检索结果-内蒙古大学图书馆

21st International Computer Conference on Wavelet Active Media Technology and Information Processing, ICCWAMTIP 2024

作者： Sahlu, Ameneshewa Abush Yohannes, Gobena Mulugeta Dawit, Bekalu Nigus Teshome, Molla Woretaw School of Computer Science and Engineering University of Electronic Science and Technology of China Sichuan Chengdu China School of System Design and Intelligent Manufacturing Southern University of Science and Technology Shenzhen China School of Information and Software Engineering University of Electronic Science and Technology of China Sichuan Chengdu China

ISBN: (纸本)9798331519254

Mental stress poses significant health risks, manifesting in various psychological and physical issues such as depression, anxiety, and cardiovascular complications. Establishing a reliable method for swiftly and accurately classifying mental stress levels is crucial due to its widespread impact. Existing classification approaches predominantly rely on either data-intensive traditional machine learning algorithms or opaque deep learning models, hindering real-world use and interpretability. This study addresses these limitations by proposing a novel approach for stress level classification using the WESAD dataset. Our method combines an attention-based ResN et with CBAM. This innovative approach synergistically integrates shallow and deep features, harnessing both low-level details and high-level semantic information to effectively classify stress levels into four distinct states: baseline, stress, amusement, and meditation, or three states excluding the meditation state. The experimental result shows 92.74% accuracy and 91.19% F1 score for the 4-class setup, and 97.58% accuracy and 97.51 % F1 score for the 3-class setup. Our approach offers a more robust and interpretable framework for mental stress classification, with potential applications in health care and stress management. © 2024 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Video2Reward: Generating Reward Function from Videos for Legged Robot Behavior Learning 27

Video2Reward: Generating Reward Function from Videos for Leg...

引用

27th European Conference on Artificial Intelligence, ECAI 2024

作者： Zeng, Runhao Zhou, Dingjie Liang, Qiwei Liu, Junlin Li, Hui Huang, Changxin Li, Jianqiang Hu, Xiping Sun, Fuchun Artificial Intelligence Research Institute Shenzhen MSU-BIT University China College of Mechatronics and Control Engineering Shenzhen University China College of Computer Science and Software Engineering Shenzhen University China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China Department of Computer Science and Technology Tsinghua University China

ISBN: (纸本)9781643685489

Learning behavior in legged robots presents a significant challenge due to its inherent instability and complex constraints. Recent research has proposed the use of a large language model (LLM) to generate reward functions in reinforcement learning, thereby replacing the need for manually designed rewards by experts. However, this approach, which relies on textual descriptions to define learning objectives, fails to achieve controllable and precise behavior learning with clear directionality. In this paper, we introduce a new video2reward method, which directly generates reward functions from videos depicting the behaviors to be mimicked and learned. Specifically, we first process videos containing the target behaviors, converting the motion information of individuals in the videos into keypoint trajectories represented as coordinates through a video2text transforming module. These trajectories are then fed into an LLM to generate the reward function, which in turn is used to train the policy. To enhance the quality of the reward function, we develop a video-assisted iterative reward refinement scheme that visually assesses the learned behaviors and provides textual feedback to the LLM. This feedback guides the LLM to continually refine the reward function, ultimately facilitating more efficient behavior learning. Experimental results on tasks involving bipedal and quadrupedal robot motion control demonstrate that our method surpasses the performance of state-of-the-art LLM-based reward generation methods by over 37.6% in terms of human normalized score. More importantly, by switching video inputs, we find our method can rapidly learn diverse motion behaviors such as walking and running. © 2024 The Authors.

关键词： Robots

来源：评论

学校读者我要写书评

暂无评论

Towards enabling learnware to handle heterogeneous feature spaces

引用

MACHINE LEARNING 2024年第4期113卷 1839-1860页

作者： Tan, Peng Tan, Zhi-Hao Jiang, Yuan Zhou, Zhi-Hua Nanjing Univ Natl Key Lab Novel Software Technol Nanjing 210023 Peoples R China

The learnware paradigm was recently proposed by Zhou (2016) with the wish of developing the learnware market to help users build models more efficiently by reusing existing well-performed models rather than starting from scratch. Specifically, a learnware in the learnware market is a well-performed pre-trained model with a specification describing its specialty and utility, and the market identifies helpful learnware(s) for the user's task based on the specification. Recent studies have attempted to realize a homogeneous prototype learnware market initially through Reduced Kernel Mean Embedding (RKME) specification, which requires all models in the market to share the same feature space. However, this limits the application scope of the learnware paradigm because various pre-trained models are often obtained from different feature spaces in real-world scenarios. In this paper, we make the first attempt to enable the learnware to handle heterogeneous feature spaces. We propose a more powerful specification to manage heterogeneous learnwares by integrating subspace learning in the specification design, along with a practical approach for identifying and reusing helpful learnwares for the user's task. Empirical studies on both synthetic data and real-world tasks validate the efficacy of our approach.

