检索结果-内蒙古大学图书馆

Learn to explain transformer via interpretation path by reinforcement learning

Neural Networks 2025年 188卷 107496页

作者： Niu, Runliang Wang, Qi Kong, He Xing, Qianli Chang, Yi Yu, Philip S. School of Artificial Intelligence Jilin University ChangChun130012 China Engineering Research Center of Knowledge-Driven Human–Machine Intelligence Ministry of Education China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University ChangChun China Department of Computer Science University of Illinois at Chicago United States

In recent years, the Transformer model has become a key part of many AI systems, making it important to understand how it works. The large parameter size and complex structure of the Transformer make interpretation more difficult and less efficient. Fortunately, there are many internal variables in Transformers that can aid in the explanation process, including attention matrices, gradients, hidden states, and activations between layers. Effectively utilizing these internal factors can help us better understand the decision-making process of Transformers. Also, most existing works focus on one type of these features and fail to investigate the interpretability of different variables in a unified model. To address these issues, this paper introduces a Reinforcement Learning environment in which the agent makes step-by-step modifications to input sequences to construct perturbed samples and gradually reduces the model's confidence in classification labels. The environment guides the agent to choose the token modification strategy along a more targeted interpretation path instead of random sampling, which can significantly improve the interpretation effectiveness. The flexibly designed agent can utilize multiple internal variables or even combinations of variables as observation, allowing for comparisons of their contributions to the model's interpretability. Extensive experiments conducted on three real-world datasets demonstrate the superior performance of our proposed model in both model interpretation and model adversarial attack tasks. We obtain a set of interesting findings, which can inspire further research on Transformer and Transformer-based model interpretation works. The code of this paper is available at https://***/niuzaisheng/Learn-to-Explain. © 2025

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

ADAC: Actor-Double-Attention-Critic for Multi-Agent Cooperation in Mixed Cooperative-Competitive Environments

引用

IEEE Transactions on Intelligent Transportation Systems 2025年

作者： Kong, He Xing, Qianli Wang, Qi Niu, Runliang Chen, Hechang Wang, Yu Wang, Shiqi Duan, Zhiyi Chang, Yi Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineer of Ministry of Education School of Artificial Intelligence Jilin Changchun130012 China Engineering Research Center of Knowledge-Driven Human-Machine Intelligence Changchun130010 China Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineer of Ministry of Education College of Computer Science and Technology Jilin Changchun130012 China Jilin University School of Artificial Intelligence Changchun130012 China Ministry of Education Engineering Research Center of Knowledge-Driven Human-Machine Intelligence Changchun130010 China Jilin University College of Computer Science and Technology Jilin Changchun130012 China Inner Mongolia University Department of Computer Science Hohhot010021 China

The cooperation in mixed cooperative-competitive tasks has drawn significant attention in multi-agent deep reinforcement learning. Agents need to cooperate with their teammates while competing against their opponents. However, most existing works treat the cooperative agents and competitive agents equally as they perform the same operation on all the agents. As a result, without distinguishing between cooperative and competitive agents, they may suffer from information disorder in learning an optimally cooperative policy and struggle to decide on the next step action. To address the above issues, we decompose the final Q-value into a weighted combination of three parts: the Q-values of the cooperative agents, the competitive group, and the current agent. A theoretical proof of the correctness of the decomposition is provided. With this decomposition, we are able to consider cooperative and competitive agents separately. Accordingly, we propose a multi-agent actor-critic algorithm called actor-double-attention-critic (ADAC) under centralized training and decentralized execution according to the decomposition. In ADAC, networks with group-specific attention and an attentional weighting network are specially designed. With the designed double-attention structure, ADAC can capture the distributions from different agents and improve cooperation performance. Extensive experiments are conducted in three scenarios with nine settings against six representative methods. The results demonstrate the superiority of the proposed ADAC model against state-of-the-art methods in various mixed cooperative-competitive tasks. © 2000-2011 IEEE.

