检索结果-内蒙古大学图书馆

Policy learning from tutorial books via understanding, rehearsing and introspecting 24

学校读者我要写书评

暂无评论

Policy learning from tutorial books via understanding, rehea...

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

作者： Xiong-Hui Chen Ziyan Wang Yali Du Shengyi Jiang Meng Fang Yang Yu Jun Wang National Key Laboratory for Novel Software Technology Nanjing University China & School of Artificial Intelligence Nanjing University China Cooperative AI Lab Department of Informatics King's College London The University of Hong Kong University of Liverpool AI Centre Department of Computer Science University College London

ISBN: (纸本)9798331314385

When humans need to learn a new skill, we can acquire knowledge through written books, including textbooks, tutorials, etc. However, current research for decisionmaking, like reinforcement learning (RL), has primarily required numerous real interactions with the target environment to learn a skill, while failing to utilize the existing knowledge already summarized in the text. The success of Large Language Models (LLMs) sheds light on utilizing such knowledge behind the books. In this paper, we discuss a new policy learning problem called Policy Learning from tutorial Books (PLfB) upon the shoulders of LLMs' systems, which aims to leverage rich resources such as tutorial books to derive a policy network. Inspired by how humans learn from books, we solve the problem via a three-stage framework: Understanding, Rehearsing, and Introspecting (URI). In particular, it first rehearses decision-making trajectories based on the derived knowledge after understanding the books, then introspects about the imaginary dataset to distill a policy network. We build two benchmarks for PLfB based on Tic-Tac-Toe and Football games. In the experiment, URI's policy achieves a minimum of 44% net winning rate against GPT-based agents without any real data. In the much more complex football game, URI's policy beat the built-in AIs with a 37% winning rate while GPT-based agents can only achieve a 6% winning rate. The project page: https://***/.

关键词：

Not All Semantics are Created Equal: Contrastive Self-supervised Learning with Automatic Temperature Individualization

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Qiu, Zi-Hao Hu, Quanqi Yuan, Zhuoning Zhou, Denny Zhang, Lijun Yang, Tianbao National Key Laboratory for Novel Software Technology Nanjing University Nanjing China Computer Science and Engineering Texas A&M University College Station United States Department of Computer Science The University of Iowa Iowa City United States Google Research United States

In this paper, we aim to optimize a contrastive loss with individualized temperatures in a principled and systematic manner for self-supervised learning. The common practice of using a global temperature parameter τ ignores the fact that "not all semantics are created equal", meaning that different anchor data may have different numbers of samples with similar semantics, especially when data exhibits long-tails. First, we propose a new robust contrastive loss inspired by distributionally robust optimization (DRO), providing us an intuition about the effect of τ and a mechanism for automatic temperature individualization. Then, we propose an efficient stochastic algorithm for optimizing the robust contrastive loss with a provable convergence guarantee without using large mini-batch sizes. Theoretical and experimental results show that our algorithm automatically learns a suitable τ for each sample. Specifically, samples with frequent semantics use large temperatures to keep local semantic structures, while samples with rare semantics use small temperatures to induce more separable features. Our method not only outperforms prior strong baselines (e.g., SimCLR, CLIP) on unimodal and bimodal datasets with larger improvements on imbalanced data but also is less sensitive to hyper-parameters. To our best knowledge, this is the first methodical approach to optimizing a contrastive loss with individualized temperatures. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

Dynamic network embedding via incremental skip-gram with negative sampling

学校读者我要写书评

暂无评论

science China(Information sciences) 2020年第10期63卷 89-107页

作者： Hao PENG Jianxin LI Hao YAN Qiran GONG Senzhang WANG Lin LIU Lihong WANG Xiang REN Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University State Key Laboratory of Software Development Environment Beihang University Collage of Computer Science and Technology Nanjing University of Aeronautics and Astronautics National Computer Network Emergency Response Technical Team/Coordination Center of China Department of Computer Science University of Southern California

