检索结果-内蒙古大学图书馆

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Yao, Jiashu Huang, Heyan Liu, Zeming Wen, Haoyu Su, Wei Qian, Boao Guo, Yuhang School of Computer Science and Technology Beijing Institute of Technology China School of Computer Science and Engineering Beihang University China

ISBN: (纸本)157735897X

Following formatting instructions to generate well-structured content is a fundamental yet often unmet capability for large language models (LLMs). To study this capability, which we refer to as format faithfulness, we present FORMATBENCH, a comprehensive format-related benchmark. Compared to previous format-related benchmarks, FORMATBENCH involves a greater variety of tasks in terms of application scenes (traditional NLP tasks, creative works, autonomous agency tasks), human-LLM interaction styles (single-turn instruction, multi-turn chat), and format types (inclusion, wrapping, length, coding). Moreover, each task in FORMATBENCH is attached with a format checker program. Extensive experiments on the benchmark reveal that state-of-the-art open- and closed-source LLMs still suffer from severe deficiency in format faithfulness. By virtue of the decidable nature of formats, we propose to Reinforce Format Faithfulness (REFF) to help LLMs generate formatted output as instructed without compromising general quality. Without any annotated data, REFF can substantially improve the format faithfulness rate (e.g., from 21.6% in original LLaMA3 to 95.0% on caption segmentation task), while keep the general quality comparable (e.g., from 47.3 to 46.4 in F1 scores). Combined with labeled training data, REFF can simultaneously improve both format faithfulness (e.g., from 21.6% in original LLaMA3 to 75.5%) and general quality (e.g., from 47.3 to 61.6 in F1 scores). We further offer an interpretability analysis to explain how REFF improves both format faithfulness and general quality. © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Classification and Segmentation of Gastroenterology Images Using Saliency Mapping in Convolutional Neural Network 21

Classification and Segmentation of Gastroenterology Images U...

引用

21st International Bhurban Conference on Applied sciences and Technology, IBCAST 2024

作者： Khan, Yusra Shaukat, Arslan Riaz, Farhan Department of Computer and Software Engineering Islamabad Pakistan School of Computer Science University of Lincoln Lincoln United Kingdom

ISBN: (纸本)9798331516680

This research work has been done towards the treatment of Gastrointestinal (GI) Cancer by experimenting as to how a computer aided design can help oncologists classify and segment GI Cancer using Gastroenterology (GE) Images. The concept of Attention Module by the name of Convolution Block Attention Module (CBAM) using Convolutional Neural Network (CNN) layers has been utilized to achieve saliency mapping in these images. The different layers of the CNN are tweaked and the hyper parameters are consistently checked and changed to achieve maximum accuracy. The CBAM Algorithm is divided into Channel Attention (CA) and Spatial Attention (SA) module. The Channel Attention has been implemented in the HSV Color Space for finding the best color space for segmentation purposes of the Chromo endoscopy medical images. Classification, followed by segmentation, is again implemented using CNN based model for normal and abnormal cancer images. For classification, 92% test accuracy has been achieved, while 70% Dice Similarity Coefficient has been obtained for segmentation. The results achieved are competitive as compared to the previous reported results on the GE images dataset. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

An Overview of LoRa Localization Technologies

引用

computers, Materials & Continua 2025年第2期82卷 1645-1680页

作者： Huajiang Ruan Panjun Sun Yuanyuan Dong Hamid Tahaei Zhaoxi Fang Institute of Artificial Intelligence Shaoxing UniversityShaoxing312000China Department of Computer Science and Engineering School of Mechanical and Electrical EngineeringShaoxing UniversityShaoxing312000China

Traditional Global Positioning System(GPS)technology,with its high power consumption and limited perfor-mance in obstructed environments,is unsuitable for many Internet of Things(IoT)*** paper explores LoRa as an alternative localization technology,leveraging its low power consumption,robust indoor penetration,and extensive coverage area,which render it highly suitable for diverse IoT *** comprehensively review several LoRa-based localization techniques,including time of arrival(ToA),time difference of arrival(TDoA),round trip time(RTT),received signal strength indicator(RSSI),and fingerprinting *** this review,we evaluate the strengths and limitations of each technique and investigate hybrid models to potentially improve positioning *** studies in smart cities,agriculture,and logistics exemplify the versatility of LoRa for indoor and outdoor *** findings demonstrate that LoRa technology not only overcomes the limitations of GPS regarding power consumption and coverage but also enhances the scalability and efficiency of IoT deployments in complex environments.

