检索结果-内蒙古大学图书馆

DemoCraft: Using In-Context Learning to Improve Code Generation in Large Language Models 3

DemoCraft: Using In-Context Learning to Improve Code Generat...

3rd International Conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics, IITCEE 2025

作者： Joshua, Kapu Nirmal Sreejith, Mihit Indian Institute of Technology Department of Electrical Engineering Kanpur India Indian Institute of Technology Department of Computer Science and Engineering Guwahati India

ISBN: (纸本)9798331515911

Producing executable code from natural-language directives via Large Language Models (LLMs) involves obstacles like semantic uncertainty and the requirement for task-focused context interpretation. To resolve these difficulties, we introduce DemoCraft, a framework that refines code generation through in-context learning and demonstration selection, in tandem with latent concept learning. Latent concept learning appends extra concept tokens, which act as trainable embeddings encapsulating task-oriented insights. We evaluate our approach on two promi- nent datasets: MBPP and Humaneval. Our experimental findings reveal that this method attains roughly a 2x boost in the pass@k metric compared to baseline models. Moreover, we propose two new evaluation measurements: correctness@k and similarity@k. Empirical analyses indicate that our system achieves nearly a 3x advancement in these measures as well. © 2025 IEEE.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

KGFakeNet: A Knowledge Graph-Enhanced Model for Fake News Detection 31

KGFakeNet: A Knowledge Graph-Enhanced Model for Fake News De...

引用

2025 Workshop on Generative AI and Knowledge Graphs, GenAIK 2025, co-located with the 31st International Conference on Computational Linguistics, COLING 2025

作者： Kumar, Anuj Kumar, Pardeep Yadav, Abhishek Ahlawat, Satyadev Prasad, Yamuna Department of Computer Science & Engineering Indian Institute of Technology Jammu India Department of Electrical Engineering Indian Institute of Technology Jammu India

ISBN: (纸本)9798891762121

The proliferation of fake news on social media has intensified the spread of misinformation, promoting societal biases, hate, and violence. While recent advancements in Generative AI (GenAI), particularly large language models (LLMs), have shown promise, these models often need more structured representation for accurate verification, as they rely on pre-trained data patterns without access to real-time or validated information. This study presents a framework that utilizes Open information Extractor 6 (OpenIE6) to extract triplet relationships (subject-predicate-object) from statements and justifications to compute the cosine similarity between the Knowledge Graphs (KGs) of the statements and their supporting justification to precisely measure the relevance and alignment between them. This similarity feature is integrated with an attention mechanism over GenAI-generated embeddings to enhance the model’s ability to capture semantic features accurately. In addition, a Multi-Layer Perceptron (MLP) classifier is employed to integrate all features, resulting in a 4% improvement in accuracy and a 5% increase in F1-score over state-of-the-art LLM-based approaches. ©2025 International Committee on Computational Linguistics (ICCL)

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

ResNet-CPDS: Colonoscopy Polyp Detection and Segmentation Using Modified ResNet101V2 1

引用

10th International Conference on Mathematics and Computing, ICMC 2024

作者： Islam, S. K. Hafizul Pradhan, Jitesh Vashistha, Purnendu Singh, Aman P. Kishore, Aman Department of Computer Science and Engineering Indian Institute of Information Technology West Bengal Kalyani741235 India Department of Computer Science and Engineering National Institute of Technology Jharkhand Jamshedpur831014 India

ISBN: (数字)9789819720699

ISBN: (纸本)9789819720682

Colorectal cancer (CRC) is a global public health concern, and early detection through screening reduces mortality rates. It is one of the common types of cancer with a high mortality rate. Traditionally, colonoscopy is used to detect CRC, which is inefficient. Therefore, an automated Colonoscopy Polyp Detection and Segmentation (CPDS) system can significantly increase the efficiency of colonoscopy. We propose an automated model, ResNet-CPDS, using the modified ResNet101V2 model. We evaluate the performance of ResNet-CPDS and other CPDS models, and compare their accuracy. We also demonstrate that the ResNet-CPDS model outperforms other models for the CVC-ClinicDB dataset. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Efficient Real-Time Human Activity Recognition with IoT Sensor Data Using Conditionally Parametrized Convolutions 4

Efficient Real-Time Human Activity Recognition with IoT Sens...

