检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Lawrence, Hannah Portilheiro, Vasco Zhang, Yan Kaba, Sékou-Oumar Department of Electrical Engineering and Computer Science MIT United States Gatsby Computational Neuroscience Unit UCL United Kingdom Samsung – SAIT AI Lab Montreal Canada Mila – Quebec Artficial Intelligence Institute Canada McGill University Canada

Equivariance encodes known symmetries into neural networks, often enhancing generalization. However, equivariant networks cannot break symmetries: the output of an equivariant network must, by definition, have at least the same self-symmetries as the input. This poses an important problem, both (1) for prediction tasks on domains where self-symmetries are common, and (2) for generative models, which must break symmetries in order to reconstruct from highly symmetric latent spaces. This fundamental limitation can be addressed by considering equivariant conditional distributions, instead of equivariant functions. We present novel theoretical results that establish necessary and sufficient conditions for representing such distributions. Concretely, this representation provides a practical framework for breaking symmetries in any equivariant network via randomized canonicalization. Our method, SymPE (Symmetry-breaking Positional Encodings), admits a simple interpretation in terms of positional encodings. This approach expands the representational power of equivariant networks while retaining the inductive bias of symmetry, which we justify through generalization bounds. Experimental results demonstrate that SymPE significantly improves performance of group-equivariant and graph neural networks across diffusion models for graphs, graph autoencoders, and lattice spin system modeling. © 2025, CC BY-NC-SA.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Efficient Leverage Score Sampling for Tensor Train Decomposition

arXiv

引用

arXiv 2024年

作者： Bharadwaj, Vivek Rakhshan, Beheshteh T. Malik, Osman Asif Rabusseau, Guillaume Electrical Engineering and Computer Science Department UC Berkeley United States Computational Research Division Lawrence Berkeley National Lab United States Mila DIRO Université de Montréal Canada CIFAR AI Canada

Tensor Train (TT) decomposition is widely used in the machine learning and quantum physics communities as a popular tool to efficiently compress high-dimensional tensor data. In this paper, we propose an efficient algorithm to accelerate computing the TT decomposition with the Alternating Least Squares (ALS) algorithm relying on exact leverage scores sampling. For this purpose, we propose a data structure that allows us to efficiently sample from the tensor with time complexity logarithmic in the tensor size. Our contribution specifically leverages the canonical form of the TT decomposition. By maintaining the canonical form through each iteration of ALS, we can efficiently compute (and sample from) the leverage scores, thus achieving significant speed-up in solving each sketched least-square problem. Experiments on synthetic and real data on dense and sparse tensors demonstrate that our method outperforms SVD-based and ALS-based algorithms. © 2024, CC BY-NC-SA.

关键词： Tensors

来源：评论

学校读者我要写书评

暂无评论

Editorial

引用

International Journal of Social robotics 2023年第1期15卷 1-1页

作者： Ivaldi, Serena Wykowska, Agnieszka Khatib, Oussama Inria/CNRS/Université de Lorraine Nancy France IstitutoItaliano di Tecnologia Genova16153 Italy Robotics Lab Computer Science Department Stanford University StanfordCA United States

来源：评论

学校读者我要写书评

暂无评论

Physics Representation Learning for Dexterous Manipulation Planning

Physics Representation Learning for Dexterous Manipulation P...

引用

IEEE International Workshop on Robot and Human Communication (ROMAN)

作者： Yi Wu Mengsha Hu Runxiang Jin Rui Liu Informatique et Sciences du Numérique (Computer Research Laboratory) CentraléSupelec Paris-Saclay University Gif-Sur-Yvette France Cognitive Robotics and AI Lab (CRAI) College of Aeronautics and Engineering Department of Computer Science Kent State University Kent OH USA

ISBN: (数字)9798350375022

ISBN: (纸本)9798350375039

Dexterous manipulation in robotics, particularly with high degrees of freedom (DoF) devices like the 24-joint Shadow Hand, confronts complexities in search space and execution precision. Humans, however, manipulate objects effortlessly, thanks to their innate grasp of physics. Inspired by this, we introduce the Physics Representation Learning (PRL) framework for robotic hand manipulation. PRL uses physics principles for action conceptualization, such as aligning axes at specific angles (Figure 1). It is a natural language that describes the physics laws, thus bridging the gap between high-level semantics planning and joint-level execution through inverse kinematics (IK). Built on top of the physics-informed action space, PRL deploys a Reinforcement Learning (RL) network with expert demonstrations. Its effectiveness was validated in Nvidia Isaac Sim on four tasks: hammering, unscrewing, sweeping, and pinching. Results show that PRL significantly outperforms conventional joint-control algorithms and RL without demonstrations, underscoring the benefits of physics-based action representations for complex tasks.

