检索结果-内蒙古大学图书馆

Layer-Wise Learning Rate Optimization for Task-Dependent Fine-Tuning of Pre-Trained Models: An Evolutionary Approach

引用

ACM Transactions on Evolutionary Learning and Optimization 2024年第4期4卷 1-23页

作者： Bu, Chenyang Liu, Yuxin Huang, Manzong Shao, Jianxuan Ji, Shengwei Luo, Wenjian Wu, Xindong Key Laboratory of Knowledge Engineering with Big Data Ministry of Education and School of Computer Science and Information Engineering Hefei University of Technology Hefei China School of Artificial Intelligence and Big Data Hefei University Hefei China Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies School of Computer Science and Technology Harbin Institute of Technology Shenzhen China

The superior performance of large-scale pre-Trained models, such as Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-Trained Transformer (GPT), has received increasing attention in both academic and industrial research and has become one of the current research hotspots. A pre-Trained model refers to a model trained on large-scale unlabeled data, whose purpose is to learn general language representation or features for fine-Tuning or transfer learning in subsequent tasks. After pre-Training is complete, a small amount of labeled data can be used to fine-Tune the model for a specific task or domain. This two-stage method of "pre-Training+fine-Tuning"has achieved advanced results in natural language processing (NLP) tasks. Despite widespread adoption, existing fixed fine-Tuning schemes that adapt well to one NLP task may perform inconsistently on other NLP tasks given that different tasks have different latent semantic structures. In this article, we explore the effectiveness of automatic fine-Tuning pattern search for layer-wise learning rates from an evolutionary optimization perspective. Our goal is to use evolutionary algorithms to search for better task-dependent fine-Tuning patterns for specific NLP tasks than typical fixed fine-Tuning patterns. Experimental results on two real-world language benchmarks and three advanced pre-Training language models show the effectiveness and generality of the proposed framework. © 2024 held by the owner/author(s).

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Multi-Cluster Feature Selection Based on Isometric Mapping

引用

IEEE/CAA Journal of Automatica Sinica 2022年第3期9卷 570-572页

作者： Yadi Wang Zefeng Zhang Yinghao Lin Henan Key Laboratory of Big Data Analysis and Processing Henan UniversityKaifeng 475004 Institute of Data and Knowledge Engineering School of Computer and Information EngineeringHenan UniversityKaifeng 475004China

Dear editor,This letter presents an unsupervised feature selection method based on machine *** selection is an important component of artificial intelligence,machine learning,which can effectively solve the curse of d... 详细信息

关键词： problem letter dimensionality

来源：评论

学校读者我要写书评

暂无评论

knowledge Graph for China's Genealogy1

引用

IEEE Transactions on knowledge and data engineering 2023年第1期35卷 634-646页

作者： Wu, Xindong Jiang, Tingting Zhu, Yi Bu, Chenyang Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education Anhui Hefei230009 China Hefei University of Technology Research Institute of Big Knowledge Anhui Hefei230009 China Mininglamp Technology Mininglamp Academy of Sciences Beijing100102 China Hefei University of Technology School of Computer Science and Information Engineering Anhui Hefei230009 China Hefei University of Technology Ministry of Education Key Laboratory of Knowledge Engineering with Big Data Anhui Hefei230009 China

Genealogical knowledge graphs depict the relationships of family networks and the development of family histories. They can help researchers to analyze and understand genealogical data, search for genealogical descendant paths, and explore the origins of a family. However, the heterogenous, autonomous, complex, and evolving natures of genealogical data bring challenges to the development of contemporary genealogical knowledge graph models. Applying existing methods to genealogical data may be improper because general knowledge graph models lack in-depth domain knowledge. In this paper, we propose a genealogical knowledge graph model named Huapu-KG that combines HAO intelligence (human intelligence + artificial intelligence + organizational intelligence) to implement the construction and applications of genealogical knowledge graphs. Furthermore, challenges in constructing genealogical knowledge graphs are demonstrated, and experiments conducted on real-world genealogical datasets verify the feasibility and effectiveness of our proposed model. © 1989-2012 IEEE.