引用

4th IEEE International Conference on ICT in Business Industry and Government, ICTBIG 2024

作者： Lovanshi, Mayank Tiwari, Vivek Singh, Anurag Ingle, Rajesh International Institute of Information Technology Department of Computer Science Engineering Naya Raipur India ABV-Indian Institute of Information Technology Department of Computer Science Engineering Gwalior India International Institute of Information Technology Department of Electronics Communication Engineering Naya Raipur India Symbiosis Skills of Professional University Department of Computer Science Engineering Pune India

ISBN: (纸本)9798331518981

Human Activity Recognition (HAR) is a trading area in computer vision and deep learning. However, boosting the performance of deep learning models often necessitates increasing their size or capacity, which raises computational demands. The high operational costs present challenges for real-time HAR on sensor-based devices. Shallow learning approaches, while computationally efficient, generally fail to deliver comparable performance. Consequently, there is a critical need for deep learning methods that can effectively balance accuracy and computational cost, a topic that remains largely underexplored. This paper introduces a computationally efficient CNN that utilizes conditionally parametrized convolution, designed specifically for real HAR on sensor-based devices. The model is tested on the OPPORTUNITY dataset, achieving existing accuracy without increasing computational load. Through extensive ablation studies, we show that this high-capacity network surpasses baseline models while maintaining a similar level of computational efficiency. The proposed method can directly replace existing deep HAR frameworks and be easily implemented on sensor-based platforms for real-time applications. © 2024 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Explainable Graph Neural Networks to Identify Potential Atoms and Chemical Bonds of Drug Candidates from Drug Target Interactions 12th

Explainable Graph Neural Networks to Identify Potential Ato...

引用

12th International Conference on Big Data and Artificial Intelligence, BDA 2024

作者： Mohanty, Satansu Dasari, Chandra Mohan Computer Science and Engineering Group Indian Institute of Information Technology Sri City Andhra Pradesh Sri City India

ISBN: (纸本)9783031818202

Drug-Target Interaction (DTI) involves the observation and recognition of interactions that occur between chemical molecules and target proteins in the human body. However, lab experimentation for DTI can be time-consuming and repetitive. Approaching these experiments from a computational perspective can save time and increase the scope of research. Graph Neural Networks (GNNs) have emerged as a leading approach in the field of DTI predictions, demonstrating state-of-the-art performance, but there is still scope for improvement in these models with new features. Thus, we propose a novel GNN-based model, GNN-DTI, for DTI predictions. The model takes the chemical molecules (as molecular graphs) and target proteins (as embedding vectors) and predicts a binding affinity score. Due to the complexity of the model, the inner workings of the models are unexplained, and it is challenging to understand the reasoning behind their outputs. Therefore, we also propose a GNN-DTI Explainer model to generate suitable explanations for drug candidates and identify the most important nodes and edges. The proposed model effectively visualizes these explanations for each drug candidate. This allows us to identify the key atoms and their bonds involved during the binding process. It provides further context for the behavior of certain drugs and their interactions with the human body, which can be used to advance drug development and research. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

Deep learning intelligence for influencer-based topological classification for online social networks

引用

International Journal of Innovative Computing and Applications 2023年第4期14卷 211-219页

作者： Jain, Somya Sinha, Adwitiya Department of Computer Science and Engineering and Information Technology Jaypee Institute of Information Technology 62 Uttar Pradesh Noida India

