检索结果-内蒙古大学图书馆

international Joint conference on Neural Networks (IJCNN)

作者： Wang, ZhenYu Wu, Yifei Donghua Univ Sch Comp Sci & Technol Shanghai Peoples R China

ISBN: (纸本)9798350359329;9798350359312

Sentence representation is a major challenge in natural language processing, especially in multilingual environments. Current approaches to sentence representation using Large language Models (LLMs) often require large amounts of data for fine-tuning, and research has focused on English content. In addition, comparative datasets translated directly from English can contain many semantic and syntactic errors. To address these issues, we propose a new approach to enhance multilingual sentence embeddings using LLMs and knowledge graphs. We first present a dedicated designed prompt that exploits in-context learning of LLMs for sentence embedding without fine-tuning. We further introduce an innovative method that utilizes knowledge graphs, such as Wikidata, for generating diverse multilingual training data for contrastive finetuning. This approach significantly reduces the reliance on translated sentences and mitigates issues related to translation accuracy. Furthermore, we develop a unique multilingual contrastive learning loss function, which, when combined with QLora's efficient fine-tuning technique, enables LLMs to achieve state-of-the-art performance in Sentence Text Similarity (STS) tasks, even with limited computational resources.

关键词： sentence embedding contrastive learning large language model data argumentation

来源：评论

学校读者我要写书评

暂无评论

AI-Powered Mental Health Assessment using Emotion Detection for Real-Time Analysis

AI-Powered Mental Health Assessment using Emotion Detection ...

引用

2025 ieee international conference on Computational, Communication and Information Technology, ICCCIT 2025

作者： Vijay, K. Raghakeerthana, R. Milinda, Renata Rachael Thusheel, S. Rajalakshmi Engineering College Department of Artificial Intelligence and Machine Learning Chennai India

ISBN: (纸本)9798331512965

This paper is an AI-driven mental health assessment tailored to each patient's distinct emotional needs, addressing the global mental health crisis. Using advanced machine learning and natural language processing, it aims to deliver personalized guidance for psychological and physical wellness activities, along with tools for tracking individual progress. The papers employ exploratory case studies and knowledge discovery techniques to evaluate the effectiveness of these interventions. Data mining methods reveal hidden patterns within mental health datasets, with the potential to reshape therapeutic paradigms. Expected outcomes include enhanced accessibility to mental health resources, early intervention, personalized progress monitoring, improved quality of life, and reduced stigma. By integrating sophisticated data processing, machine learning, and empathetic user interaction, it operates by redefining mental health support. The system's architecture is designed to promote both emotional and cognitive well-being, fostering positive change in mental health care delivery. © 2025 ieee.

关键词： Computer Vision conversational bot Convolution neural network Emotion detection Machine learning natural language processing Real-Time Analysis

来源：评论

学校读者我要写书评

暂无评论

Semantic-aware Source Code Modeling 24

Semantic-aware Source Code Modeling

引用

39th ACM/ieee international conference on Automated Software engineering (ASE)

作者： Ding, Yangruibo Columbia Univ New York NY 10027 USA

ISBN: (数字)9798400712487

ISBN: (纸本)9798400712487

Source code modeling represents a promising avenue for automating software development, such as code generation, bug repair, and program analysis. This research direction aims to train deep neural nets to learn the statistical predictability inherent in human-written programs to enhance developer productivity, code quality, and the overall software development life cycle. Although existing code modeling approaches, particularly those underpinned by Transformer-based language models, have demonstrated effectiveness across various software engineering tasks, most of them have directly adopted learning schemes from natural language processing (e.g., data collection and processing, training objectives) to source code, primarily focusing on learning code text and syntax. However, such a direct transplant limits the models' capability to capture deep program semantics, such as code functionality, dependencies, and program states during execution. In this research proposal, we highlight the critical role of program semantics in source code modeling. We propose a range of innovative methodologies to bridge the gap between the text-based language models for large-scale code training and the requirement of deep semantic understanding to assist with software engineering tasks effectively. Furthermore, we showcase the efficacy of the proposed semantic-aware code modeling through a handful of published papers and preliminary results, with motivations to delve deeper into this avenue during doctoral research.

关键词： Training Codes Source coding Semantics Predictive models Syntactics Transformers Data models Software engineering Software development management

来源：评论

学校读者我要写书评

暂无评论

Intelligent customer service system of subway based on semantic decision-making judgment 4

Intelligent customer service system of subway based on seman...

引用

4th international conference on Machine Learning and Intelligent Systems engineering (MLISE)

作者： Liu, Xinyu Shanghai Univ Engn Sci Sch Urban Railway Transportat Shanghai Peoples R China

ISBN: (纸本)9798350375084;9798350375077

This study introduces an intelligent customer service system for urban rail transit, addressing inefficiencies and high labor costs in existing systems' problem classification and feedback mechanisms. Leveraging semantic understanding, the system integrates advanced speech recognition, intention classification, and entity extraction to automate the processing and intelligent classification of passenger hotline calls. Post-evaluation, the system demonstrated a minimum of 4.55% enhancement in data processing accuracy and F1 score, respectively, validating its efficacy in enhancing efficiency and precision. Furthermore, the research delves into the practical application of entity recognition, offering insights into the integration of natural language processing in intelligent customer service systems.