关键词： LoRa localization IoT TDoA RSSI application hybrid models

来源：评论

学校读者我要写书评

暂无评论

Hybrid approach for the detection of epileptic seizure using electroencephalography input

引用

International Journal of Information Technology (Singapore) 2024年第1期16卷 569-575页

作者： Basha, Niha Kamal Surendiran, B. Benzikar, Amutha Joyal, S. School of Computer Science and Engineering Vellore Institute of Technology (VIT) Vellore India Department of Computer Science and Engineering National Institute of Technology (NIT) Puducherry Karaikal India School of Computer Science and Engineering Vellore Institute of Technology (VIT) Chennai India Saveetha Engineering College Chennai India

In the early days, it was difficult to study bio-electric signals, but now a days these problems have been solved by many hardware devices which are available at low cost. Even then there is a need for technical improvements to process bio-electric signals. Enormous effort has been taken for the detection of epileptic seizure from EEG signal. Considerable evidence available on Bonn university EEG dataset for epileptic seizure detection. However, deep learning algorithms have not applied often on Bonn university data like other machine learning algorithms for the detection of epileptic seizure due to the less availability of data. This work adopted machine learning and deep learning models for the detection of epileptic seizure. The data consists of a hundred subjects with sampling rate as 173.61 Hz of ‘S’ (ictal) and ‘Z’ (normal) dataset. The results show how the choice of healthy and ictic subjects through high order functional math decision like variance (STD), Power, Skewness, and Kurtosis values are analyzed to extract epileptic features. The performance of classifiers has been evaluated based on the evaluation metrics within which CGRU-SVM outperforms all other models with 97.54% accuracy. The features are extracted based on the statistical measures to detect epileptic seizure. Further, the proposed work provides more evidence on the epileptic seizure features, and it shows the new possibility for using deep learning models for epileptic seizure detection. In future, an analysis must be performed with dynamic sets of input and different analysis techniques. © 2023, The Author(s), under exclusive licence to Bharati Vidyapeeth's institute of computer Applications and Management.

关键词： Confusion matrix Deep learning model Electroencephalography Epilepsy Ictal subject Rhythmic frequency Statistical feature

来源：评论

学校读者我要写书评

暂无评论

XAI based Photovoltaic Energy Management Framework for Smart Cities

引用

IEEE Access 2025年 13卷 98349-98361页

作者： Alagarsamy, Malini Rajasekaran, Umamaheswari Ganesan, Sriram Palanivel, Ramyavarshini Vellore Institute of Technology School of Computer Science and Engineering Tamil Nadu Chennai India Thiagarajar College of Engineering Tamil Nadu Madurai India

The demand for explainable energy forecasting models has been emphasized in recent studies to enhance decision-making transparency. The efficient conversion of solar energy is stochastic and depends on environmental conditions. To address this, we propose a model-chain framework named XPEMM (XAI-based Photovoltaic Energy Management Model chain) for Solar Energy Production and Management in Smart Cities, ensuring explainability using LIME. Our model chain consists of three stages: Long-term forecasting of Global Horizontal Irradiance (GHI) as the first stage, Regression Modeling of GHI with power as the second stage, and Model-Agnostic explanation generation as the third stage. A recursive forecasting strategy was employed, leveraging Recurrent Neural Networks (RNN) to reduce error accumulation over a 5-year forecast horizon. The output of the proposed model chain assists grid cell operators in real-time monitoring, resulting in optimized performance by predicting the causes of high and low performance in advance, allowing for necessary adjustments. It also aids in site selection for high-budget PV-plant installations. We used two different datasets for training and benchmarking the Recursive Multistep GHI Forecasting Model (RMGFM). The proposed recursive forecasting strategy has been evaluated using RMSE and MAE error metrics which are 91.40 and 47.32 respectively, which are 2.9% 29.39%, and 7.9% less than the previously established approach using LSTM, PCR, and SVR based on RMSE measure and 15.9%, 49.9%, 34.8% less based on MAE metric. This study concludes that a recursive approach is best for long-term forecasting of GHI in regions with cyclic climatic patterns based on RMSE and MAE values. The interpretation of the LIME output establishes the fitness of model agnosticity in the Photovoltaic model chain. © 2013 IEEE.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Graph-Based Multitask Transfer Learning for Fault Detection and Diagnosis of Few-Shot Analog Circuits