关键词： History

来源：评论

学校读者我要写书评

暂无评论

PCQPR: Proactive Conversational Question Planning with Reflection

PCQPR: Proactive Conversational Question Planning with Refle...

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Guo, Shasha Liao, Lizi Zhang, Jing Li, Cuiping Chen, Hong School of Information Renmin University of China Beijing China Key Laboratory of Data Engineering and Knowledge Engineering of Ministry of Education China Singapore Management University Singapore

ISBN: (纸本)9798891761643

Conversational Question Generation (CQG) enhances the interactivity of conversational question-answering systems in fields such as education, customer service, and entertainment. However, traditional CQG, focusing primarily on the immediate context, lacks the conversational foresight necessary to guide conversations toward specified conclusions. This limitation significantly restricts their ability to achieve conclusion-oriented conversational outcomes. In this work, we redefine the CQG task as Conclusion-driven Conversational Question Generation (CCQG) by focusing on proactivity, not merely reacting to the unfolding conversation but actively steering it towards a conclusion-oriented question-answer pair. To address this, we propose a novel approach, called Proactive Conversational Question Planning with self-Refining (PCQPR). Concretely, by integrating a planning algorithm inspired by Monte Carlo Tree Search (MCTS) with the analytical capabilities of large language models (LLMs), PCQPR predicts future conversation turns and continuously refines its questioning strategies. This iterative self-refining mechanism ensures the generation of contextually relevant questions strategically devised to reach a specified outcome. Our extensive evaluations demonstrate that PCQPR significantly surpasses existing CQG methods, marking a paradigm shift towards conclusion-oriented conversational question-answering systems. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Enhancing Extractive Question Answering in Multiparty Dialogues with Logical Inference Memory Network 31

Enhancing Extractive Question Answering in Multiparty Dialog...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Zhou, Shu Zhao, Rui Zhou, Zhengda Yi, Haohan Zheng, Xuhui Wang, Hao Nanjing University China Key Laboratory of Data Engineering and Knowledge Services in Jiangsu Provincial Universities Nanjing University China University of Technology Sydney Australia

ISBN: (纸本)9798891761964

Multiparty dialogue question answering (QA) within machine reading comprehension (MRC) presents significant challenges due to the complex interplay of information across multiple speakers and the need for advanced logical reasoning. While existing models often focus on separating dialogue information based on speakers and utterances, they rarely address the crucial aspect of logical inference, leading to suboptimal performance in understanding and answering questions. To bridge this gap, we introduce the Logical Inference Memory Network (LIMN), a novel architecture designed for extractive QA in multiparty dialogues. LIMN incorporates a unique inference module pretrained on plain text QA datasets (like SQuAD 2.0), enabling it to transfer robust logical reasoning abilities to the dialogue domain. This module generates representations that are specifically attuned to logical inference, which are then integrated into the dialogue context. Furthermore, we propose a key-utterance-based interaction mechanism that dynamically focuses on the most relevant utterances within the dialogue, enhancing the model's ability to pinpoint answers. To ensure robust performance, LIMN employs a multitask learning strategy that jointly optimizes for answer extraction, answerability prediction, key-utterance identification, and masked speaker prediction. Extensive experiments on the Molweni and FriendsQA benchmarks, encompassing 25,000 and 10,000 questions respectively, demonstrate that LIMN achieves state-of-the-art results, affirming the effectiveness of incorporating logical inference in multiparty dialogue QA. © 2025 Association for Computational Linguistics.

关键词： Memory architecture

来源：评论

学校读者我要写书评

暂无评论

Face micro-expression recognition algorithm based on ResNet depth model 8

Face micro-expression recognition algorithm based on ResNet ...