关键词： natural language processing semantic recognition text classification entity recognition

来源：评论

学校读者我要写书评

暂无评论

Axolotl: Fairness through Assisted Prompt Rewriting of Large language Model Outputs 15

Axolotl: Fairness through Assisted Prompt Rewriting of Large...

引用

15th ieee international conference on knowledge Graph, ICKG 2024

作者： Ebrahimi, Sana Chen, Kaiwen Asudeh, Abolfazl Das, Gautam Koudas, Nick University of Illinois Chicago Department of Computer Science United States University of Toronto Department of Computer Science Canada University of Texas at Arlington Department of Computer Science United States

ISBN: (纸本)9798331508821

Pre-trained Large language Models (LLMs) have significantly advanced natural language processing capabilities but are susceptible to biases present in their training data, leading to unfair outcomes in various applications. While numerous strategies have been proposed to mitigate bias, they often require extensive computational resources and may compromise model *** this work, inspired by the concept of query rewriting, we propose prompt rewriting to minimize the unfairness of LLM outputs. We introduce Axolotl, a novel post-processing framework that operates agnostically across tasks and models, leveraging public APIs to interact with LLMs without direct access to internal parameters. Through a three-step process, Axolotl identifies biases, proposes resolutions, and guides the model to self-debias its outputs. This approach minimizes computational costs and preserves model performance, making Axolotl a promising tool for debiasing LLM outputs with broad applicability and ease of use. © 2024 ieee.

关键词： Structured Query language

来源：评论

学校读者我要写书评

暂无评论

Factoid Question Answering System using knowledge Graph 7

Factoid Question Answering System using Knowledge Graph

引用

2023 7th international conference On Computing, Communication, Control And Automation, ICCUBEA 2023

作者： Gaikwad, Vikas Patil, Archana College of Engineering Dept. of Computer Engineering and It Pune India

ISBN: (纸本)9798350304268

Here we discusses various topics related to natural language processing and question answering systems. It explains the difference between reading comprehension QA and knowledge Graph Question Answering (KGQA) and highlights the importance of world knowledge for dialogue systems. The text then goes on to describe a simple question answering system and the limitations of such systems in handling complex questions. Finally, the concept of knowledge graphs is introduced and their role in NLP applications is discussed. Overall, the text provides an overview of key concepts and techniques used in question answering systems and their potential applications in various fields. © 2023 ieee.

关键词： Dialogue Systems knowledge Graph natural language processing Question Answering System

来源：评论

学校读者我要写书评

暂无评论

Enhancing Telugu Abusive language Detection Using Word Embeddings and BERT Models 2

Enhancing Telugu Abusive Language Detection Using Word Embed...

引用

2nd international conference on Recent Trends in Microelectronics, Automation, Computing, and Communications Systems, ICMACC 2024

作者： Battula, Venkataramana Nune, Nikhil Teja Guptha, Vivek Rampuri, Shiva Sai Chacham, Pradyumna Pyarasani, Koushik Bojji, Anirudh MVSR Engineering College Computer Science and Engineering Hyderabad India

ISBN: (纸本)9798350366570

In the digital era, the surge of abusive content online underscores the necessity for robust content moderation. This study targets the challenges posed by abusive content in Telugu, a language with limited digital resources. The goal is an advanced framework to classify Telugu sentences as abusive or not, using FastText and IndicFast for word embedding. Various machine learning techniques like Decision Trees, SVM, KNN, Random Forest, RNN, LSTM, BiLSTM, GRU, and transfer learning with pre-trained BERT models are explored. Each method is evaluated for its ability to capture abusive language nuances. The study employs diverse word vector dimensions for comprehensive coverage, systematically evaluating techniques to identify optimal classification combinations. It also investigates transfer learning via pre-trained BERT models, leveraging knowledge from other datasets to enhance Telugu abusive language understanding. This research enriches natural language processing and content moderation knowledge. It offers insights into effective word embedding techniques and ML algorithms for low-resource languages, aiding in the development of sensitive content moderation tools for Telugu speakers, fostering a safer online space. © 2024 ieee.

关键词： Random forests

来源：评论

学校读者我要写书评

暂无评论

ENHANCED TRANSFER LEARNING WITH EFFICIENT MODELING AND ADAPTIVE FUSION OF knowledge VIA PROMPT TUNING 49

ENHANCED TRANSFER LEARNING WITH EFFICIENT MODELING AND ADAPT...

