检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Prajapat, Dharmendra Toshniwal, Durga Dept of Computer Science and Engineering Indian Institute of Technology Roorkee India

Task-oriented dialogue (TOD) system is designed to accomplish user-defined tasks through dialogues. The TOD system has progressed towards end-to-end modeling by leveraging pre-trained large language models. Fine-tuning the pre-trained language models using only supervised learning leads to the exposure bias and token loss problem and it deviates the models from completing the user's task. To address these issues, we propose a TOD system that leverages a unified pre-trained language model, GPT2, as a base model. It is optimized using supervised learning and reinforcement learning (RL). The issues in the TOD system are mitigated using a non-differentiable reward function. The reward is calculated using the weighted sum of the success rate and BLEU evaluation metrics. The success rate and BLEU metrics in reward calculation guide the language model for user task completion while ensuring a coherent and fluent response. Our model is acquired by fine-tuning a pre-trained model on the dialogue-session level which comprises user utterance, belief state, system act, and system response. Experimental results on MultiWOZ2.1 demonstrate that our model increases the inform rate by 1.60% and the success rate by 3.17% compared to the baseline. © 2024, CC0.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

KoDialogBench: Evaluating Conversational Understanding of Language Models with Korean Dialogue Benchmark

arXiv

引用

arXiv 2024年

作者： Jang, Seongbo Lee, Seonghyeon Yu, Hwanjo Dept. of Computer Science and Engineering POSTECH Pohang Korea Republic of

As language models are often deployed as chatbot assistants, it becomes a virtue for models to engage in conversations in a user’s first language. While these models are trained on a wide range of languages, a comprehensive evaluation of their proficiency in low-resource languages such as Korean has been lacking. In this work, we introduce KoDialogBench, a benchmark designed to assess language models’ conversational capabilities in Korean. To this end, we collect native Korean dialogues on daily topics from public sources, or translate dialogues from other languages. We then structure these conversations into diverse test datasets, spanning from dialogue comprehension to response selection tasks. Leveraging the proposed benchmark, we conduct extensive evaluations and analyses of various language models to measure a foundational understanding of Korean dialogues. Experimental results indicate that there exists significant room for improvement in models’ conversation skills. Furthermore, our in-depth comparisons across different language models highlight the effectiveness of recent training techniques in enhancing conversational proficiency. We anticipate that KoDialogBench will promote the progress towards conversation-aware Korean language models. © 2024, CC BY-NC-SA.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Leveraging Virtual Reality and AI Tutoring for Language Learning: A Case Study of a Virtual Campus Environment with OpenAI GPT Integration with Unity 3D

arXiv

引用

arXiv 2024年

作者： Adithya, T.G. Abhinavaram, N. Srinivasa, Gowri Dept. Computer Science & Engineering PES University Bangalore India

This paper presents a new approach to multiple language learning, with Hindi the language to be learn’t in our case, by using the integration of virtual reality environments and AI enabled tutoring systems using OpenAI’s GPT api calls. We have developed a scenario which has a virtual campus environment using Unity which focuses on a detailed representation of our university’s building's 11th floor, where most of the cultural and technological activities take place. Within this virtual environment that we have created, we have an AI tutor powered by OpenAI's GPT model which was called using an api which moves around with the user. This provided language learning support in Hindi, as GPT is able to take care of language translation. Our approach mainly involves utilising speech to text, text to text conversion and text to speech capabilities to facilitate real time interaction between users and the AI tutor in the presence of internet. This research demonstrates the use of combining VR technology with AI tutoring for immersive language learning experiences and provides interaction. Copyright © 2024, The Authors. All rights reserved.

关键词： Virtual environments

来源：评论

学校读者我要写书评

暂无评论

Machine Learning-Powered Defense Against Phishing Websites

Machine Learning-Powered Defense Against Phishing Websites

引用

IEEE Students' Conference on Electrical, Electronics and computer science (SCEECS)

作者： Dipesh Dwivedi Sajjad Ahmed Adarsh Patel H Azath Dept. of Computer Science and Engineering VIT University Sehore India