关键词： cooperation group-specific attention mixed tasks Multi-agent reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Representation Auto-fused NMF based Hierarchical Clustering

引用

Expert Systems with Applications 2025年 283卷

作者： Lin, Yunxia Hu, Haoyu Li, Bentian Zhao, Shuchang Jing, Hengchang School of Information Engineering Yangzhou University Yangzhou225127 China Jiangsu Province Engineering Research Center of Knowledge Management and Intelligent Service Yangzhou University Yangzhou225127 China MIIT Key Laboratory of Pattern Analysis and Machine Intelligence Nanjing211106 China Institute of Intelligent Information Processing Taizhou University Taizhou318000 China Taizhou University Taizhou318000 China

Nonnegative matrix factorization (NMF) as a powerful representation learning technique has been extensively applied in clustering tasks across diverse domains, leading to the proliferation of various NMF-based clustering methods. However, most of these methods suffer from the suboptimal performance arising from the disjointed process of representation learning and clustering, and the necessity to manually predetermine the optimal number of clusters which is a critical factor affecting the quality of clustering results. To fill this gap and inspired by convex clustering, we propose a simple yet effective NMF-based clustering method named Representation Auto-fused NMF based Hierarchical Clustering (RAFNMF-HC) by incorporating the l2 norm penalty on the differences between pairs of representations into the NMF framework to foster the automatic fusion of representations during representation learning. As the fusion continuously proceeds, similar representations gradually coalesce into small clusters, which subsequently merge into larger ones until a largest cluster encompassing all representations is formed, enabling a joint representation learning and clustering paradigm. Moreover, a cluster hierarchy as a byproduct adaptively emerges without requiring a predefined number of clusters, thus providing us with interpretability of cluster structure of data to some degree. Additionally, RAFNMF-HC only incorporates a single hyperparameter, significantly simplifying the adjustment process and boosting its practical efficiency. Our proposed RAFNMF-HC method also boasts straightforward scalability to diverse NMF variants, enhancing its versatility. Extensive experiments on various real-world datasets demonstrate the superiority of RAFNMF-HC. © 2025 Elsevier Ltd

关键词： Hierarchical clustering

来源：评论

学校读者我要写书评

暂无评论

Iterative cross-training: An algorithm for web page categorization

引用

Intelligent Data Analysis 2003年第3期7卷 233-253页

作者： Soonthornphisaj, Nuanwan Kijsirikul, Boonserm Machine Intelligence and Knowledge Discovery Laboratory Department of Computer Engineering Chulalongkorn University Bangkok 10330 Thailand not available Bangkruay-Sainoi Rd. Bangbuathong Nonthaburi 11110 Thailand

The goal of Web page categorization is to classify Web documents into a certain number of predefined categories. Previous works in this area employed a large number of labeled training documents for supervised learning. The problem is that, it is difficult to create labeled training documents. Though it is not so easy to manually categorize unlabeled documents for creating training data, it is easy to collect unlabeled ones. Therefore, a new machine learning algorithm is investigated to overcome these difficulties and effectively utilize unlabeled documents. We propose a novel approach called Iterative Cross-Training (ICT). In this paper, we applied the algorithm to Web page categorization on three data sets. The performance of ICT was evaluated and analyzed with the supervised learning algorithms, Co-Training and Expectation Maximization. We found that ICT is considered to be an effective approach for the Web page categorization task. © 2003-IOS Press.

关键词： Websites

来源：评论

学校读者我要写书评

暂无评论

Image-point cloud registration method based on image matching 3

Image-point cloud registration method based on image matchin...

引用

3rd International Conference on machine Vision, Automatic Identification, and Detection, MVAID 2024

作者： Ai, Yang Yang, Xi School of Artificial Intelligence Jilin University Changchun China Engineering Research Center of Knowledge-Driven Human-Machine Intelligence MoE China Key Laboratory of Ancient Chinese Script Culture Relics and Artificial Intelligence Jilin University China

ISBN: (纸本)9781510681880

Image-to-point cloud registration is the task of fusing 2D images with 3D point clouds, typically combining the rich texture information from images with the spatial information from point clouds for a more comprehensive scene representation. However, the feature discrepancy between different modalities often makes it challenging to accurately match. Addressing this issue, this paper proposes an image-matching-based algorithm for image-to-point cloud registration. Initially, for cross-modal issues, we use 3D visualization tools to screenshot the 3D models, reading the camera information of the screenshots to establish a correspondence between the 3D models and 2D images. By employing a pretrained image matching model, the algorithm extracts keypoints and matches images of the screenshots and images, establishing image keypoints pairing relationships. Then, using the camera projection model, it maps the 3D coordinates to image keypoints, obtaining image-to-point cloud keypoint pairing relationships and calculating the transformation matrix between the image and 3D point cloud using PnP-RANSAC. The proposed method, compared with the manual annotations method, proves to significantly improve efficiency and stability in real-world scenarios. © 2024 SPIE.

