检索结果-内蒙古大学图书馆

Personalized Federated Learning with Collaborative Aggregation networks for Multi-Site Brain Disorder Diagnosis 4

学校读者我要写书评

暂无评论

Personalized Federated Learning with Collaborative Aggregati...

4th International Conference on Industrial Automation, Robotics and Control Engineering, IARCE 2024

作者： Si, Qian Li, Yang School of Cyber Science and Technology Beihang University Beijing China Department of Automation Science and Electrical Engineering The Beijing Advanced Innovation Center for Big Data and Brain Computing State Key Laboratory of Virtual Reality Technology and Systems Beijing China Advanced Institute of Information Technology Peking University Beijing China Beihang University Beijing China

ISBN: (纸本)9798350380323

In multi-site brain disease diagnosis studies, traditional centralized training methods necessitate sharing medical data, posing significant privacy risks. Federated learning (FL) offers a privacy-preserving solution by enabling global model training through aggregating locally trained models from multiple data centers without sharing raw data. However, current FL approaches rely on a server-based network topology, where central server failure disrupts training. Additionally, data heterogeneity across sites often slows convergence and reduces accuracy. To overcome these issues, we introduce a decentralized personalized federated learning collaborative aggregation network (pFedCAN). This framework has two core components: (1) separating local models into shared and personalized layers, and (2) forming a collaborative aggregation network via similarity detection in the shared layers. Specifically, each center trains its local model, then separates it into shared and personalized layers. The shared layer is exchanged with other centers, while the personalized layer remains local. data centers analyze similarities in received shared layers to build a collaborative network, where shared layers from similar centers are aggregated to refine the model. This approach flexibly adapts to varying levels of data heterogeneity, enhancing model training efficiency. Validation on public datasets, ABIDE I and ADHD, shows that the proposed method outperforms current leading techniques. © 2024 IEEE.

关键词： Federated learning

Retrieval-Augmented Code Generation for Universal Information Extraction

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Guo, Yucan Li, Zixuan Jin, Xiaolong Liu, Yantao Zeng, Yutao Liu, Wenxuan Li, Xiang Yang, Pan Bai, Long Guo, Jiafeng Cheng, Xueqi CAS Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences China School of Computer Science and Technology University of Chinese Academy of Sciences China Platform and Content Group Tencent

Information Extraction (IE) aims to extract structural knowledge (e.g., entities, relations, events) from natural language texts, which brings challenges to existing methods due to task-specific schemas and complex text expressions. Code, as a typical kind of formalized language, is capable of describing structural knowledge under various schemas in a universal way. On the other hand, Large Language Models (LLMs) trained on both codes and texts have demonstrated powerful capabilities of transforming texts into codes, which provides a feasible solution to IE tasks. Therefore, in this paper, we propose a universal retrieval-augmented code generation framework based on LLMs, called Code4UIE, for IE tasks. Specifically, Code4UIE adopts Python classes to define task-specific schemas of various structural knowledge in a universal way. By so doing, extracting knowledge under these schemas can be transformed into generating codes that instantiate the predefined Python classes with the information in texts. To generate these codes more precisely, Code4UIE adopts the in-context learning mechanism to instruct LLMs with examples. In order to obtain appropriate examples for different tasks, Code4UIE explores several example retrieval strategies, which can retrieve examples semantically similar to the given texts. Extensive experiments on five representative IE tasks across nine datasets demonstrate the effectiveness of the Code4UIE framework. Copyright © 2023, The Authors. All rights reserved.

关键词： Information retrieval

Gradient-Based Differential kWTA network With Application to Competitive Coordination of Multiple Robots

学校读者我要写书评

暂无评论

IEEE/CAA Journal of Automatica Sinica 2022年第8期9卷 1452-1463页

作者： Mei Liu Xiaoyan Zhang Mingsheng Shang Long Jin School of Information Science and Engineering Lanzhou UniversityLanzhou 730000 State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining 810016China Chongqing Key Laboratory of Big Data and Intelligent Computing Chongqing Institute of Green and Intelligent TechnologyChinese Academy of SciencesChongqing 400714China Shenzhen Research Institute Lanzhou UniversityShenzhen 518063China

Aiming at the k-winners-take-all(kWTA)operation,this paper proposes a gradient-based differential kWTA(GDk WTA)*** obtaining the network,theorems and related proofs are provided to guarantee the exponential convergence and noise resistance of the proposed GD-kWTA ***,numerical simulations are conducted to substantiate the preferable performance of the proposed network as compared with the traditional ***,the GD-k WTA network,backed with a consensus filter,is utilized as a robust control scheme for modeling the competition behavior in the multi-robot coordination,thereby further demonstrating its effectiveness and feasibility.

