检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Karten, Seth Nguyen, Andy Luu Jin, Chi Department of Computer Science Princeton University United States Department of Electrical and Computer Engineering Princeton University United States

We introduce PokéChamp, a minimax agent powered by Large Language Models (LLMs) for Pokémon battles. Built on a general framework for two-player competitive games, PokéChamp leverages the generalist capabilities of LLMs to enhance minimax tree search. Specifically, LLMs replace three key modules: (1) player action sampling, (2) opponent modeling, and (3) value function estimation, enabling the agent to effectively utilize gameplay history and human knowledge to reduce the search space and address partial observability. Notably, our framework requires no additional LLM training. We evaluate PokéChamp in the popular Gen 9 OU format. When powered by GPT-4o, it achieves a win rate of 76% against the best existing LLM-based bot and 84% against the strongest rule-based bot, demonstrating its superior performance. Even with an open-source 8-billion-parameter Llama 3.1 model, PokéChamp consistently outperforms the previous best LLM-based bot, Pokéllmon powered by GPT-4o, with a 64% win rate. PokéChamp attains a projected Elo of 1300-1500 on the Pokémon Showdown online ladder, placing it among the top 30%-10% of human players. In addition, this work compiles the largest real-player Pokémon battle dataset, featuring over 3 million games, including more than 500k high-Elo matches. Based on this dataset, we establish a series of battle benchmarks and puzzles to evaluate specific battling skills. We further provide key updates to the local game engine. We hope this work fosters further research that leverage Pokémon battle as benchmark to integrate LLM technologies with game-theoretic algorithms addressing general multiagent problems. Videos, code, and dataset available at https://***/view/pokechamp-llm. © 2025, CC BY.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Enhanced Suicidal Ideation Detection via Text Classification and Ensemble Methods

Enhanced Suicidal Ideation Detection via Text Classification...

引用

International Conference on Electrical, computer and Communication engineering (ECCE)

作者： Meherunnesa Hossain Ibnath Miftahul Sheikh Misbahul Sheikh Jannatul Ferdous Shormy Department of Computer Science and Engineering Ahsanullah University of Science and Technology Dhaka Bangladesh

ISBN: (数字)9798350357509

ISBN: (纸本)9798350357516

Suicidal ideation detection in textual data has emerged as a critical area of research, with implications for mental health support and intervention. This study presents a novel approach that integrates ensemble modeling and keyword-based categorization to improve the detection and analysis of suicidal tendencies in text. Using a labeled dataset with balanced class distribution (10,009 suicide and 9,988 non-suicide instances), we performed comprehensive preprocessing, including the removal of URLs, mentions, hashtags, emojis, stopwords, and applying lemmatization and tokenization. Features were extracted using CountVectorizer, and multiple machine learning models, including K-Nearest Neighbors (KNN), SVM, Naive Bayes, Decision Tree, Logistic Regression, and LSTM, were evaluated. An ensemble model combining KNN, Naive Bayes, and Decision Tree achieved the highest binary classification accuracy of 94.87%, demonstrating its superiority. Our novelty lies in the creation of a new dataset by leveraging the ensemble model to identify suicide-related text and categorizing entries into explicit, implicit, help-seeking, and emotional distress categories using domain-specific keywords. Multi-class classification on this enriched dataset was performed using advanced algorithms, including XGBoost, CatBoost, and LightGBM, with LightGBM achieving the highest 96.32% accuracy. This innovative integration of ensemble modeling and keyword-based categorization highlights the potential for enhancing suicidal ideation detection and mental health monitoring systems.

关键词： Accuracy Text categorization Mental health Nearest neighbor methods Feature extraction Tokenization Bayes methods Ensemble learning Decision trees Long short term memory

来源：评论

学校读者我要写书评

暂无评论

AI Integrated Holistic Resume Analysis

AI Integrated Holistic Resume Analysis

引用

International Conference on ent Computing and Control Systems (ICICCS)

作者： Potu Ram Prasad Mutyala Uday Kiran V Subapriya Department of Computer Science and Engineering Sathyabama Institute of Science and Technology Chennai India