关键词： Image matching

来源：评论

学校读者我要写书评

暂无评论

DIPR: Efficient Point Cloud Registration via Dynamic Iteration

arXiv

引用

arXiv 2023年

作者： Ai, Yang Bai, Qiang Li, Jindong Yang, Xi School of Artificial Intelligence Jilin University China Engineering Research Center of Knowledge-Driven Human-Machine Intelligence MoE China Key Laboratory of Ancient Chinese Script Culture Relics and Artificial Intelligence Jilin University China

Point cloud registration (PCR) is an essential task in 3D vision. Existing methods achieve increasingly higher accuracy. However, a large proportion of non-overlapping points in point cloud registration consume a lot of computational resources while negatively affecting registration accuracy. To overcome this challenge, we introduce a novel Efficient Point Cloud Registration via Dynamic Iteration framework, DIPR, that makes the neural network interactively focus on overlapping points based on sparser input points. We design global and local registration stages to achieve efficient course-to-fine processing. Beyond basic matching modules, we propose the Refined Nodes to narrow down the scope of overlapping points by using adopted density-based clustering to significantly reduce the computation amount. And our SC Classifier serves as an early-exit mechanism to terminate the registration process in time according to matching accuracy. Extensive experiments on multiple datasets show that our proposed approach achieves superior registration accuracy while significantly reducing computational time and GPU memory consumption compared to state-of-the-art methods. © 2023, CC BY-NC-SA.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Quaternion factorization machines: A lightweight solution to intricate feature interaction modelling

arXiv

引用

arXiv 2021年

作者： Chen, Tong Yin, Hongzhi Zhang, Xiangliang Huang, Zi Wang, Yang Wang, Meng The School of Information Technology and Electrical Engineering The University of Queensland Australia The Machine Intelligence and Knowledge Engineering Laboratory King Abdullah University of Science and Technology Saudi Arabia The Key Laboratory of Knowledge Engineering with Big Data Ministry of Education School of Computer Science and Information Engineering Hefei University of Technology China

Due to the sparsity of available features in web-scale predictive analytics, combinatorial features become a crucial means for deriving accurate predictions. As a well-established approach, factorization machine (FM) is capable of automatically learning high-order interactions among features to make predictions without the need for manual feature engineering. With the prominent development of deep neural networks (DNNs), there is a recent and ongoing trend of enhancing the expressiveness of FM-based models with DNNs. However, though better results are obtained with DNN-based FM variants, such performance gain is paid off by an enormous amount (usually millions) of excessive model parameters on top of the plain FM. Consequently, the heavy parameterization impedes the real-life practicality of those deep models, especially efficient deployment on resource-constrained IoT and edge devices. In this paper, we move beyond the traditional real space where most deep FM-based models are defined, and seek solutions from quaternion representations within the hypercomplex space. Specifically, we propose the quaternion factorization machine (QFM) and quaternion neural factorization machine (QNFM), which are two novel lightweight and memory-efficient quaternion-valued models for sparse predictive analytics. By introducing a brand new take on FM-based models with the notion of quaternion algebra, our models not only enable expressive inter-component feature interactions, but also significantly reduce the parameter size due to lower degrees of freedom in the hypercomplex Hamilton product compared with real-valued matrix multiplication. Extensive experimental results on three large-scale datasets demonstrate that QFM achieves 4.36% performance improvement over the plain FM without introducing any extra parameters, while QNFM outperforms all baselines with up to two magnitudes' parameter size reduction in comparison to state-of-the-art peer methods. Copyright © 2021, The Authors. All

关键词： Large dataset

来源：评论

学校读者我要写书评

暂无评论

Uniting heterogeneity, inductiveness, and efficiency for graph representation learning

arXiv

引用

arXiv 2021年

作者： Chen, Tong Yin, Hongzhi Ren, Jie Huang, Zi Zhang, Xiangliang Wang, Hao The School of Information Technology and Electrical Engineering The University of Queensland Australia The Machine Intelligence and Knowledge Engineering Laboratory King Abdullah University of Science and Technology Saudi Arabia Alibaba AI Labs China