Social network analysis provides quantifiable methods and topological metrics to examine the networked structure for several interdisciplinary applications. In our research, a social network of GitHub community is constructed that forms a dense network of 37,700 developers with 289,003 associations amongst them. The research involves finding the central developers in the GitHub network using graph analytics and benchmark centrality metrics, including degree, betweenness, closeness, PageRank and eigenvector;which is based upon network structural information. Our research methodology provides a breakthrough towards predicting the classification of GitHub users using artificial intelligence-based learning model trained with derived topological network centrality metrics. The proposed approach performs feature extraction for the developers by computing centrality score of each user followed by building correlation matrix using centrality parameters based on network topology. Further, the derived topological centrality scores were used as input features to train and build artificial intelligence-based models for classification. Our experimentation is better performance of artificial neural network over autoencoders, logistic regression and hyper-parameter tuned support vector machine. Certain intermediate outcomes include correlation, principal component analysis, loss monitoring, etc. The performance evaluation was performed in terms of macro and weighted F1-score, recall, precision, and accuracy. Copyright © 2023 Inderscience Enterprises Ltd.

关键词： Social networking (online)

来源：评论

学校读者我要写书评

暂无评论

On Applying Preprocessing Techniques on Gallbladder Ultrasound Images 8

On Applying Preprocessing Techniques on Gallbladder Ultrasou...

引用

8th IEEE International Conference on information and Communication technology, CICT 2024

作者： Suji, R. Jenkin Godfrey, W. Wilfred Madhav Institute of Technology and Science Department of Electronics Engineering Gwalior India ABV-Indian Institute of Information Technology and Management Department of Computer Science and Engineering Gwalior India

ISBN: (纸本)9798331531539

Ultrasound imaging is a common and non-invasive method for diagnosing gallbladder diseases, including gallbladder cancer (GBC). However, the inherent challenges of ultrasound images-such as noise, low contrast, and variable image quality-necessitate the use of preprocessing techniques to enhance image clarity and improve diagnostic accuracy. In this study, we apply a series of preprocessing methods, including median filtering, Gaussian filtering, bilateral filtering, adaptive histogram equalization, and wavelet transforms, to gallbladder ultrasound images. We evaluate the effectiveness of these techniques using a variety of quality metrics such as sharpness, signal-to-noise ratio (SNR), entropy, structural similarity index (SSIM), and perceptual image quality evaluator (PIQE). A detailed analysis of the preprocessing methods was conducted, where the performance of each technique was systematically compared. Additionally, we highlight the statistical variations in these metrics across different preprocessing approaches to identify the most suitable methods for enhancing GBC detection. The results show that specific preprocessing methods improve image quality and feature clarity, aiding accurate gallbladder disease detection models. This work is an initial groundwork for enhancing the diagnostic value of ultrasound images for GBC. It suggests future directions for integrating these methods into machine-learning pipelines for automated disease detection. © 2024 IEEE.

关键词： Wavelet transforms

来源：评论

学校读者我要写书评

暂无评论

WSO-KELM: War Strategy Optimization-Based Kernel Extreme Learning Machine for Automatic Software Fault Prediction Model

引用

Journal of The Institution of Engineers (India): Series B 2025年第1期106卷 145-163页

作者： Elci, J. Brundha Nandagopalan, S. Department of Computer Science and Engineering Vemana Institute of Technology Visvesvaraya Technological University Karnataka Belagavi India Department of Information Science and Engineering Vemana Institute of Technology Visvesvaraya Technological University Karnataka Belagavi India

