检索结果-内蒙古大学图书馆

Constrained learning with Non-Convex Losses

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Chamon, Luiz F.O. Paternain, Santiago Calvo-Fullana, Miguel Ribeiro, Alejandro The Simons Institute for the Theory of Computation University of California Berkeley United States The Department of Electrical Computer and Systems Engineering Rensselaer Polytechnic Institute United States The Department of Aeronautics and Astronautics Massachusetts Institute of Technology United States The Department of Electrical and Systems Engineering University of Pennsylvania United States

Though learning has become a core component of modern information processing, there is now ample evidence that it can lead to biased, unsafe, and prejudiced systems. The need to impose requirements on learning is therefore paramount, especially as it reaches critical applications in social, industrial, and medical domains. However, the non-convexity of most modern statistical problems is only exacerbated by the introduction of constraints. Whereas good unconstrained solutions can often be learned using empirical risk minimization, even obtaining a model that satisfies statistical constraints can be challenging. All the more so, a good one. In this paper, we overcome this issue by learning in the empirical dual domain, where constrained statistical learning problems become unconstrained and deterministic. We analyze the generalization properties of this approach by bounding the empirical duality gap—i.e., the difference between our approximate, tractable solution and the solution of the original (non-convex) statistical problem—and provide a practical constrained learning algorithm. These results establish a constrained counterpart to classical learning theory, enabling the explicit use of constraints in learning. We illustrate this theory and algorithm in rate-constrained learning applications arising in fairness and adversarial robustness. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

A relaxed technical assumption for posterior sampling-based reinforcement learning for control of unknown linear systems

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Gagrani, Mukul Sudhakara, Sagar Mahajan, Aditya Nayyar, Ashutosh Ouyang, Yi Qualcomm AI research San Diego United States The Department of Electrical and Computer Engineering University of Southern California Los AngelesCA United States The department of Electrical and Computer Engineering McGill University MontrealQC Canada Preferred Networks America BurlingameCA United States

We revisit the Thompson sampling-based learning algorithm for controlling an unknown linear system with quadratic cost proposed in [1]. This algorithm operates in episodes of dynamic length and it is shown to have a regret bound of Õ(√T), where T is the time-horizon. The regret bound of this algorithm is obtained under a technical assumption on the induced norm of the closed loop system. We propose a variation of this algorithm that enforces a lower bound Tmin on the episode length. We show that a careful choice of Tmin (that depends on the uncertainty about the system model) allows us to recover the Õ(√T) regret bound under a milder technical condition about the closed loop system. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Emerging Trends in Federated learning: From Model Fusion to Federated X learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Ji, Shaoxiong Tan, Yue Saravirta, Teemu Yang, Zhiqin Liu, Yixin Vasankari, Lauri Pan, Shirui Long, Guodong Walid, Anwar University of Helsinki Finland University of Technology Sydney Australia Aalto University Finland Beihang University China Monash University Australia Griffith University Australia Amazon United States Columbia University United States

Federated learning is a new learning paradigm that decouples data collection and model training via multi-party computation and model aggregation. As a flexible learning setting, federated learning has the potential to integrate with other learning frameworks. We conduct a focused survey of federated learning in conjunction with other learning algorithms. Specifically, we explore various learning algorithms to improve the vanilla federated averaging algorithm and review model fusion methods such as adaptive aggregation, regularization, clustered methods, and Bayesian methods. Following the emerging trends, we also discuss federated learning in the intersection with other learning paradigms, termed federated X learning, where X includes multitask learning, meta-learning, transfer learning, unsupervised learning, and reinforcement learning. In addition to reviewing state-of-the-art studies, this paper also identifies key challenges and applications in this field, while also highlighting promising future directions. © 2021, CC BY.

关键词： learning algorithms

On the Benefits of Inducing Local Lipschitzness for Robust Generative Adversarial Imitation learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Memarian, Farzan Hashemi, Abolfazl Niekum, Scott Topcu, Ufuk NVIDIA Corporation United States Purdue University United States The University of Massachusetts Amherst United States The University of Texas Austin United States

We explore methodologies to improve the robustness of generative adversarial imitation learning (GAIL) algorithms to observation noise. Towards this objective, we study the effect of local Lipschitzness of the discriminator and the generator on the robustness of policies learned by GAIL. In many robotics applications, the learned policies by GAIL typically suffer from a degraded performance at test time since the observations from the environment might be corrupted by noise. Hence, robustifying the learned policies against the observation noise is of critical importance. To this end, we propose a regularization method to induce local Lipschitzness in the generator and the discriminator of adversarial imitation learning methods. We show that the modified objective leads to learning significantly more robust policies. Moreover, we demonstrate - both theoretically and experimentally - that training a locally Lipschitz discriminator leads to a locally Lipschitz generator, thereby improving the robustness of the resultant policy. We perform extensive experiments on simulated robot locomotion environments from the MuJoCo suite that demonstrate the proposed method learns policies that significantly outperform the state-of-the-art generative adversarial imitation learning algorithm when applied to test scenarios with noise-corrupted observations. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Winning the ICCV'2021 VALUE challenge: Task-aware ensemble and transfer learning with visual concepts

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Shin, Minchul Mun, Jonghwan On, Kyoung-Woon Kang, Woo-Young Han, Gunsoo Kim, Eun-Sol Kakao Brain Hanyang University

The VALUE (Video-And-Language Understanding Evaluation) benchmark is newly introduced to evaluate and analyze multi-modal representation learning algorithms on three video-and-language tasks: Retrieval, QA, and Captioning. The main objective of the VALUE challenge is to train a task-agnostic model that is simultaneously applicable for various tasks with different characteristics. This technical report describes our winning strategies for the VALUE challenge: 1) single model optimization, 2) transfer learning with visual concepts, and 3) task-aware ensemble. The first and third strategies are designed to address heterogeneous characteristics of each task, and the second one is to leverage rich and fine-grained visual information. We provide a detailed and comprehensive analysis with extensive experimental results. Based on our approach, we ranked first place on the VALUE and QA phases for the competition. © 2021, CC BY.