Network representation learning, as an approach to learn low dimensional representations of vertices, has attracted considerable research attention recently. It has been proven extremely useful in many machine learning tasks over large graph. Most existing methods focus on learning the structural representations of vertices in a static network, but cannot guarantee an accurate and efficient embedding in a dynamic network scenario. The fundamental problem of continuously capturing the dynamic properties in an efficient way for a dynamic network remains unsolved. To address this issue, we present an efficient incremental skip-gram algorithm with negative sampling for dynamic network embedding, and provide a set of theoretical analyses to characterize the performance guarantee. Specifically, we first partition a dynamic network into the updated, including addition/deletion of links and vertices, and the retained networks over time. Then we factorize the objective function of network embedding into the added, vanished and retained parts of the network. Next we provide a new stochastic gradient-based method, guided by the partitions of the network, to update the nodes and the parameter vectors. The proposed algorithm is proven to yield an objective function value with a bounded difference to that of the original objective function. The first order moment of the objective difference converges in order of O（1/n;）, and the second order moment of the objective difference can be stabilized in order of O（1）. Experimental results show that our proposal can significantly reduce the training time while preserving the comparable performance. We also demonstrate the correctness of the theoretical analysis and the practical usefulness of the dynamic network embedding. We perform extensive experiments on multiple real-world large network datasets over multi-label classification and link prediction tasks to evaluate the effectiveness and efficiency of the proposed framework, and up to

关键词： dynamic network embedding bound and convergence analysis multi-label classification link prediction

Verifying Fairness in Quantum Machine Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Guan, Ji Fang, Wang Ying, Mingsheng State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing100190 China University of Chinese Academy of Sciences Beijing100049 China Department of Computer Science and Technology Tsinghua University Beijing100084 China

Due to the beyond-classical capability of quantum computing, quantum machine learning is applied independently or embedded in classical models for decision making, especially in the field of finance. Fairness and other ethical issues are often one of the main concerns in decision making. In this work, we define a formal framework for the fairness verification and analysis of quantum machine learning decision models, where we adopt one of the most popular notions of fairness in the literature based on the intuition - any two similar individuals must be treated similarly and are thus unbiased. We show that quantum noise can improve fairness and develop an algorithm to check whether a (noisy) quantum machine learning model is fair. In particular, this algorithm can find bias kernels of quantum data (encoding individuals) during checking. These bias kernels generate infinitely many bias pairs for investigating the unfairness of the model. Our algorithm is designed based on a highly efficient data structure - Tensor Networks - and implemented on Google's TensorFlow Quantum. The utility and effectiveness of our algorithm are confirmed by the experimental results, including income prediction and credit scoring on real-world data, for a class of random (noisy) quantum decision models with 27 qubits (227-dimensional state space) tripling (218 times more than) that of the state-of-the-art algorithms for verifying quantum machine learning models. © 2022, CC BY.

关键词： Quantum noise

Multi-MedVit: a deep learning approach for the diagnosis of COVID-19 with the CT images

学校读者我要写书评

暂无评论

Multi-MedVit: a deep learning approach for the diagnosis of ...

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Yunjie Cai Zeqi Zheng Shanling Nie Yue Guo Ruijie Zhang Hai Yang Department of Computer Science and Engineering East China University of Science and Technology Shanghai PR China Shanghai Key Laboratory of Computer Software Evaluating and Testing Shanghai China School of Electronics and Computer Science University of Southampton UK

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

The grim situation of novel coronavirus pneumonia 2019 (COVID-19) and its terrible spreading speed have already constituted a severe risk to human life, so it is ultimately essential to rapidly and accurately diagnose for COVID-19 pneumonia. Based on this study’s 746 lung CT images, we propose Multi-MedVit, a novel auxiliary COVID-19 diagnosis framework based on the multi-input Transformer. We compare Multi-MedVit with state-of-the-art deep learning methods, such as CNN, VGG16, and ResNet50. Multi-MedVit outperformed the other methods on the benchmark dataset and proved that multiscale data input for data augmentation helped enhance model stability. Based on an interpretable analysis of the input and output of Multi-MedVit, we found that with the support of the training set data, the model has been possible to accurately focus on the lesion area for diagnosis of COVID-19 without expert annotations, which can provide initial references containing more potential information to doctors more precisely and fleetly.