The software development projects’ testing part is usually expensive and complex, but it is essential to gauge the effectiveness of the developed software. Software Fault Prediction (SFP) primarily serves to detect faults within software components. The scale of software projects is getting larger, making it impossible and time-consuming to analyze the flaws manually. Artificial Intelligence (AI) techniques are mainly incorporated to solve this problem. The massive data retrieved from the software repositories helps in identifying the features that are highly correlated with fault prediction and the features that are not. This paper introduces a Kernel Extreme Learning Machine (KELM) for SFP. To optimize the hyperparameters of the classifier and also enhance the classification accuracy, the War Strategy Optimization (WSO) algorithm is employed here. The experimental evaluations are conducted using different baseline actual-world datasets of software programs such as Java-coded open-source and PROMISE datasets. The experimental outcomes are conducted using different performance metrics that include sensitivity, Accuracy, and F1-score with values of 0.96%, 0.97%, and 0.96%, respectively. The results show that KELM is beneficial, and the optimization provided by the WSO algorithm also helps us obtain higher classification accuracy for the SFP problem and improve the fault detection process. © The Institution of Engineers (India) 2024.

关键词： Software reliability

来源：评论

学校读者我要写书评

暂无评论

OCRBench: on the hidden mystery of OCR in large multimodal models

引用

science China(information sciences) 2024年第12期67卷 23-35页

作者： Yuliang LIU Zhang LI Mingxin HUANG Biao YANG Wenwen YU Chunyuan LI Xu-Cheng YIN Cheng-Lin LIU Lianwen JIN Xiang BAI School of Artificial Intelligence and Automation Huazhong University of Science and Technology School of Electronic and Information Engineering South China University of Technology Microsoft Research School of Computer & Communication Engineering University of Science and Technology Beijing Institute of Automation Chinese Academy of Sciences School of Software Engineering Huazhong University of Science and Technology

Large models have recently played a dominant role in natural language processing and multimodal vision-language learning. However, their effectiveness in text-related visual tasks remains relatively unexplored. In this paper, we conducted a comprehensive evaluation of large multimodal models, such as GPT4V and Gemini, in various text-related visual tasks including text recognition, scene text-centric visual question answering(VQA), document-oriented VQA, key information extraction(KIE), and handwritten mathematical expression recognition(HMER). To facilitate the assessment of optical character recognition(OCR) capabilities in large multimodal models, we propose OCRBench, a comprehensive evaluation benchmark. OCRBench contains 29 datasets, making it the most comprehensive OCR evaluation benchmark available. Furthermore, our study reveals both the strengths and weaknesses of these models, particularly in handling multilingual text, handwritten text, non-semantic text, and mathematical expression *** importantly, the baseline results presented in this study could provide a foundational framework for the conception and assessment of innovative strategies targeted at enhancing zero-shot multimodal *** evaluation pipeline and benchmark are available at https://***/Yuliang-Liu/Multimodal OCR.

关键词： large multimodal model OCR text recognition scene text-centric VQA document-oriented VQA key information extraction handwritten mathematical expression recognition

来源：评论

学校读者我要写书评

暂无评论

Secure probe-based latency measurement using NTP and PTP for optimized gateway placement in LoRaWAN

引用

International Journal of information technology (Singapore) 2025年 1-10页

作者： Bhatnagar, Kumar Vaibhav Kushwah, Rashmi Department of Computer Science & Engineering and Information Technology Jaypee Institute of Information Technology Uttar Pradesh Noida 201304 India

The paper presents a novel framework for optimizing LoRaWAN gateway placement to enhance network performance and reliability. By integrating Network Time Protocol (NTP) for global synchronization and Precision Time Protocol (PTP) for local precision, the approach ensures accurate and secure latency measurements. A robust hash algorithm is employed to safeguard the integrity of latency data. The Performance Efficiency Index (PEI) is introduced as a comprehensive metric to evaluate key performance indicators such as latency, throughput, energy efficiency, and packet delivery ratio (PDR). The model incorporates factors like geographical distribution, signal strength, and gateway density to achieve optimal placement. Simulation results demonstrate the framework’s effectiveness in improving network latency, coverage, energy efficiency, and security, providing actionable insights for LoRaWAN-based IoT network design. © Bharati Vidyapeeth's institute of computer Applications and Management 2025.

关键词： Gateway placement LoRaWAN Network performance NTP PTP Timestamp

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：