引用

8th International Conference on Intelligent Computing and Signal Processing, ICSP 2023

作者： Wang, Liquan Zhan, Shu Hefei University of Technology Key Laboratory of Big Data Knowledge Engineering Ministry of Education School of Computer and Information Engineering Hefei China

ISBN: (纸本)9798350302455

Micro Expression (ME) is the subtle facial expressions that people show when they express their inner feelings. To address the problem that micro-expression recognition is difficult and less accurate due to the small number of samples and uneven distribution of different categories, we propose a model framework to improve the accuracy of micro-expression recognition. The peak frames containing more key expression information in the micro-expression video sequences are extracted;SE-ResNeXt-50, an improved residual network with SE module, is used to extract features from the peak frames of micro-expressions, where the SE module can better learn the key information in the features, and ResNeXt simplifies the structure by replacing the dense structure with the sparse structure through group convolution, which improves the recognition efficiency. The recognition efficiency is improved by replacing the dense structure with the sparse structure by group convolution. At the same time, the Focal Loss loss function can better solve the model performance problem caused by the imbalance of micro-expression data. Simulation experiments are conducted on the micro-expression dataset CASME, and it is found that the improved residual network and peak frame improve the accuracy and F1 value of micro-expression recognition. The improved residual network and peak frame can reduce the effect of small data set, make the model have good fitting effect, and improve the performance of different categories, improve the recognition accuracy of micro-expressions, and have better recognition performance for micro-expression recognition. © 2023 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Bootstrap-Based Layerwise Refining for Causal Structure Learning

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2024年第6期5卷 2708-2722页

作者： Xiang, Guodu Wang, Hao Yu, Kui Guo, Xianjie Cao, Fuyuan Song, Yukun Hefei University of Technology Key Laboratory of Knowledge Engineering with the Big Data of Ministry of Education Hefei230601 China Hefei University of Technology School of Computer Science and Information Engineering Hefei230601 China Shanxi University School of Computer and Information Technology Taiyuan030006 China

Learning causal structures from observational data is critical for causal discovery and many machine learning tasks. Traditional constraint-based methods first adopt conditional independence (CI) tests to learn a global skeleton layer by layer and then orient the undirected edges to obtain a causal structure. However, the reliability of these statistical tests largely depends on the quality of data samples. In real-life scenarios, the presence of data noise or limited samples often makes many CI tests unreliable at each layer in the skeleton learning phase, leading to an inaccurate skeleton. As the number of layers increases, the inaccurate skeleton will continue to impair the skeleton construction of subsequent layers. Furthermore, an unreliable skeleton hampers the skeleton orientation procedure, resulting in an unsatisfactory causal structure. In this article, we propose a Bootstrap-based layerwise refining (BLR) algorithm for causal structure learning, which includes two new procedures to solve the above problems. First, BLR utilizes a novel layerwise skeleton refining procedure to construct the global skeleton layer by layer based on the bootstrap sampling. Second, BLR employs a collective skeleton orientation procedure that incorporates scoring techniques to collectively orient the global skeleton. The experimental results show that BLR outperforms the state-of-the-art methods on the benchmark Bayesian Network datasets. © 2020 IEEE.

关键词： Refining

来源：评论

学校读者我要写书评

暂无评论

Histological Image Diagnosis of Breast Cancer Based on Multi-Attention Convolution Neural Network

引用

Journal of Shanghai Jiaotong university(Science) 2025年第1期30卷 91-106页

作者： XU Wangwang XU Liangfeng LIU Ninghui LU Na Key Laboratory of Knowledge Engineering with Big Data(Hefei University of Technology) Ministry of EducationHefei230601China School of Computer and Information Hefei University of TechnologyHefei230601China First Affiliated Hospital of Anhui Medical University Hefei230022China

Breast cancer is a serious and high morbidity disease in women,and it is the main cause of cancer death in ***,getting tested and diagnosed early can reduce the risk of *** present,there are clinical examinations,imaging screening and biopsies,among which histopathological examination is the gold ***,the process is complicated and time-consuming,and misdiagnosis may *** paper puts forward a classification framework based on deep learning,introducing multi-attention mechanism,selecting kernel convolution instead of ordinary convolution,and using different weights and combinations to pay attention to the accuracy index and growth rate of the *** addition,we also compared the learning rate *** function can fine-tune the learning rate to achieve good performance,using label softening to reduce the loss error caused by model error recognition in the label,and assigning different category weights in the loss function to balance the positive and negative *** used the BreakHis data set to automatically classify histological images into benign and malignant,four categories and eight *** results showed that the accuracy of binary classifications ranged from 98.23%to 98.83%,and that of multiple classifications ranged from 97.89%to 98.11%.