ISBN: (数字)9798331512088

ISBN: (纸本)9798331512095

Analytical resume is the best tool used to deeply and thoroughly evaluate any candidate resume by considering various elements like education, work experience, and skills. This approach incorporates different angles that complement one another to build the entire profile of the professional identity and prospects of the candidate. By means of the improvement of specific areas, e.g. formatting, language usage, and personal brand, holistic analysis is able to provide an employer with powerful resume content that efficiently presents a candidate's value proposition to the employer. It tells the reader how a resume not only reveals qualifications but also fits well with the desired audience. The analysis not only examines how the document is designed and if the experiences listed are relevant, but it also considers if the language used is appropriate, the whole package of which a candidate produces for recruiters. Holistic resume analysis addresses and points out possible omissions or inconsistencies that may adversely affect the profile of a candidate, thereby serving as a catalyst to his or her improvement. The holistic resume analysis mainly serves as a guide for job seekers to better their applications and at the same time it is a hiring manager's instruction for understanding that a candidate's skills must coincide with the organization's requirements at a given time. By bringing together the data and people. This analysis helps the hiring manager to benefit more in regard to taking right decisions that will ensure that the best firms and the right people will meet and work together. In a highly competitive job market, a comprehensive review of such details is the main means of getting ahead and thus it is the only means of successfully overcoming the pitfalls that intermingle in the promising career field.

关键词： Reviews engineering profession Navigation Resumes Organizations Market research Resonance Real-time systems Natural language processing Qualifications

来源：评论

学校读者我要写书评

暂无评论

HISIM: Analytical Performance Modeling and Design Space Exploration of 2.5D/3D Integration for AI Computing

引用

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年

作者： Wang, Zhenyu Nalla, Pragnya Sudershan Sun, Jingbo Goksoy, A. Alper Mandal, Sumit K. Seo, Jae-Sun Chhabria, Vidya A. Zhang, Jeff Chakrabarti, Chaitali Ogras, Umit Y. Cao, Yu University of Minnesota Department of Electrical and Computer Engineering MinneapolisMN55455 United States Arizona State University School of Electrical Computer and Energy Engineering TempeAZ85287 United States Cornell Tech Department of Electrical and Computer Engineering New YorkNY10044 United States University of Wisconsin-Madison Department of Electrical and Computer Engineering MadisonWI53706 United States Indian Institute of Science Department of Computer Science and Automation Bengaluru560012 India

Monolithic designs face significant fabrication cost and data movement challenges, especially when executing complex and diverse AI models. Advanced 2.5D/3D packaging promises high bandwidth and connection density to overcome these challenges, yet it also introduces new electro-thermal constraints. This paper develops a suite of analytical performance models to enable efficient benchmarking of a 2.5D/3D heterogeneous system for energy-efficient AI computing. These models encompass various performance metrics related to computing units, network-on-chip, and network-on-package. The results are summarized into a new tool, HISIM, which is 104-106× faster than state-of-the-art AI benchmark tools. Furthermore, HISIM integrates rapid thermal simulation for the 2.5D/3D system, helping shed light on both the potential and limitations of 2.5D/3D heterogeneous integration on representative AI algorithms. The code of HISIM is available at https://***/mec-UMN/HISIM. © 1982-2012 IEEE.

关键词： Space research

来源：评论

学校读者我要写书评

暂无评论

Using Machine Learning to Predict Internship Effectiveness in Employability

Using Machine Learning to Predict Internship Effectiveness i...

引用

International Conference on Electrical, computer and Communication engineering (ECCE)

作者： Md. Sakibul Hassan Omi Tanvirul Islam Ayon Mazumder M Saif Sartaz Arpit Christian Md. Hasan Imam Bijoy Department of Computer Science and Engineering Daffodil International University Dhaka Bangladesh Department of Computer Science and Engineering Daffodil International Univarsity Dhaka Bangladesh Department of Information Technology Towson University Towson USA Department of Electrical and Computer Engineering Florida Atlantic University Florida USA Department of Computer & Information Science Gannon University Erie USA

ISBN: (数字)9798350357509

ISBN: (纸本)9798350357516

The main purpose of this research shall be to assess the suitability of applying internships when estimating students' employability through machine learning techniques. One analysis was done using a new database containing 1000 records with the variables of General Point Average (GPA), number of internships, supervisor ratings, organization type, employment offer status and demographic data. The data collection took place among structured undergraduate students who participated in business internship programs as their course requirements. In this study, eight machine learning methods were evaluated, including Random Forest (RF), Gradient Boosting, AdaBoost, Multi-Layer Perceptron (MLP), Gaussian Naive Bayes (GNB), XGBoost, k-Nearest Neighbors (KNN), and LightGBM. The alumni employability outcomes were predicted using these models and assessed based on four measures: accuracy, precision, recall, and F1-score. Among these models, Gaussian Naive Bayes demonstrated the highest performance, with an accuracy of 65% and an F1-score of 65.02%. AdaBoost also performed relatively well, achieving an accuracy of 59% and an F1-score of 58.89%. In contrast, the other models exhibited varying levels of performance. Random Forest achieved an accuracy of 52%with an F1-score of 52.04%, while XGBoost performed the worst, with an accuracy and F1-score of 49%. These results underscore the need to consider some internship characteristics, including task content and relevance, ratings for intern performance, and subsequent employment opportunities as critical dimensions in employability. While the obtained accuracy values suggest directions for enhancing the use of machine learning models to estimate internship effectiveness and determine essential factors influencing the graduate employability, this research confirms the continuing relevance of the field. The findings suggest the need for further research trying to improve the accuracy of the existing predictive models for practical use