The proliferation of phishing URLs has experienced rapid growth in recent years, necessitating urgent attention to phishing attack detection in cybersecurity. In response, we introduce an improved predictive model that leverages both machine learning and a Deep Learning Model. Our dataset consists of 88,674 instances, encompassing 112 features, including the class level. Notably, this dataset comprises 58,000 legitimate instances and 30,647 phishing *** proposed method incorporates six distinct algorithms: Decision Tree, Support Vector Machine, K-nearest neighbors, Logistic Regression, Ensemble Classifier, and Neural Network algorithm. we systematically evaluated total 28 models performance with different types of feature changes. Following enhancements, every algorithm demonstrated an accuracy performance surpassing 93.16%. Among them, the Ensemble Classifier emerged as the most effective, boasting an accuracy 97.45%, MCC 94.37%, precision a96.35%, and an F1 Score 96.31%.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Automatic Authoring of Physical and Perceptual/Affective Motion Effects for Virtual Reality

arXiv

引用

arXiv 2024年

作者： Lee, Jiwan Choi, Seungmoon Dept. of Computer Science and Engineering POSTECH Pohang Korea Republic of

This video demo is about automatic authoring of various motion effects that are provided with audiovisual content to improve user experiences. Traditionally, motion effects have been used for simulators, e.g., flight simulators for pilots and astronauts, to present physically accurate vestibular feedback. At present, we have greatly wider use of motion effects for entertainment purposes, such as 4D rides in amusement parks and even shopping malls, 4D films in theaters, and relative new virtual reality games with head-mounted displays and personal motion platforms. However, the production of motion effects is done solely by manual authoring or coding, and this costly process prevents the faster and wider dissemination of 4D content. It is imperative to facilitate motion effect production by providing automatic synthesis algorithms. This demo introduces nine different automatic synthesis algorithms for motion effects and a recorded demonstration of each. All of these have been validated to deliver a reasonably competitive user experience in their respective studies. © 2024, CC BY.

关键词： Flight simulators

来源：评论

学校读者我要写书评

暂无评论

Adchat-TVQA: Innovative Application of LLMs-Based Text-Visual Question Answering Method in Advertising Legal Compliance Review

Adchat-TVQA: Innovative Application of LLMs-Based Text-Visua...

引用

Machine Learning and computer Application (ICMLCA), International Conference on

作者： Ruiling Gao Yingqi Hou Weijie Tang Jin Zhang Yanbo Hu Haiyun Huang Wen'an Tan Liping Li Dept. of Computer Engineering Shanghai Polytechnic University Shanghai China Dept. of Computer Science Donghua University Shanghai China Dept. of Computer Science Tufts University Medford MA USA

ISBN: (数字)9798331530334

ISBN: (纸本)9798331530341

Advertising legal compliance reviews have always been time-consuming and labor-intensive, and existing Large Language Models(LLMs) are far worse performing than senior industry experts. In this paper, we propose a novel LLMs-based text-visual question answering method called Adchat-TVQA for advertising legal compliance review. After reading an advertising image and the corresponding review question, it will first understand the content on the image through multimodal learning, then optimize the question with Zero-Shot Chain-of-Thought (CoT) prompting method based on predefined industry expert experience, and finally input the augmented question into the large language model for inquiry. The feasibility of this method is verified through system prototype implementation and end-to-end functional tests. In addition, we summarized the characteristics of advertising images and added image segmentation and text box transpose to the data processing process for system performance improvement. Performance testing of the text visual cognitive question task on the AIWIN2021 dataset shows that the method scores higher than Microsoft's LayoutLM, with an increasement of the score from 51.52% to 52.74%.

关键词： Industries Visualization Reviews Law System performance Prototypes Question answering (information retrieval) Prompt engineering Advertising Testing

来源：评论

学校读者我要写书评

暂无评论

Distributed Intrusion Detection System using Kafka and Spark Streaming

Distributed Intrusion Detection System using Kafka and Spark...

引用

2025 International Conference on Visual Analytics and Data Visualization, ICVADV 2025

作者： Kiran Kumar, Kotyada Mohan Srikar Reddy, M. Vivek Ullas, Karthik Supriya, M. Amrita School of Computing Dept. of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798331521394

Intrusion Detection Systems (IDS) are critical for identifying and mitigating potential security threats within network traffic. However, traditional IDS solutions often struggle with scalability and real-time threat detection, particularly in high-volume, high-velocity environments. The proposed work introduces a scalable and efficient IDS that leverages Apache Kafka and Apache Spark to address these challenges. Kafka's robust streaming platform ensures reliable, low-latency data flow, while Apache Spark's parallelized machine learning algorithms (MLlib) enable rapid and accurate classification of network traffic. By combining Kafka's data handling capabilities with Spark's processing efficiency, the proposed system provides fast, adaptive threat detection in real- time. This approach not only enhances IDS performance but also sets the stage for future developments in scalable, distributed IDS solutions. The work demonstrates the potential of big data technologies to improve network security in complex and dynamic network environments. © 2025 IEEE.

关键词： Network security

来源：评论

学校读者我要写书评

暂无评论

Vehicle Tracking Using Lane Cameras for Personalized Parking Guidance

Vehicle Tracking Using Lane Cameras for Personalized Parking...