With the ubiquitous graph-structured data in various applications, models that can learn compact but expressive vector representations of nodes have become highly desirable. Recently, bearing the message passing paradigm, graph neural networks (GNNs) have greatly advanced the performance of node representation learning on graphs. However, a majority class of GNNs are only designed for homogeneous graphs, leading to inferior adaptivity to the more informative heterogeneous graphs with various types of nodes and edges. Also, despite the necessity of inductively producing representations for completely new nodes (e.g., in streaming scenarios), few heterogeneous GNNs can bypass the transductive learning scheme where all nodes must be known during training. Furthermore, the training efficiency of most heterogeneous GNNs has been hindered by their sophisticated designs for extracting the semantics associated with each meta path or relation. In this paper, we propose wide and deep message passing network (WIDEN) to cope with the aforementioned problems about heterogeneity, inductiveness, and efficiency that are rarely investigated together in graph representation learning. In WIDEN, we propose a novel inductive, meta path-free message passing scheme that packs up heterogeneous node features with their associated edges from both low- and high-order neighbor nodes. To further improve the training efficiency, we innovatively present an active downsampling strategy that drops unimportant neighbor nodes to facilitate faster information propagation. Experiments on three real-world heterogeneous graphs have further validated the efficacy of WIDEN on both transductive and inductive node representation learning, as well as the superior training efficiency against state-of-the-art baselines. Copyright © 2021, The Authors. All rights reserved.

关键词： Message passing

来源：评论

学校读者我要写书评

暂无评论

Adversarial Driving Behavior Generation Incorporating Human Risk Cognition for Autonomous Vehicle Evaluation

Adversarial Driving Behavior Generation Incorporating Human ...

引用

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

作者： Zhen Liu Hang Gao Hao Ma Shuo Cai Yunfeng Hu Ting Qu Hong Chen Xun Gong School of Artificial Intelligence Jilin University China and Engineering Research Center of Knowledge-Driven Human-Machine Intelligence MOE China College of Communication Engineering Jilin University Changchun China State Key Laboratory of Automotive Simulation and Control Jilin University Changchun China School of Electronic and Information Engineering Tongji University Shanghai China

Autonomous vehicle (AV) evaluation has been the subject of increased interest in recent years both in industry and in academia. This paper focuses on the development of a novel framework for generating adversarial driving behavior of background vehicle interfering against the AV to expose effective and rational risky events. Specifically, the adversarial behavior is learned by a reinforcement learning (RL) approach incorporated with the cumulative prospect theory (CPT) which allows representation of human risk cognition. Then, the extended version of deep deterministic policy gradient (DDPG) technique is proposed for training the adversarial policy while ensuring training stability as the CPT action-value function is leveraged. A comparative case study regarding the cut-in scenario is conducted on a high fidelity Hardware-in-the-Loop (HiL) platform and the results demonstrate the adversarial effectiveness to infer the weakness of the tested AV.

关键词：

来源：评论

学校读者我要写书评

暂无评论

On the generalization of stochastic gradient descent with momentum

The Journal of Machine Learning Research

引用

The Journal of machine Learning Research 2024年第1期25卷 1132-1187页

作者： Ali Ramezani-Kebrya Kimon Antonakopoulos Volkan Cevher Ashish Khisti Ben Liang Department of Informatics University of Oslo and Visual Intelligence Centre Integreat Norwegian Centre for Knowledge-driven Machine Learning Oslo Norway Laboratory for Information and Inference Systems (LIONS) EPFL EPFL STI IEL LIONS Lausanne Switzerland Department of Electrical and Computer Engineering University of Toronto Toronto ON Canada

While momentum-based accelerated variants of stochastic gradient descent (SGD) are widely used when training machine learning models, there is little theoretical understanding on the generalization error of such methods. In this work, we first show that there exists a convex loss function for which the stability gap for multiple epochs of SGD with standard heavy-ball momentum (SGDM) becomes unbounded. Then, for smooth Lipschitz loss functions, we analyze a profiles momentum-based update rule, i.e., SGD with early momentum (SGDEM) under a broad range of step-sizes, and show that it can train machine learning models for multiple epochs with a guarantee for generalization. Finally, for the special case of strongly convex loss functions, we find a range of momentum such that multiple epochs of standard SGDM, as a special form of SGDEM, also generalizes. Extending our results on generalization, we also develop an upper bound on the expected true risk, in terms of the number of training steps, sample size, and momentum. Our experimental evaluations verify the consistency between the numerical results and our theoretical bounds. SGDEM improves the generalization error of SGDM when training ResNet-18 on ImageNet in practical distributed settings.

关键词： uniform stability generalization error heavy-ball momentum stochastic gradient descent non-convex

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：