关键词： Radio frequency Accuracy Databases Employment Gaussian processes Organizations Predictive models Nearest neighbor methods Bayes methods Random forests

来源：评论

学校读者我要写书评

暂无评论

deCIFer: Crystal Structure Prediction from Powder Diffraction Data using Autoregressive Language Models

arXiv

引用

arXiv 2025年

作者： Johansen, Frederik Lizak Friis-Jensen, Ulrik Dam, Erik Bjørnager Jensen, Kirsten Marie Ørnsbjerg Mercado, Rocío Selvan, Raghavendra Department of Computer Science University of Copenhagen Denmark Department of Chemistry University of Copenhagen Denmark Department of Computer Science & Engineering Chalmers University of Technology Sweden

Novel materials drive progress across applications from energy storage to electronics. Automated characterization of material structures with machine learning methods offers a promising strategy for accelerating this key step in material design. In this work, we introduce an autoregressive language model that performs crystal structure prediction (CSP) from powder diffraction data. The presented model, deCIFer, generates crystal structures in the widely used Crystallographic Information File (CIF) format and can be conditioned on powder X-ray diffraction (PXRD) data. Unlike earlier works that primarily rely on high-level descriptors like composition, deCIFer performs CSP from diffraction data. We train deCIFer on nearly 2.3M unique crystal structures and validate on diverse sets of PXRD patterns for characterizing challenging inorganic crystal systems. Qualitative and quantitative assessments using the residual weighted profile and Wasserstein distance show that deCIFer produces structures that more accurately match the target diffraction data when conditioned, compared to the unconditioned case. Notably, deCIFer can achieve a 94% match rate on unseen data. deCIFer bridges experimental diffraction data with computational CSP, lending itself as a powerful tool for crystal structure characterization and accelerating materials discovery. 1 © 2025, CC BY.

关键词： Crystal structure

来源：评论

学校读者我要写书评

暂无评论

A systematic literature review to implement large language model in higher education: issues and solutions

引用

Discover Education 2025年第1期4卷 1-25页

作者： Guizani, Sghaier Mazhar, Tehseen Shahzad, Tariq Ahmad, Wasim Bibi, Afsha Hamam, Habib College of Engineering Alfaisal University Riyadh 11533 Saudi Arabia School of Computer Science National College of Business Administration and Economics Lahore 54000 Pakistan Department of Computer Science and Information Technology School Education Department Government of Punjab Layyah 31200 Pakistan Department of Computer Engineering COMSATS University Islamabad Sahiwal Campus Sahiwal 57000 Pakistan Department of Computing School of Arts and Creative Technology University of Greater Manchester Manchester BL15AB United Kingdom Department of Computer and Software Technology University of Swat Swat 19200 Pakistan Faculty of Engineering Uni de Moncton Moncton E1A3E9 NB Canada School of Electrical Engineering University of Johannesburg Johannesburg 2006 South Africa International Institute of Technology and Management (IITG) Av. Grandes Ecoles Libreville BP 1989 Gabon Bridges for Academic Excellence - Spectrum Tunis Tunisia

Artificial intelligence-driven Chatbots, especially large language models (LLMs) like GPT-4, represent significant progress in digital education. These models excel in mimicking human-like text and transforming learning and teaching methods. This study examines the development, application, and impact of LLMs in education. It highlights their role in automating instructional tasks and promoting personalized learning experiences. Despite integration concerns and ethical debates, LLMs showcase the potential of AI to improve educational practices. Our research concludes that LLMs offer transformative opportunities for education. However, their incorporation requires careful ethical considerations, data privacy measures, and a balance between human educators and AI technologies. The findings suggest strategies for integrating LLMs into educational frameworks to enhance learning outcomes while preserving educational integrity. © The Author(s) 2025.