关键词： COVID-19 Deep learning Training Computed tomography Pulmonary diseases Lung Transformers

MrTF: Model Refinery for Transductive Federated Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Xin-Chun Yang, Yang Zhan, De-Chuan State Key Laboratory for Novel Software Technology Nanjing University Jiangsu Nanjing210023 China School of Artificial Intelligence Nanjing University Jiangsu Nanjing210023 China School of Computer Science and Engineering Nanjing University of Science and Technology Jiangsu Nanjing210094 China Department of Computing The Hong Kong Polytechnic University 100872 Hong Kong

We consider a real-world scenario in which a newly-established pilot project needs to make inferences for newly-collected data with the help of other parties under privacy protection policies. Current federated learning (FL) paradigms are devoted to solving the data heterogeneity problem without considering the to-be-inferred data. We propose a novel learning paradigm named transductive federated learning (TFL) to simultaneously consider the structural information of the to-be-inferred data. On the one hand, the server could use the pre-available test samples to refine the aggregated models for robust model fusion, which tackles the data heterogeneity problem in FL. On the other hand, the refinery process incorporates test samples into training and could generate better predictions in a transductive manner. We propose several techniques including stabilized teachers, rectified distillation, and clustered label refinery to facilitate the model refinery process. Abundant experimental studies verify the superiorities of the proposed Model refinery framework for Transductive Federated learning (MrTF). The source code is available at https://***/lxcnju/MrTF. © 2023, CC BY-NC-ND.

关键词： Distillation

Efficient Algorithms for Generalized Linear Bandits with Heavy-tailed Rewards

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Xue, Bo Wang, Yimu Wan, Yuanyu Yi, Jinfeng Zhang, Lijun Department of Computer Science City University of Hong Kong Hong Kong The City University of Hong Kong Shenzhen Research Institute Shenzhen China Cheriton School of Computer Science University of Waterloo Waterloo Canada School of Software Technology Zhejiang University Ningbo China JD AI Research Beijing China National Key Laboratory for Novel Software Technology Nanjing University Nanjing China Peng Cheng Laboratory Shenzhen China

This paper investigates the problem of generalized linear bandits with heavy-tailed rewards, whose (1 + ϵ)-th moment is bounded for some ϵ ∈ (0, 1]. Although there exist methods for generalized linear bandits, most of them focus on bounded or sub-Gaussian rewards and are not well-suited for many real-world scenarios, such as financial markets and web-advertising. To address this issue, we propose two novel algorithms based on truncation and mean of medians. These algorithms achieve an almost optimal regret bound of Oe(dT 1+ 1 ϵ ), where d is the dimension of contextual information and T is the time horizon. Our truncation-based algorithm supports online learning, distinguishing it from existing truncation-based approaches. Additionally, our mean-of-medians-based algorithm requires only O(log T) rewards and one estimator per epoch, making it more practical. Moreover, our algorithms improve the regret bounds by a logarithmic factor compared to existing algorithms when ϵ = 1. Numerical experimental results confirm the merits of our algorithms. Copyright © 2023, The Authors. All rights reserved.