关键词： Consensus filter k-winners-take-all(kWTA) multirobot coordination noise resistance

Multi-video Moment Ranking with Multimodal Clue

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Hou, Danyang Pang, Liang Lan, Yanyan Shen, Huawei Cheng, Xueqi Data Intelligence System Research Center Institute of Computing Technology CAS Beijing China CAS Key Lab of Network Data Science and Technology Institute of Computing Technology CAS Beijing China University of Chinese Academy of Sciences Beijing China Institute for AI Industry Research Tsinghua University Beijing China

Video corpus moment retrieval (VCMR) is the task of retrieving a relevant video moment from a large corpus of untrimmed videos via a natural language query. State-of-the-art work for VCMR is based on two-stage method. In this paper, we focus on improving two problems of two-stage method: (1) Moment prediction bias: The predicted moments for most queries come from the top retrieved videos, ignoring the possibility that the target moment is in the bottom retrieved videos, which is caused by the inconsistency of Shared Normalization during training and inference. (2) Latent key content: Different modalities of video have different key information for moment localization. To this end, we propose a two-stage model MultI-video raNking with mUlTimodal cluE (MINUTE). MINUTE uses Shared Normalization during both training and inference to rank candidate moments from multiple videos to solve moment predict bias, making it more efficient to predict target moment. In addition, Mutilmdaol Clue Mining (MCM) of MINUTE can discover key content of different modalities in video to localize moment more accurately. MINUTE outperforms the baselines on TVR and DiDeMo datasets, achieving a new state-of-the-art of VCMR. Our code will be available at GitHub. Copyright © 2023, The Authors. All rights reserved.

关键词： Forecasting

TrustRAG: An Information Assistant with Retrieval Augmented Generation

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Fan, Yixing Yan, Qiang Wang, Wenshan Guo, Jiafeng Zhang, Ruqing Cheng, Xueqi Key Lab of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

Retrieval-Augmented Generation (RAG) has emerged as a crucial technique for enhancing large models with real-time and domain-specific knowledge. While numerous improvements and open-source tools have been proposed to refine the RAG framework for accuracy, relatively little attention has been given to improving the trustworthiness of generated results. To address this gap, we introduce TrustRAG, a novel framework that enhances RAG from three perspectives: indexing, retrieval, and generation. Specifically, in the indexing stage, we propose a semantic-enhanced chunking strategy that incorporates hierarchical indexing to supplement each chunk with contextual information, ensuring semantic completeness. In the retrieval stage, we introduce a utility-based filtering mechanism to identify high-quality information, supporting answer generation while reducing input length. In the generation stage, we propose fine-grained citation enhancement, which detects opinion-bearing sentences in responses and infers citation relationships at the sentence-level, thereby improving citation accuracy. We open-source the TrustRAG framework and provide a demonstration studio designed for excerpt-based question answering tasks 1. Based on these, we aim to help researchers: 1) systematically enhancing the trustworthiness of RAG systems and (2) developing their own RAG systems with more reliable outputs. © 2025, CC BY.

关键词： Semantics

Visual Transformation Telling

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Cui, Wanqing Hong, Xin Lan, Yanyan Pang, Liang Guo, Jiafeng Cheng, Xueqi CAS Key Laboratory of Network Data Science and Technology Institute of Computing Technology Chinese Academy of Sciences Beijing China Data Intelligence System Research Center Institute of Computing Technology CAS Beijing China University of Chinese Academy of Sciences Beijing China Institute for AI Industry Research Tsinghua University Beijing China

Humans can naturally reason from superficial state differences (e.g. ground wetness) to transformations descriptions (e.g. raining) according to their life experience. In this paper, we propose a new visual reasoning task to test this transformation reasoning ability in real-world scenarios, called Visual Transformation Telling (VTT). Given a series of states (i.e. images), VTT requires to describe the transformation occurring between every two adjacent states. Different from existing visual reasoning tasks that focus on surface state reasoning, the advantage of VTT is that it captures the underlying causes, e.g. actions or events, behind the differences among states. We collect a novel dataset to support the study of transformation reasoning from two existing instructional video datasets, CrossTask and COIN, comprising 13,547 samples. Each sample involves the key state images along with their transformation descriptions. Our dataset covers diverse real-world activities, providing a rich resource for training and evaluation. To construct an initial benchmark for VTT, we test several models, including traditional visual storytelling methods (CST, GLACNet, Densecap) and advanced multimodal large language models (LLaVA v1.5-7B, Qwen-VL-chat, Gemini Pro Vision, GPT-4o, and GPT-4). Experimental results reveal that even state-of-the-art models still face challenges in VTT, highlighting substantial areas for improvement. © 2023, CC BY-NC-SA.

关键词： Visual languages

AUCSeg: AUC-oriented pixel-level long-tail semantic segmentation 24

学校读者我要写书评

暂无评论

AUCSeg: AUC-oriented pixel-level long-tail semantic segmenta...