关键词： Representation learning Semantics Natural languages Reinforcement learning Kinematics Planning Complexity theory Physics Robots

来源：评论

学校读者我要写书评

暂无评论

Towards Analysis and Interpretation of Large Language Models for Arithmetic Reasoning

Towards Analysis and Interpretation of Large Language Models...

引用

Swiss Conference on Data science (SDS)

作者： Mst. Shapna Akter Hossain Shahriar Alfredo Cuzzocrea Department of Intelligent Systems and Robotics University of West Florida Pensacola FL USA Center for Cybersecurity University of West Florida Pensacola FL USA iDEA Lab University of Calabria Rende Italy Department of Computer Science University of Paris City Paris France

ISBN: (数字)9798350309294

ISBN: (纸本)9798350309300

Large Language Models (LLMs) have recently conquered the research scene, with particular regards to the Transformer architecture in the context of arithmetic reasoning. In this so-delineated scenario, this paper puts the basis for a causal mediation analysis about the approach of Transformer-based LLMs to complex arithmetic problems. In particular, we try to discover which parameters are crucial for complex reasoning tasks such as model activations. Our preliminary results state that, for complex arithmetic operations, information is channeled from mid-layer activations to the final token through enhanced attention mechanisms. Preliminary experiments are reported.

关键词： Attention mechanisms Large language models Data science Transformers Cognition Mediation Arithmetic

来源：评论

学校读者我要写书评

暂无评论

Explaining Black-Box Predictions by Generating Local Meaningful Perturbations 3

Explaining Black-Box Predictions by Generating Local Meaning...

引用

3rd IEEE International Conference on Transdisciplinary AI, TransAI 2021

作者： Verma, Tejaswani Lingenfelder, Christoph Klakow, Dietrich Saarbrücken Graduate School of Computer Science Saarland University Mercedes-Benz Innovation Lab Berlin Germany Berlin Germany Department of Computational Linguistics Saarland University Germany

ISBN: (纸本)9781665434126

Generating explanations of predictions made by machine learning models is a difficult task, especially for black-box models. One possible way to explain an individual decision or recommendation for a given instance is to build an interpretable local surrogate for the underlying black-box model in the vicinity of the given instance. This approach has been adopted by many algorithms, for example LIME and LEAFAGE. These algorithms suffer from shortcomings, strict assumptions and prerequisites, which not only limit their applicability but also affect black-box fidelity of their local approximations. We present ways to overcome their shortcomings including the definition of neighborhood, removal of prerequisites and assumption of linearity in local model. The main contribution of this paper is a novel algorithm (LEMP) which provides explanation for the given instance by building a surrogate model using generated perturbations in the neighborhood of the given instance as training data. Experiments show that our approach is more widely applicable and generates interpretable models with better fidelity to the underlying black-box model than previous algorithms. © 2021 IEEE.

关键词： Approximation algorithms

来源：评论

学校读者我要写书评

暂无评论

Robust Cross-Domain Speaker Verification with Multi-Level Domain Adapters

Robust Cross-Domain Speaker Verification with Multi-Level Do...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Wen Huang Bing Han Shuai Wang Zhengyang Chen Yanmin Qian AI Institute Department of Computer Science and Engineering Auditory Cognition and Computational Acoustics Lab MoE Key Lab of Artificial Intelligence Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

Speaker verification encounters significant challenges when confronted with diverse domain data, often resulting in performance degradation due to domain mismatch. To enhance performance in cross-domain scenarios, we introduce the Domain Adapter, an adaptable module designed for specific domains. This module learns and integrates domain-specific information with speaker-related data, mitigating domain-related variations and promoting convergence of utterance embeddings from the same speaker across diverse domains. It offers configurability across multiple levels and is adaptable to various backbone architectures. Our proposed module substantially enhances cross-domain performance with minimal parameter increments while effectively generalizing to previously unseen domains. In our experiments, we present results on the 3D-Speaker dataset, which provides acoustically-relevant attributes crucial for domain categorization and the subsequent learning of domain information. The top-performing system integrated with domain adapters achieved 10.8%, 14.8%, and 21.1% EER improvements over the baseline across three 3D-Speaker dataset trials.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multi-State-Action Tokenisation in Decision Transformers for Multi-Discrete Action Spaces

arXiv

引用

arXiv 2024年

作者： Moodley, Perusha Kaushik, Pramod Thambi, Dhillu Trovinger, Mark Paruchuri, Praveen Hong, Xia Rosman, Benjamin School of Mathematical Physical and Computational Sciences University of Reading Reading United Kingdom TCS Research Pune India Machine Learning Lab IIIT Hyderabad Hyderabad India Department of Computer Science and Engineering Mississippi State University Mississippi United States School of Computer Science and Applied Mathematics University of Witwatersrand Johannesburg South Africa