关键词： learning algorithms

Distributional Shift Adaptation using Domain-Specific Features

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Tahir, Anique Cheng, Lu Guo, Ruocheng Liu, Huan Arizona State University TempeAZ United States University of Illinois Chicago ChicagoIL United States Bytedance AI Lab London United Kingdom

Machine learning algorithms typically assume that the training and test samples come from the same distributions, i.e., in-distribution. However, in open-world scenarios, streaming big data can be Out-Of-Distribution (OOD), rendering these algorithms ineffective. Prior solutions to the OOD challenge seek to identify invariant features across different training domains. The underlying assumption is that these invariant features should also work reasonably well in the unlabeled target domain. By contrast, this work is interested in the domain-specific features that include both invariant features and features unique to the target domain. We propose a simple yet effective approach that relies on correlations in general regardless of whether the features are invariant or not. Our approach uses the most confidently predicted samples identified by an OOD base model (teacher model) to train a new model (student model) that effectively adapts to the target domain. Empirical evaluations on benchmark datasets show that the performance is improved over the SOTA by ∼10-20%1 © 2022, CC BY-NC-ND.

关键词： learning algorithms

Improved Crow Search Algorithm for Optimal Flexible Process Planning

学校读者我要写书评

暂无评论

SSRN

SSRN 2022年

作者： Avalos, Omar Haro, Eduardo H. Camarena, Octavio Díaz, Primitivo Departamento de Electrónica Universidad de Guadalajara CUCEI Av. Revolución 1500 Jal Guadalajara Mexico

The enhancement of manufacturing processes due to the growing demand for products and services is a problem that must be constantly updated to fulfill such requests. Under such circumstances, several schemas have been conducted to improve the manufacturing processes, such as the flexible use of machines, tools, and tools access directions to determine the best process planning (PP) for a specific piece or part. Including different tools and machine combinations for a particular process becomes the application in a complicated combinatory problem that can be treated as an optimization problem. On the other hand, bioinspired optimization techniques have been widely employed for solving several complex engineering and real-life problems showing competitive results. The crow search algorithm (CSA) is an optimization technique based on the intelligent foraging behavior of crows. This work uses the improved version of the CSA to determine the optimal flexible process planning (FPP) with a low computational effort. Several AND/OR networks FPP problems are employed in the experimentation stage, where the production cost and production are considered fitness functions for this study. Moreover, to determine the performance and accuracy of the proposed technique, different results previously reported in the literature are used in the experimentation stage to compare the proposal in which the ICSA-FPP is able to outperform similar approaches. © 2022, The Authors. All rights reserved.

关键词： learning algorithms

EDU-level Extractive Summarization with Varying Summary Lengths

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wu, Yuping Tseng, Ching-Hsun Shang, Jiayu Mao, Shengzhong Nenadic, Goran Zeng, Xiao-Jun Department of Computer Science University of Manchester United Kingdom

Extractive models usually formulate text summarization as extracting fixed top-k salient sentences from the document as a summary. Few works exploited extracting finer-grained Elementary Discourse Unit (EDU) with little analysis and justification for the extractive unit selection. Further, the selection strategy of the fixed top-k salient sentences fits the summarization need poorly, as the number of salient sentences in different documents varies and therefore a common or best k does not exist in reality. To fill these gaps, this paper first conducts the comparison analysis of oracle summaries based on EDUs and sentences, which provides evidence from both theoretical and experimental perspectives to justify and quantify that EDUs make summaries with higher automatic evaluation scores than sentences. Then, considering this merit of EDUs, this paper further proposes an EDU-level extractive model with Varying summary Lengths (EDUVL1) and develops the corresponding learning algorithm. EDU-VL learns to encode and predict probabilities of EDUs in the document, generate multiple candidate summaries with varying lengths based on various k values, and encode and score candidate summaries, in an end-to-end training manner. Finally, EDU-VL is experimented on single and multi-document benchmark datasets and shows improved performances on ROUGE scores in comparison with state-of-the-art extractive models, and further human evaluation suggests that EDU-constituent summaries maintain good grammaticality and readability. © 2022, CC BY-NC-SA.

关键词： learning algorithms

FAIR GROUP-SHARED REPRESENTATIONS WITH NORMALIZING FLOWS

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Cerrato, Mattia Köppel, Marius Segner, Alexander Kramer, Stefan Johannes Gutenberg-Universität Mainz Saarstraße 21 Mainz Germany

The issue of fairness in machine learning stems from the fact that historical data often displays biases against specific groups of people which have been underprivileged in the recent past, or still are. In this context, one of the possible approaches is to employ fair representation learning algorithms which are able to remove biases from data, making groups statistically indistinguishable. In this paper, we instead develop a fair representation learning algorithm which is able to map individuals belonging to different groups in a single group. This is made possible by training a pair of Normalizing Flow models and constraining them to not remove information about the ground truth by training a ranking or classification model on top of them. The overall, "chained" model is invertible and has a tractable Jacobian, which allows to relate together the probability densities for different groups and "translate" individuals from one group to another. We show experimentally that our methodology is competitive with other fair representation learning algorithms. Furthermore, our algorithm achieves stronger invariance w.r.t. the sensitive attribute. © 2022, CC BY.

关键词： learning algorithms