Proceedings of the 38th International Conference on Neural Information Processing Systems

作者： Boyu Han Qianqian Xu Zhiyong Yang Shilong Bao Peisong Wen Yangbangyan Jiang Qingming Huang Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS and School of Computer Science and Tech. University of Chinese Academy of Sciences Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS and Peng Cheng Laboratory School of Computer Science and Tech. University of Chinese Academy of Sciences School of Computer Science and Tech. University of Chinese Academy of Sciences and Key Lab. of Intelligent Information Processing Institute of Computing Technology CAS and Key Laboratory of Big Data Mining and Knowledge Management CAS

ISBN: (纸本)9798331314385

The Area Under the ROC Curve (AUC) is a well-known metric for evaluating instance-level long-tail learning problems. In the past two decades, many AUC optimization methods have been proposed to improve model performance under long-tail distributions. In this paper, we explore AUC optimization methods in the context of pixel-level long-tail semantic segmentation, a much more complicated scenario. This task introduces two major challenges for AUC optimization techniques. On one hand, AUC optimization in a pixel-level task involves complex coupling across loss terms, with structured inner-image and pairwise inter-image dependencies, complicating theoretical analysis. On the other hand, we find that mini-batch estimation of AUC loss in this case requires a larger batch size, resulting in an unaffordable space complexity. To address these issues, we develop a pixel-level AUC loss function and conduct a dependency-graph-based theoretical analysis of the algorithm's generalization ability. Additionally, we design a Tail-Classes Memory Bank (T-Memory Bank) to manage the significant memory demand. Finally, comprehensive experiments across various benchmarks confirm the effectiveness of our proposed AUCSeg method. The code is available at https://***/boyuh/AUCSeg.

关键词：

Combat data shift in few-shot learning with knowledge graph

学校读者我要写书评

暂无评论

Frontiers of Computer science 2023年第1期17卷 101-111页

作者： Yongchun ZHU Fuzhen ZHUANG Xiangliang ZHANG Zhiyuan QI Zhiping SHI Juan CAO Qing HE Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(CAS) Institute of Computing TechnologyCASBeijing 100190China University of Chinese Academy of Sciences Beijing 100049China Institute of Artificial Intelligence Beihang UniversityBeijing 100191China Xiamen Institute of Data Intelligence Xiamen 361021China Computer Science and Engineering University of Notre DameIN 46556USA University of California San Diego La JollaCA 92093USA College of Information Engineering&Academy for Multidisciplinary Studies Capital Normal UniversityBeijing 100089China

Many few-shot learning approaches have been designed under the meta-learning framework, which learns from a variety of learning tasks and generalizes to new tasks. These meta-learning approaches achieve the expected performance in the scenario where all samples are drawn from the same distributions (i.i.d. observations). However, in real-world applications, few-shot learning paradigm often suffers from data shift, i.e., samples in different tasks, even in the same task, could be drawn from various data distributions. Most existing few-shot learning approaches are not designed with the consideration of data shift, and thus show downgraded performance when data distribution shifts. However, it is non-trivial to address the data shift problem in few-shot learning, due to the limited number of labeled samples in each task. Targeting at addressing this problem, we propose a novel metric-based meta-learning framework to extract task-specific representations and task-shared representations with the help of knowledge graph. The data shift within/between tasks can thus be combated by the combination of task-shared and task-specific representations. The proposed model is evaluated on popular benchmarks and two constructed new challenging datasets. The evaluation results demonstrate its remarkable performance.

关键词： few-shot data shift knowledge graph

Tensor Graph Convolutional network for Dynamic Graph Representation Learning 7

学校读者我要写书评

暂无评论

Tensor Graph Convolutional Network for Dynamic Graph Represe...

7th International Symposium on Autonomous Systems, ISAS 2024

作者： Wang, Ling Yuan, Ye The School of Computer Science and Technology Chongqing University of Posts and Telecommunications Chongqing400065 China The Chongqing Key Laboratory of Big Data and Intelligent Computing Chongqing Engineering Research Center of Big Data Application For Smart Cities Chongqing Institute of Green and Intelligent Technology Chinese Academy of Sciences Chongqing400714 China College of Computer and Information Science Southwest University Chongqing400715 China

ISBN: (纸本)9798350363173

Dynamic graphs (DG) represent evolving interactions between entities in various real-world scenarios. Many existing DG representation learning models employ a combination of graph convolutional networks and sequence neural networks to capture spatial-temporal dependencies. However, this hybrid approach often fails to effectively capture the spatial-temporal continuity inherent in DGs. In this paper, we propose a novel approach called the Tensor Graph Convolutional network (TGCN) to learn DG representations within a unified convolution framework based on tensor products. Our approach is founded on two key principles: (a) representing DG information in tensor form, and (b) leveraging tensor product operations to design a graph convolutional network that simultaneously models spatial and temporal features. Experimental results on three real-world DG datasets demonstrate that our proposed model achieves state-of-the-art performance. © 2024 IEEE.

关键词： Tensors