关键词： Learnware Heterogeneous feature spaces Model reuse Subspace learning

来源：评论

学校读者我要写书评

暂无评论

Pre-Trained Model-Based Automated software Vulnerability Repair: How Far are We?

引用

IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING 2024年第4期21卷 2507-2525页

作者： Zhang, Quanjun Fang, Chunrong Yu, Bowen Sun, Weisong Zhang, Tongke Chen, Zhenyu Nanjing Univ State Key Lab Novel Software Technol Nanjing 210093 Peoples R China

Various approaches are proposed to help under-resourced security researchers to detect and analyze software vulnerabilities. It is still incredibly time-consuming and labor-intensive for security researchers to fix such reported vulnerabilities due to the increasing size and complexity of modern software systems. The time lag between the reporting and fixing of a security vulnerability causes software systems to suffer from significant exposure to possible attacks. Very recently, some techniques propose to apply pre-trained models to fix security vulnerabilities and have proved their success in improving repair accuracy. However, the effectiveness of existing pre-trained models has not been systematically compared and little is known about their advantages and disadvantages. To bridge this gap, we perform the first extensive study on applying various pre-trained models to automated vulnerability repair. The experimental results on two vulnerability datasets show that all studied pre-trained models consistently outperform the state-of-the-art technique VRepair with a prediction accuracy of 32.94%$\sim$similar to 44.96%. We also investigate the impact of three major phases (i.e., data pre-processing, model training and repair inference) in the vulnerability repair workflow. Inspired by the findings, we construct a simplistic vulnerability repair approach that adopts the transfer learning from bug fixing. Surprisingly, such a simplistic approach can further improve the prediction accuracy of pre-trained models by 9.40% on average. Besides, we provide additional discussion from different aspects (e.g., code representation and a preliminary study with ChatGPT) to illustrate the capacity and limitation of pre-trained model-based techniques. Finally, we further pinpoint various practical guidelines (e.g., the improvement of fine-tuning) for advanced pre-trained model-based vulnerability repair in the near future. Our study highlights the promising future of adopting pre-tr

关键词： Maintenance engineering Security Codes Predictive models Training Task analysis Transfer learning Security vulnerability pre-trained model vulnerability repair

来源：评论

学校读者我要写书评

暂无评论

Learning Joint 2-D and 3-D Graph Diffusion Models for Complete Molecule Generation

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2024年第9期35卷 11857-11871页

作者： Huang, Han Sun, Leilei Du, Bowen Lv, Weifeng Beihang Univ State Key Lab Software Dev Environm Beijing 100191 Peoples R China

Designing new molecules is essential for drug discovery and material science. Recently, deep generative models that aim to model molecule distribution have made promising progress in narrowing down the chemical research space and generating high-fidelity molecules. However, current generative models only focus on modeling 2-D bonding graphs or 3-D geometries, which are two complementary descriptors for molecules. The lack of ability to jointly model them limits the improvement of generation quality and further downstream applications. In this article, we propose a joint 2-D and 3-D graph diffusion model (JODO) that generates geometric graphs representing complete molecules with atom types, formal charges, bond information, and 3-D coordinates. To capture the correlation between 2-D molecular graphs and 3-D geometries in the diffusion process, we develop a diffusion graph transformer (DGT) to parameterize the data prediction model that recovers the original data from noisy data. The DGT uses a relational attention mechanism that enhances the interaction between node and edge representations. This mechanism operates concurrently with the propagation and update of scalar attributes and geometric vectors. Our model can also be extended for inverse molecular design targeting single or multiple quantum properties. In our comprehensive evaluation pipeline for unconditional joint generation, the experimental results show that JODO remarkably outperforms the baselines on the QM9 and GEOM-Drugs datasets. Furthermore, our model excels in few-step fast sampling, as well as in inverse molecule design and molecular graph generation. Our code is provided in https://***/GRAPH-0/JODO.