关键词： Artificial intelligence Education Large language models Learning systems Natural language processing systems Neural networks

来源：评论

学校读者我要写书评

暂无评论

Conflict resolution in nested transactions using ConvLTSM and Crow search algorithm in distributed optical data center databases with fiber delay lines and load balancing

引用

Journal of Optical Communications 2025年

作者： Meenu Shanker, Udai Department of Computer Science and Engineering Madan Mohan Malaviya University of Technology Gorakhpur India

This paper addresses the issue of conflict resolution in nested transactions within distributed databases of optical data centers, where efficient data processing and management are crucial. We propose a novel approach using conflict resolution in nested transactions via ConvLTSM (Convolutional long short-term memory) along with the Crow Search algorithm (CSA) for optimizing database performance. A key component of our approach is the integration of fiber delay lines (FDLs), which play a critical role in enhancing system efficiency by minimizing overall delays and ensuring optimal resource utilization. The use of FDLs helps in controlling the timing of data transmission, ensuring that transactions are processed in a timely manner, and reducing bottlenecks within the system. The incorporation of FDLs along with load balancing techniques significantly improves the throughput of nested transactions, reduces latency, and ensures consistency across the distributed system. By optimizing the timing and distribution of transaction data, FDLs facilitate better resource management, leading to more efficient and reliable database operations. Our experimental results, with a load of 0.6 and a buffering of 10, show that the transaction loss probability is 7.27 × 10-5 and the average delay is 1.54 slots. These results demonstrate the effectiveness of the proposed approach in mitigating conflicts, improving data integrity, and optimizing the performance of optical data centers. © 2025 Walter de Gruyter GmbH, Berlin/Boston.

关键词： Resource allocation

来源：评论

学校读者我要写书评

暂无评论

Generalized Few-shot 3D Point Cloud Segmentation with Vision-Language Model

arXiv

引用

arXiv 2025年

作者： An, Zhaochong Sun, Guolei Liu, Yun Li, Runjia Han, Junlin Konukoglu, Ender Belongie, Serge Department of Computer Science University of Copenhagen Denmark Computer Vision Laboratory ETH Zurich Switzerland College of Computer Science Nankai University China Department of Engineering Science University of Oxford United Kingdom

Generalized few-shot 3D point cloud segmentation (GFS-PCS) adapts models to new classes with few support samples while retaining base class segmentation. Existing GFS-PCS methods enhance prototypes via interacting with support or query features but remain limited by sparse knowledge from few-shot samples. Meanwhile, 3D vision-language models (3D VLMs), generalizing across open-world novel classes, contain rich but noisy novel class knowledge. In this work, we introduce a GFS-PCS framework that synergizes dense but noisy pseudo-labels from 3D VLMs with precise yet sparse few-shot samples to maximize the strengths of both, named GFS-VL. Specifically, we present a prototype-guided pseudo-label selection to filter low-quality regions, followed by an adaptive infilling strategy that combines knowledge from pseudo-label contexts and few-shot samples to adaptively label the filtered, unlabeled areas. Additionally, we design a novel-base mix strategy to embed few-shot samples into training scenes, preserving essential context for improved novel class learning. Moreover, recognizing the limited diversity in current GFS-PCS benchmarks, we introduce two challenging benchmarks with diverse novel classes for comprehensive generalization evaluation. Experiments validate the effectiveness of our framework across models and datasets. Our approach and benchmarks provide a solid foundation for advancing GFS-PCS in the real world. The code is at here. © 2025, CC BY-NC-SA.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

Advanced Fraud Detection in Smart Grids Using Deep Neural Networks

Advanced Fraud Detection in Smart Grids Using Deep Neural Ne...

引用

Advances in Renewable Energy & Electric Vehicles (AREEV), International Conference on

作者： Pallapothu Sasank Teja Pane Pavan Kumar Sankari M Department of Computer Science and Engineering Sathyabama Institute of Science and Technology Chennai India

ISBN: (数字)9798331529574

ISBN: (纸本)9798331529581

Fraud detection in smart grids is critical to ensure reliable and efficient energy distribution, preventing significant financial losses and maintaining grid stability. This project presents an advanced fraud detection system utilizing deep neural networks (DNN) to automatically identify and prevent fraudulent activities such as energy theft, meter tampering, and unauthorized energy consumption. The DNN model processes vast amounts of data generated by smart meters and other grid sensors, learning complex patterns and anomalies that indicate potential fraud. By integrating feature extraction techniques and real-time monitoring, the system is designed for scalability and accuracy, capable of handling high-dimensional data from modern energy grids. Additionally, this solution improves the grid's resilience and operational efficiency by enabling proactive fraud detection and reducing false positives, helping utility companies safeguard their assets.

关键词： Accuracy Scalability Artificial neural networks Feature extraction Real-time systems Smart meters Fraud Smart grids Security Reliability

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：