2
firstpage: 2608
lastpage: 2617
published: 2023-07-02 00:00:00 +0000