引用

Consumer Technology (ICCT-Pacific), International Conference on

作者： Chun-Yao Zheng Chi-Yi Lin Dept. Computer Science and Information Engineering Tamkang University Taipei Taiwan

ISBN: (数字)9798331504120

ISBN: (纸本)9798331504137

Real-time monitoring of vehicle movements within a parking lot is crucial for creating a personalized parking guidance service. This applied research proposes using lane cameras to monitor vehicle dynamics, employing YOLOv8-based license plate object tracking and character recognition to determine the real-time location and movement direction of vehicles, while achieving cross-camera object tracking capabilities. With this critical information, the parking management system can provide tailored parking guidance instructions for each vehicle, reducing the burden on drivers to locate available parking spaces and contributing to lower carbon emissions.

关键词： Space vehicles Object detection Carbon dioxide Cameras Real-time systems Object tracking Character recognition Vehicle dynamics Monitoring License plate recognition

来源：评论

学校读者我要写书评

暂无评论

Forensic science: AI-Powered Image and Audio Analysis

Forensic Science: AI-Powered Image and Audio Analysis

引用

Smart Electronics and Communication (ICOSEC), International Conference on

作者： P Renukadevi Sincy John N Shivani Dept. of Computer Science and Engineering Jain University Bangalore India

ISBN: (数字)9798331504403

ISBN: (纸本)9798331504410

Forensic science demands the precise application of advanced scientific principles and procedures to investigate crimes and ensure justice. This process involves the analysis of vast and complex data. The research explores the potential of modern science and technologies, particularly Artificial Intelligence (AI), to develop innovative techniques across various branches of forensic science. It provides a comprehensive examination of AI applications in forensic fields, focusing on image and audio analysis to identify instruments used in crimes and determine causes of death. By integrating AI, the research aims to enhance the accuracy and efficiency of forensic investigations while maintaining ethical standards. The paper also discusses the difficulties with AI in forensic science, highlighting the need for open and moral AI systems. It highlights that forensic errors leading to wrongful convictions often result from factors such as incompetence, spam, weak scientific foundations, or organizational deficiencies, rather than blunders made by forensic scientists. These issues highlight the importance of addressing the root causes of forensic errors to improve the reliability of forensic practices. The study also delves into the specific challenges AI encounters, such as data quality, interpretability of AI models, and the unification of AI into current forensic procedures. By tackling these challenges, the research aims to ensure that AI applications in forensic science are effective and trustworthy. Overall, this work highlights the critical role of AI in advancing forensic science and the importance of addressing the underlying issues that contribute to forensic errors, ultimately striving for a more reliable and just forensic process

关键词： Ethics Accuracy Forensics Instruments Data integrity Focusing Data models Reliability Artificial intelligence Standards

来源：评论

学校读者我要写书评

暂无评论

A Novel High Frame Rate and High Contrast Coherent Plane Wave Compounding Approach Utilizing Euclidean Distance Transform

A Novel High Frame Rate and High Contrast Coherent Plane Wav...

引用

IEEE International Symposium on Applications of Ferroelectrics (ISAF)

作者： Sajjad Afrakhteh Libertario Demi Dept. of Information Engineering and Computer Science University of Trento Italy

ISBN: (数字)9798350371901

ISBN: (纸本)9798350371918

To address the challenge of reduced image quality in plane-wave imaging (PWI), coherent plane-wave compounding (CPWC) has been introduced. CPWC combines plane wave images from various directions (i.e., with different angles) to enhance image quality. However, the number of angles required to achieve satisfactory image quality directly impacts the maximum attainable frame rate in CPWC. Consequently, there exists a tradeoff between image quality, particularly contrast, and frame rate. In this study, our goal is to mitigate this compromise by enhancing image contrast while simultaneously improving the frame rate of CPWC imaging. Our proposed technique includes skeleton-based spatial filtering of each beamformed radio frequency (RF) image at each angle before coherent compounding. For this purpose, we introduce a novel filtering technique based on the Euclidean distance transform (EDT). EDT assigns the minimum distance from each foreground pixel to the background. Given that lower contrast in CPWC imaging stems from information leakage (i.e., higher sidelobe levels), we propose using the EDT to reduce this effect, thereby improving contrast. We implemented our proposed technique using the Plane Wave Imaging Challenge in Medical Ultrasound (PICMUS) datasets.

关键词： Image quality Radio frequency Ultrasonic imaging Filtering Databases Transforms Euclidean distance Information leakage Frequency control Biomedical imaging

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：