关键词： Solid modeling Diffusion models Geometry Data models Predictive models Bonding Transformers Deep generative model geometric graph learning graph transformer molecule design

来源：评论

学校读者我要写书评

暂无评论

WINGS: Learning Multimodal LLMs without Text-only Forgetting 38

WINGS: Learning Multimodal LLMs without Text-only Forgetting

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Zhang, Yi-Kai Lu, Shiyin Li, Yang Ma, Yanqing Chen, Qing-Guo Xu, Zhao Luo, Weihua Zhang, Kaifu Zhan, De-Chuan Ye, Han-Jia School of Artificial Intelligence Nanjing University China National Key Laboratory for Novel Software Technology Nanjing University China Alibaba International Digital Commerce China

Multimodal large language models (MLLMs), initiated with a trained LLM, first align images with text and then fine-tune on multimodal mixed inputs. However, during the continued training, the MLLM catastrophically forgets the text-only instructions that the initial LLM masters. In this paper, we present WINGS, a novel MLLM that excels in both text-only and multimodal instructions. By examining attention across layers of MLLM, we find that text-only forgetting is related to the attention shifts from pre-image to post-image text. From that, we construct an additional Low-Rank Residual Attention (LoRRA) block that acts as the "modality learner" to expand the learnable space and compensate for the attention shift. The complementary learners, like "wings" on either side, are connected in parallel to each layer's attention block. The LoRRA mirrors the structure of attention but utilizes low-rank connections to ensure efficiency. Initially, image and text inputs are aligned with visual learners operating alongside the main attention, balancing focus on visual elements. Later, textual learners are integrated with token-wise routing, blending the outputs of both modality learners collaboratively. Our experimental results demonstrate that WINGS outperforms equally-scaled MLLMs in both text-only and visual question-answering tasks. WINGS with compensation of learners addresses text-only forgetting during visual modality expansion in general MLLMs. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

HierarchyNet: Learning to Summarize Source Code with Heterogeneous Representations 18

HierarchyNet: Learning to Summarize Source Code with Heterog...

引用

18th Conference of the European Chapter of the Association for computational Linguistics, EACL 2024 - Findings of EACL 2024

作者： Nguyen, Minh Huynh Bui, Nghi D.Q. Hy, Truong Son Thanh, Long Tran Nguyen, Tien N. FPT Software AI Center Department of Computer Science Fulbright University Viet Nam Department of Mathematics and Computer Science Indiana State University United States Department of Computer Science University of Warwick United Kingdom Computer Science Department The University of Texas Dallas United States

ISBN: (纸本)9798891760936

Code representation is important to machine learning models in the code-related applications. Existing code summarization approaches primarily leverage Abstract Syntax Trees (ASTs) and sequential information from source code to generate code summaries while often overlooking the critical consideration of the interplay of dependencies among code elements and code hierarchy. However, effective summarization necessitates a holistic analysis of code snippets from three distinct aspects: lexical, syntactic, and semantic information. In this paper, we propose a novel code summarization approach utilizing Heterogeneous Code Representations (HCRs) and our specially designed HIERARCHYNET. HCRs adeptly capture essential code features at lexical, syntactic, and semantic levels within a hierarchical structure. HIERARCHYNET processes each layer of the HCR separately, employing a Heterogeneous Graph Transformer, a Tree-based CNN, and a Transformer Encoder. In addition, HIERARCHYNET demonstrates superior performance compared to fine-tuned pretrained models, including CodeT5, and CodeBERT, as well as large language models that employ zero/few-shot settings, such as CodeLlama, StarCoder, and CodeGen. Implementation details can be found at https://***/FSoft-AI4Code/HierarchyNet. © 2024 Association for computational Linguistics.

关键词： Syntactics

来源：评论

学校读者我要写书评

暂无评论

CM-OOA：An Energy-Efficient Clustering Algorithm for Wireless Sensor Networks Using Chaotic Mapping and Osprey Optimization

引用

Informatica (Slovenia) 2025年第12期49卷 173-190页

作者： Jia, Songhao Shao, Wenqian Yang, Cai Jia, Shuya Yuan, Yaohui Chen, Huiyuan Zhang, Haiyu School of Artificial Intelligence and Software Engineering Nanyang Normal University Henan Nanyang473061 China