Decision Transformers, in their vanilla form, struggle to perform on image-based environments with multi-discrete action spaces. Although enhanced Decision Transformer architectures have been developed to improve performance, these methods have not specifically addressed this problem of multi-discrete action spaces which hampers existing Decision Transformer architectures from learning good representations. To mitigate this, we propose Multi-State Action Tokenisation (M-SAT), an approach for tokenising actions in multi-discrete action spaces that enhances the model’s performance in such environments. Our approach involves two key changes: disentangling actions to the individual action level and tokenising the actions with auxiliary state information. These two key changes also improve individual action level interpretability and visibility within the attention layers. We demonstrate the performance gains of M-SAT on challenging ViZDoom environments with multi-discrete action spaces and image-based state spaces, including the Deadly Corridor and My Way Home scenarios, where M-SAT outperforms the baseline Decision Transformer without any additional data or heavy computational overheads. Additionally, we find that removing positional encoding does not adversely affect M-SAT’s performance and, in some cases, even improves it. © 2024, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prototype and Instance Contrastive Learning for Unsupervised Domain Adaptation in Speaker Verification

arXiv

引用

arXiv 2024年

作者： Huang, Wen Han, Bing Chen, Zhengyang Wang, Shuai Qian, Yanmin Auditory Cognition and Computational Acoustics Lab MoE Key Lab of Artificial Intelligence AI Institute Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

Speaker verification system trained on one domain usually suffers performance degradation when applied to another domain. To address this challenge, researchers commonly use feature distribution matching-based methods in unsupervised domain adaptation scenarios where some unlabeled target domain data is available. However, these methods often have limited performance improvement and lack generalization in various mismatch situations. In this paper, we propose Prototype and Instance Contrastive Learning (PICL), a novel method for unsupervised domain adaptation in speaker verification through dual-level contrastive learning. For prototype contrastive learning, we generate pseudo labels via clustering to create dynamically updated prototype representations, aligning instances with their corresponding class or cluster prototypes. For instance contrastive learning, we minimize the distance between different views or augmentations of the same instance, ensuring robust and invariant representations resilient to variations like noise. This dual-level approach provides both high-level and low-level supervision, leading to improved generalization and robustness of the speaker verification model. Unlike previous studies that only evaluated mismatches in one situation, we have conducted relevant explorations on various datasets and achieved state-of-the-art performance currently, which also proves the generalization of our method. Copyright © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Enhancing Post-Stroke Upper Limb Rehabilitation Through Haptic Feedback in Virtual Reality-Based Gaming

Enhancing Post-Stroke Upper Limb Rehabilitation Through Hapt...

引用

IEEE International Conference on robotics and Biomimetics

作者： Branesh M. Pillai Napas Siripala Kongtup Sai-Aroon Dileep Sivaraman Ramana Vinjamuri Sorayouth Chumnanvej Jackrit Suthakorn Department of Biomedical Engineering Faculty of Engineering Center for Biomedical and Robotics Technology (BART LAB) Mahidol University Thailand Vinjamuri lab in the Department of Computer Science and Electrical Engineering University of Maryland Baltimore County Baltimore MD USA Department of Surgery Neurosurgery Division Faculty of Medicine Ramathibodi Hospital Mahidol University Bangkok Thailand

ISBN: (数字)9798331509644

ISBN: (纸本)9798331509651

This preliminary study explores developing and implementing upper-limb rehabilitation tools utilizing haptic feedback and 3D spatial recognition analysis in a virtual reality (VR) environment. The effectiveness of upper limb rehabilitation in post-stroke patients was investigated by analyzing the joint and gimbal range of motion (ROM), ROM ratios, and force distribution through various drawing tasks. This study evaluated the impact of different drawing patterns on joint and gimbal ROM, and the force applied during rehabilitation exercises. This study revealed significant variability in ROM across different tasks by combining ROM measurements with force data and employing heat maps for visual representation. These findings emphasize the need for task-specific adjustments to improve rehabilitation outcomes. This study highlights the potential of incorporating advanced technologies, such as VR-based tools with haptic feedback and 3D spatial recognition, to enhance patient engagement and rehabilitation effectiveness.

关键词： Limbs Visualization Three-dimensional displays Tracking Force Medical treatment Virtual reality Haptic interfaces Motion control Read only memory

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：