引用

IEEE Internet of Things Journal 2025年第12期12卷 21264-21279页

作者： Gao, Zhongyu Yan, Aibin Huang, Zhengfeng Cui, Jie Roh, Byeong-Hee Liu, Guangzhu Girard, Patrick Wen, Xiaoqing Anhui University School of Computer Science and Technology Hefei230601 China Hefei University of Technology School of Microelectronics Hefei230009 China Ajou University Department of Software and Computer Engineering Suwon16499 Korea Republic of University of Montpellier CNRS LIRMM Montpellier34095 France Kyushu Institute of Technology Department of Computer Science and Networks Fukuoka820-8502 Japan

Building an interpretable fault detection and diagnostic model based on few-shot circuit samples and prior information about circuit structures is of significant importance. To fill these gaps, we propose a graph-based multitask transfer learning (TL) method for fault detection and diagnosis of circuits under few-shot conditions. First, in order to model the interconnections of nodes in a circuit, the sample data is organized into a graph structure, and a semi-supervised graph-based structural feature fusion method is proposed. The proposed method can accept graph-structured data and process the data using feature fusion methods. Second, to improve the model performance under few-shot conditions, two TL mechanisms are proposed for the topological structure characteristics of analog circuits as well as circuit signal characteristics. Finally, through a parameter-shared strategy, we propose a task transfer-based fault diagnosis approach. Experimental results on three different circuits show that the proposed method has the best diagnostic accuracy compared to typical detection and diagnosis schemes. © 2014 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Generation for Zero-Shot Natural Language Video Localization 39

Boundary-Aware Temporal Dynamic Pseudo-Supervision Pairs Gen...

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Deng, Xiongwen Tang, Haoyu Jiang, Han Zheng, Qinghai Zhu, Jihua School of Software Engineering Xi’an Jiaotong University China School of Software Shandong University China College of Computer and Data Science Fuzhou University China

ISBN: (纸本)157735897X

Zero-shot Natural Language Video Localization (NLVL) aims to automatically generate moments and corresponding pseudo queries from raw videos for the training of the localization model without any manual annotations. Existing approaches typically produce pseudo queries as simple words, which overlook the complexity of queries in real-world scenarios. Considering the powerful text modeling capabilities of large language models (LLMs), leveraging LLMs to generate complete queries that are closer to human descriptions is a potential solution. However, directly integrating LLMs into existing approaches introduces several issues, including insensitivity, isolation, and lack of regulation, which prevent the full exploitation of LLMs to enhance zero-shot NLVL performance. To address these issues, we propose BTDP, an innovative framework for Boundary-aware Temporal Dynamic Pseudo-supervision pairs generation. Our method contains two crucial operations: 1) Boundary Segmentation that identifies both visual boundaries and semantic boundaries to generate the atomic segments and activity descriptions, tackling the issue of insensitivity. 2) Context Aggregation that employs the LLMs with a self-evaluation process to aggregate and summarize global video information for optimized pseudo moment-query pairs, tackling the issue of isolation and lack of regulation. Comprehensive experimental results on the Charades-STA and ActivityNet Captions datasets demonstrate the effectiveness of our BTDP method. © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Federated ADMM

IEEE Networking Letters

引用

IEEE Networking Letters 2025年第1期7卷 11-15页

作者： Azimi-Abarghouyi, Seyed Mohammad Bastianello, Nicola Johansson, Karl H. Fodor, Viktoria KTH Royal Institute of Technology School of Electrical Engineering and Computer Science and Digital Futures Stockholm10044 Sweden