A wireless sensor network (WSN) represents a promising approach for establishing self-organizing wireless networks comprising a substantial number of wireless sensors, with the objective of facilitating communication in regions where the existing communication infrastructure has been severely disrupted. In order to address the issue of excessive energy consumption by cluster heads and central nodes in emergency communication networks of wireless sensor networks, this paper proposes an emergency communication algorithm for wireless sensor networks based on chaos mapping and osprey optimization. Firstly, an optimization algorithm based on chaos theory is used to select the virtual position of the initial population of the Osprey optimization algorithm. This is achieved by simulating the randomness and unpredictability of chaotic systems. Secondly, the osprey optimization algorithm and the improved fitness function are used to select the optimal cluster head combination. In the selection process, six factors, such as the energy level of network nodes, the distance between cluster heads, the distance between cluster heads and base stations, the distance between cluster heads and ordinary nodes, the variance of the distance between cluster heads and base stations and the variance of the distance between cluster heads, are comprehensively considered. Finally, the heuristic function of FA-star algorithm is used to select the next hop node to transmit the message. The results of the simulation demonstrate that the residual energy of the CM-OOA algorithm is 14% higher than that of the CGWOA algorithm following the transmission of 1000 data rounds. This figure is 54% higher than that observed for the PSO-C algorithm. The findings demonstrate that the CM-OOA algorithm effectively extends the network lifetime and preserves a favorable load balance in diverse network settings. © 2025 Slovene Society Informatika. All rights reserved.

关键词： Base stations

来源：评论

学校读者我要写书评

暂无评论

Deformation mechanism of Cu-Al-Ni shape memory alloys fabricated via laser powder b e d fusion:Tension-compression asymmetry

引用

Journal of Materials science & Technology 2023年第36期167卷 14-26页

作者： Yankun Zhang Lianyong Xu Lei Zhao Danyang Lin Minqian Liu Wei Chen Yongdian Han School of Materials Science and Engineering Tianjin UniversityTianjin 300350China Tianjin Key Laboratory of Advanced Joining Technology Tianjin 300350China State Key Laboratory of Engines Tianjin UniversityTianjin 300350China State Key Laboratory of Advanced Welding and Joining Harbin Institute of TechnologyHarbin 150001China

Introducing the unique advantage of additive manufacturing technology into copper-based shape memory alloys(SMAs)to fabricate high-performance alloys has garnered great attention in recent years,but the intrinsic relationships between microstructure and mechanical properties need to be further *** this paper,the microstructural evolution of ternary CuAlNi SMAs fabricated by laser powder bed fusion(LPBF)under the tensile-compressive loading was investigated to determine the underlying mechanism of tension-compression asymmetry,that is,excellent compressive but poor tensile *** characterization of the different deformation stages revealed the numerous activated deformation mech-anism on the 18R martensite matrix.A twin-related transformation dominated the main plastic defor-mation process due to lower stacking faults energy and high-density pre-existing planer defects in the CuAlNi *** twinning nucleated at prior austenite boundaries and developed into parallel and network structures inside the parent grain of different *** addition,the preferred orientation in different stages,the stress-inducedγphase transformation,and the interaction between dislocations and stacking faults are *** results not only provide significant insights to understand the detwinning and deformation twinning process of SMAs but also establish the essential framework of mi-crostructure and mechanical properties of Cu-based SMAs fabricated by LPBF.

关键词： Shape memory alloys Cu-based Laser powder bed fusion Microstructural evolution Deformation mechanism

来源：评论

学校读者我要写书评

暂无评论

Parkinson’s Disease Detection from Resting State EEG Using Multi-head Graph Structure Learning with Gradient Weighted Graph Attention Explanations 7th

Parkinson’s Disease Detection from Resting State EEG Using...

引用

7th International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2024, Held in Conjunction with 27th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2024

作者： Neves, Christopher Zeng, Yong Xiao, Yiming Department of Computer Science and Software Engineering Concordia University MontrealQC Canada Concordia Institute for Information Systems Engineering Concordia University MontrealQC Canada

ISBN: (纸本)9783031787607

Parkinson’s disease (PD) is a debilitating neurodegenerative disease that has severe impacts on an individual’s quality of life. Compared with structural and functional MRI-based biomarkers for the disease, electroencephalography (EEG) can provide more accessible alternatives for clinical insights. While deep learning (DL) techniques have provided excellent outcomes, many techniques fail to model spatial information and dynamic brain connectivity, and face challenges in robust feature learning, limited data sizes, and poor explainability. To address these issues, we proposed a novel graph neural network (GNN) technique for explainable PD detection using resting state EEG. Specifically, we employ structured global convolutions with contrastive learning to better model complex features with limited data, a novel multi-head graph structure learner to capture the non-Euclidean structure of EEG data, and a head-wise gradient-weighted graph attention explainer to offer neural connectivity insights. We developed and evaluated our method using the UC San Diego Parkinson’s disease EEG dataset, and achieved 69.40% detection accuracy in subject-wise leave-one-out cross-validation while generating intuitive explanations for the learnt graph topology. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Contrastive Learning

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：