关键词： breast cancer deep learning attentional mechanism classification diagnosis

来源：评论

学校读者我要写书评

暂无评论

Joint user profiling with hierarchical attention networks

引用

Frontiers of Computer Science 2023年第3期17卷 133-143页

作者： Xiaojian LIU Yi ZHU Xindong WU Key Laboratory of Knowledge Engineering with Big Data(Hefei University of Technology) Ministry of EducationHefei 230009China School of Computer Science and Information Engineering Hefei University of TechnologyHefei 230009China School of Information Engineering Yangzhou UniversityYangzhou 225009China Mininglamp Academy of Sciences MininglampBeijing 100193China

User profiling by inferring user personality traits,such as age and gender,plays an increasingly important role in many real-world *** existing methods for user profiling either use only one type of data or ignore handling the noisy information of ***,they usually consider this problem from only one *** this paper,we propose a joint user profiling model with hierarchical attention networks(JUHA)to learn informative user representations for user *** JUHA method does user profiling based on both inner-user and inter-user *** explore inner-user features from user behaviors(e.g.,purchased items and posted blogs),and inter-user features from a user-user graph(where similar users could be connected to each other).JUHA learns basic sentence and bag representations from multiple separate sources of data(user behaviors)as the first round of data *** this module,convolutional neural networks(CNNs)are introduced to capture word and sentence features of age and gender while the self-attention mechanism is exploited to weaken the noisy *** this,we build another bag which contains a user-user ***-user features are learned from this bag using propagation information between linked users in the *** acquire more robust data,inter-user features and other inner-user bag representations are joined into each sentence in the current bag to learn the final bag ***,all of the bag representations are integrated to lean comprehensive user representation by the self-attention *** experimental results demonstrate that our approach outperforms several state-of-the-art methods and improves prediction performance.

关键词： user profiling hierarchical attention joint learning inner-user feature inter-user feature

来源：评论

学校读者我要写书评

暂无评论

MacroTrend: A Write-Efficient Cache Algorithm for NVM-Based Read Cache

引用

Journal of Computer Science & Technology 2022年第1期37卷 207-230页

作者： Ning Bao Yun-Peng Chai Xiao Qin Chuan-Wen Wang Key Laboratory of Data Engineering and Knowledge Engineering Ministry of EducationBeijing 100872China School of Information Renmin University of ChinaBeijing 100872China Samuel Ginn College of Engineering Auburn UniversityAlabama 36830U.S.A.

The future storage systems are expected to contain a wide variety of storage media and layers due to the rapid development of NVM(non-volatile memory)*** NVM-based read caches,many kinds of NVM devices cannot stand frequent data updates due to limited write endurance or high energy consumption of ***,traditional cache algorithms have to update cached blocks frequently because it is difficult for them to predict long-term popularity according to such limited information about data blocks,such as only a single value or a queue that reflects frequency or *** this paper,we propose a new MacroTrend(macroscopic trend)prediction method to discover long-term hot blocks through blocks'macro trends illustrated by their access count *** then a new cache replacement algorithm is designed based on the MacroTrend prediction to greatly reduce the write amount while improving the hit *** conduct extensive experiments driven by a series of real-world traces and find that compared with LRU,MacroTrend can reduce the write amounts of NVM cache devices significantly with similar hit ratios,leading to longer NVM lifetime or less energy consumption.

关键词： non-volatile memory(NVM) solid state disk(SSD) cache endurance

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：