In this letter, we depart from the widely-used gradient descent-based hierarchical federated learning (FL) algorithms to develop a novel hierarchical FL framework based on the alternating direction method of multipliers (ADMM), leveraging a network architecture consisting of a single cloud server and multiple edge servers, where each edge server is dedicated to a specific client set. Within this framework, we propose two novel FL algorithms, which both use ADMM in the top layer: one that employs ADMM in the lower layer and another that uses the conventional gradient descent-based approach. The proposed framework enhances privacy, and experiments demonstrate the superiority of the proposed algorithms compared to the conventional algorithms in terms of learning convergence and accuracy. Additionally, gradient descent on the lower layer performs well even if the number of local steps is very limited, while ADMM on both layers lead to better performance otherwise. © 2019 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

A dynamic and resource sharing virtual network mapping algorithm

引用

Digital Communications and Networks 2023年第5期9卷 1101-1112页

作者： Xiancui Xiao Xiangwei Zheng Ji Bian Cun Ji Xinchun Cui School of Information Science and Engineering Shandong Normal UniversityJinan250014China School of Information Engineering Shandong Management UniversityJinan250358China Shandong Provincial Key Laboratory for Distributed Computer Software Novel Technology Jinan250014China School of Computer Science and Technology Qufu Normal UniversityRizhao276800China

Network virtualization can effectively establish dedicated virtual networks to implement various network ***,the existing research works have some shortcomings,for example,although computing resource properties of individual nodes are considered,node storage properties and the network topology properties are usually ignored in Virtual Network(VN)modelling,which leads to the inaccurate measurement of node availability and *** addition,most static virtual network mapping methods allocate fixed resources to users during the entire life cycle,and the users’actual resource requirements vary with the workload,which results in resource allocation *** on the above analysis,in this paper,we propose a dynamic resource sharing virtual network mapping algorithm named NMA-PRS-VNE,first,we construct a new,more realistic network framework in which the properties of nodes include computing resources,storage resources and topology *** the node mapping process,three properties of the node are used to measure its mapping ***,we consider the resources of adjacent nodes and links instead of the traditional method of measuring the availability and priority of nodes by considering only the resource properties,so as to more accurately select the physical mapping nodes that meet the constraints and conditions and improve the success rate of subsequent link ***,we divide the resource requirements of Virtual Network Requests(VNRs)into basic subrequirements and variable sub-variable requirements to complete dynamic resource *** former represents monopolizing resource requirements by the VNRs,while the latter represents shared resources by many VNRs with the probability of occupying resources,where we keep a balance between resource sharing and collision among users by calculating the collision *** results show that the proposed NMAPRS-VNE can increase the average acceptance rate and network revenu

关键词： Network virtualization VNRs Network frameworks Dynamic resource allocation Resource sharing

来源：评论

学校读者我要写书评

暂无评论

RoSense: Refining LOS Signal Phase for Robust RFID Sensing via Spinning Antenna

引用

Performance Evaluation Review 2023年第4期50卷 53-55页

作者： Zhu, Yinan Duan, Chunhui Ding, Xuan Tsinghua University School of Software and BNRist Beijing China Beijing Institute of Technology School of Computer Science and Technology Beijing China

RFID sensing leveraging backscatter signal features (e.g., phase shift) from tags has gained increasing popularity in numerous applications, but also suffers from negative impacts of environmental multipaths. Past works to address it rely on extra customized devices, labor-intensive offine training or frequency channel hopping, all of which are nonubiquitous or ineffective for real-life adoption. This paper presents RoSense, a universal method to alleviate multipath reflections' impacts by spinning the reader antenna, thus enabling more robust RFID sensing. Besides, RoSense requires no RF devices or offline training, and operates in a nonintrusive manner. The key insight of RoSense is to exploit two properties of line-of-sight (LOS) signal when spinning the antenna, i.e., the linearity of phase changes and stability of received signal strength to attenuate the non-linear and non-monotonic effect of multipath signals and refine the phase shift of LOS signal. We have implemented a prototype of RoSense with COTS devices and studied two cases for evaluation: material identification and object localization. Experimental results show that RoSense can improve the material identification accuracy by 16.22% and reduce the mean localization error by 39.93%, greatly outperforming the state-of-the-art solutions. © 2023 Copyright is held by the owner/author(s).

关键词： Mobile antennas

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：