引用

49th ieee international conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Xu, Minghui Guo, Zishan Zeng, Yulong Xiong, Deyi Tianjin Univ Coll Intelligence & Comp Tianjin Peoples R China Huawei Noahs Ark Lab Shenzhen Peoples R China

ISBN: (纸本)9798350344868;9798350344851

This work presents a novel and parameter-efficient transfer learning framework. The framework consists of two phases: knowledge modeling based on prompt decomposition and knowledge transfer based on attention. Specifically, during the first phase, we decompose the prompt into parameter spaces of different ranks and leverage their characteristics to precisely model both general knowledge and task-specific knowledge separately. During the second phase, we train an attention module to adaptively integrate task-specific knowledge and generate an instance-wise prompt, which is then further fine-tuned. Through these two stages, our approach can accurately and efficiently transfer task knowledge of different granularities and types based on the input sample during inference. Extensive experiments demonstrate that our approach outperforms the multi-task learning baselines and SOTA parameter-efficient transfer learning methods. Furthermore, despite using only 0.13% of the parameters compared to full-parameter fine-tuning, it achieves an absolute improvement ranging from 0.3 to 3.9 across different benchmarks.

关键词： Parameter-efficient fine-tuning Prompt tuning natural language processing Transfer Learning

来源：评论

学校读者我要写书评

暂无评论

An Overview of the ADReSS-M Signal processing Grand Challenge on Multilingual Alzheimer's Dementia Recognition Through Spontaneous Speech

IEEE OPEN JOURNAL OF SIGNAL PROCESSING

引用

ieee OPEN JOURNAL OF SIGNAL processing 2024年 5卷 738-749页

作者： Luz, Saturnino Haider, Fasih Fromm, Davida Lazarou, Ioulietta Kompatsiaris, Ioannis Macwhinney, Brian Univ Edinburgh Usher Inst Edinburgh Med Sch Edinburgh EH9 3JW Scotland Univ Edinburgh Sch Engn Edinburgh EH9 3JW Scotland Carnegie Mellon Univ Dept Psychol Pittsburgh PA 15213 USA CERTH Informat Technol Inst Thessaloniki 57001 Thermi Thessalo Greece

The ADReSS-M Signal processing Grand Challenge was held at the 2023 ieee international conference on Acoustics, Speech and Signal processing, ICASSP 2023. The challenge targeted difficult automatic prediction problems of great societal and medical relevance, namely, the detection of Alzheimer's Dementia (AD) and the estimation of cognitive test scoress. Participants were invited to create models for the assessment of cognitive function based on spontaneous speech data. Most of these models employed signal processing and machine learning methods. The ADReSS-M challenge was designed to assess the extent to which predictive models built based on speech in one language generalise to another language. The language data compiled and made available for ADReSS-M comprised English, for model training, and Greek, for model testing and validation. To the best of our knowledge no previous shared research task investigated acoustic features of the speech signal or linguistic characteristics in the context of multilingual AD detection. This paper describes the context of the ADReSS-M challenge, its data sets, its predictive tasks, the evaluation methodology we employed, our baseline models and results, and the top five submissions. The paper concludes with a summary discussion of the ADReSS-M results, and our critical assessment of the future outlook in this field.

关键词： Biomedical signal processing medical conditions Alzheimer's disease human disease biomarkers speech processing natural language processing multilingual Alzheimer's dementia detection

来源：评论

学校读者我要写书评

暂无评论

A Dynamic Retrieval-Augmented Generation Framework for Border Inspection Legal Question Answering

A Dynamic Retrieval-Augmented Generation Framework for Borde...

引用

28th international conference on Asian language processing (IALP)

作者： Zhang, Yanjun Li, Dapeng Peng, Gaojun Guo, Shuang Dou, Yu Yi, Ruheng Inner Mongolia Univ Technol Coll Data Sci & Applicat Hohhot Peoples R China Natl Immigrat Adm Hohhot EntryExit Frontier Inspect Stn Hohhot Peoples R China

ISBN: (纸本)9798331540869;9798331540852

Border inspection legal question answering (LQA) is designed for specific legal scenarios, aiming to address legal questions related to border inspections and provide accurate and practical legal guidance to the public and border inspection departments. Current research requires a retrieval process regardless of whether the necessary knowledge already exists, and does not evaluate the legal information in the generated responses, leading to errors and low-quality content. we propose the Dynamic Retrieval-Augmented Generation framework for Border Inspection LQA (DRAG-BILQA), which generates responses first and then dynamically controls retrieval based on confidence scores. In this framework, we innovatively introduce a legal factor recognition module to ensure the legal accuracy of generated answers and improve the quality of the generation. We also present the BorderLegal-QA dataset, filling the gap in border inspection legal datasets. We conducted extensive experiments on both the BorderLegal-QA and the general LQA dataset JEC-QA. Experimental results show that DRAG-BILQA outperforms Sota models in terms of legal factor recognition accuracy and retrieval performance, achieving a METEOR score of 39.5 and improving generation quality. Case analysis also demonstrate that DRAG-BILQA effectively Augmented the legal accuracy of the generated content.

关键词： border inspection LQA large language model retrieval-augmented generation

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：