关键词： Machine learning

Cognitive Disentanglement for Referring Multi-Object Tracking

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Liang, Shaofeng Guan, Runwei Lian, Wangwang Liu, Daizong Sun, Xiaolou Wu, Dongming Yue, Yutao Ding, Weiping Xiong, Hui Qingdao China Guangzhou China Shandong Key Laboratory of Intelligent Oil & Gas Industrial Software Qingdao China Wangxuan Institute of Computer Technology Peking University Beijing China School of Automation Southeast University Nanjing China School of Computer Science Beijing Institute of Technology Beijing China School of Artificial Intelligence and Computer Science Nantong University Nantong China Guangzhou China Department of Electrical Engineering and Electronics University of Liverpool Liverpool United Kingdom

As a significant application of multi-source information fusion in intelligent transportation perception systems, Referring Multi-Object Tracking (RMOT) involves localizing and tracking specific objects in video sequences based on language references. However, existing RMOT approaches often treat language descriptions as holistic embeddings and struggle to effectively integrate the rich semantic information contained in language expressions with visual features. This limitation is especially apparent in complex scenes requiring comprehensive understanding of both static object attributes and spatial motion information. In this paper, we propose a Cognitive Disentanglement for Referring Multi-Object Tracking (CDRMT) framework that addresses these challenges. It adapts the "what" and "where" pathways from human visual processing system to RMOT tasks. Specifically, our framework comprises three collaborative components: (1)The Bidirectional Interactive Fusion module first establishes cross-modal connections while preserving modality-specific characteristics;(2) Building upon this foundation, the Progressive Semantic-Decoupled Query Learning mechanism hierarchically injects complementary information into object queries, progressively refining object understanding from coarse to fine-grained semantic levels;(3) Finally, the Structural Consensus Constraint enforces bidirectional semantic consistency between visual features and language descriptions, ensuring that tracked objects faithfully reflect the referring expression. Extensive experiments on different benchmark datasets demonstrate that CDRMT achieves substantial improvements over state-of-the-art methods, with average gains of 6.0% in HOTA score on Refer-KITTI and 3.2% on Refer-KITTI-V2. Our approach advances the state-of-the-art in RMOT while simultaneously providing new insights into multi-source information fusion. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantics

Real-time Accident Anticipation for Autonomous Driving Through Monocular Depth-Enhanced 3D Modeling

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liao, Haicheng Li, Yongkang Wang, Chengyue Lai, Songning Li, Zhenning Bian, Zilin Lee, Jaeyoung Cui, Zhiyong Zhang, Guohui Xu, Chengzhong State Key Laboratory of Internet of Things for Smart City Department of Computer and Information Science University of Macau China Department of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China State Key Laboratory of Internet of Things for Smart City University of Macau China State Key Laboratory of Internet of Things for Smart City Departments of Civil and Environmental Engineering and Computer and Information Science University of Macau China Guangzhou China Transportation Planning and Engineering The Department of Civil and Urban Engineering New York University New York United States School of Traffic and Transportation Engineering Central South University Changsha China School of Transportation Science and Engineering Beihang University Beijing China Department of Civil and Environmental Engineering University of Hawaii HonoluluHI United States

The primary goal of traffic accident anticipation is to foresee potential accidents in real time using dashcam videos, a task that is pivotal for enhancing the safety and reliability of autonomous driving technologies. In this study, we introduce an innovative framework, AccNet, which significantly advances the prediction capabilities beyond the current state-of-the-art 2D-based methods by incorporating monocular depth cues for sophisticated 3D scene modeling. Addressing the prevalent challenge of skewed data distribution in traffic accident datasets, we propose the Binary Adaptive Loss for Early Anticipation (BA-LEA). This novel loss function, together with a multi-task learning strategy, shifts the focus of the predictive model towards the critical moments preceding an accident. We rigorously evaluate the performance of our framework on three benchmark datasets—Dashcam Accident Dataset (DAD), Car Crash Dataset (CCD), and AnAn Accident Detection (A3D), and DADA-2000 Dataset—demonstrating its superior predictive accuracy through key metrics such as Average Precision (AP) and mean Time-To-Accident (mTTA). Copyright © 2024, The Authors. All rights reserved.

关键词： Multi-task learning