检索结果-内蒙古大学图书馆

4th international Conference on computational Methods in science and technology, ICCMST 2024

作者： Sharma, Ankita Upadhyay, Aman Gautam, Prateek Raj Centre for Advanced Studies AKTU Lucknow India School of Computer Science UPES Dehradun India

ISBN: (纸本)9781032911571

River water flow, also known as water discharge or streamflow of a river pertains to the volume of water passing through a point in a specified time frame. It is quantified in cubic meters per second or cubic feet per second. Maintaining and predicting the flow of water through a stream is essential yet a very difficult task. Increase or decrease in water levels can lead to flood or drought like conditions respectively. However, monitoring water behavior is tough due to lack of data, various hydrological factors, unmonitored terrains, and non-linear nature of water. The conventional approaches are not capable of handling non-linear, non-stationary nature of water. Therefore, this research focuses on developing an innovative hybrid artificial intelligence model. It involves training a neural network rigorously with monthly data collected for 10 years from Tikarpara gauge station over Mahanadi River for predicting the flow of water. This study produces notable results with correlation value of 0.94 between actual and predicted values. Artificial intelligent methods outperform any manual method and saves time, money, energy and chances of human errors. © 2025 the Author(s).

关键词： Rivers

来源：评论

学校读者我要写书评

暂无评论

Hydra: Scale-out FHE Accelerator Architecture for Secure Deep Learning on FPGA 31

Hydra: Scale-out FHE Accelerator Architecture for Secure Dee...

引用

31st IEEE international symposium on High Performance computer Architecture, HPCA 2025

作者： Yang, Yinghao Xu, Xicheng Zhang, Haibin Song, Jie Tang, Xin Lu, Hang Li, Xiaowei Institute of Computing Technology CAS State Key Laboratory of Processors Beijing China University of Chinese Academy of Sciences Beijing China Zhongguancun Laboratory Beijing China

ISBN: (纸本)9798331506476

Deep learning, including Convolutional Neural Network (CNN) and Large Language Model (LLM), under Fully Homomorphic Encryption (FHE) is very computationally intensive because of the burdensome computations like ciphertext convolution and matrix multiplication, non-linear layers, and bootstrapping. Existing FHE accelerators focus on the high throughput computational units, stacking parallelized clusters to maximize ciphertext inference performance. Nevertheless, this design philosophy cannot leverage the substantial parallelism at the application level and is not scalable for further performance enhancement by simply adding additional compute nodes to cope with the ever-increasing model sizes in the future. In this paper, we propose the high-performance FHE acceleration architecture in a 'scale-out' manner for secure deep learning, termed as Hydra. It supports the multi-server scaling and arbitrary computational nodes theoretically, each handling a portion of the deep learning model governed by the central scheduling mechanism on the host server. Hydra exhibits excellent scalability and delivers outstanding performance across a range of compute resource sizes. We highlight the following results: (1) up to 74 × and 160 × speedup over the SOTA single card accelerator Poseidon and FAB;(2) outperforms 8-card FAB-2 by 12 × to 21 × for FHE-based CNNs and LLMs;(3) outperforms SOTA ASIC accelerators, CraterLake and SHARP, by 8.1 × and 2.5 × for LLM OPT-6.7B, and achieves comparable or superior energy efficiency under the same chip technology. © 2025 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Meter-YOLOv8n: A Lightweight and Efficient Algorithm for Word-Wheel Water Meter Reading Recognition

引用

international Journal of Advanced computer science and Applications 2025年第4期16卷 209-221页

作者： Qiao, Shichao Yuan, Yuying Qi, Ruijie School of Computer Science and Technology Shandong University of Technology Shandong Zibo255000 China

To address the issues of low efficiency and large parameters in the current word-wheel water meter reading recognition algorithms, this paper proposes a Meter-YOLOv8n algorithm based on YOLOv8n. Firstly, the C2f component of YOLOv8n is improved by introducing an enhanced inverted residual mobile block (iRMB). It enables the model to efficiently capture global features and fully extract the key information of the water meter characters. Secondly, the Slim-Neck feature fusion structure is employed in the neck network. By replacing the original convolutional kernels with GSConv, the model's ability to express the features of small object characters is enhanced, and the number of parameters in the model is reduced. Finally, Inner-EIoU is used to optimize the bounding box loss function. This simplifies the calculation process of the loss function and improves the model's ability to locate dense bounding boxes. The experimental results show that, compared with the original model, the precision, recall, mAP@0.5, and mAP@0.5:0.95 of the improved model have increased by 1.7%, 1.2%, 3.4%, and 3.3% respectively. Meanwhile, the parameters, FLOPs, and model size have decreased by 0.56M, 2.6G, and 0.7MB respectively. The improved model can better balance the relationship between detection performance and computational complexity. It is suitable for the task of recognizing word-wheel water meter readings and has practical application value. © (2025), (science and Information Organization). All Rights Reserved.

关键词： NP-hard

来源：评论

学校读者我要写书评

暂无评论

LMSNet for Real-Time Semantic Segmentation on Urban Road Images

IAENG International Journal of Computer Science

引用

IAENG international Journal of computer science 2025年第4期52卷 1030-1037页

作者： Cao, Yi Qu, Hongbo Zhengzhou University of Science and Technology Zhengzhou450064 China

Semantic segmentation algorithms are essential for technologies in autonomous systems. The development of lightweight convolutional neural networks has facilitated the deployment of deep learning-driven semantic segmentation approaches on energy-efficient mobile devices. However, these lightweight networks often overlook feature fusion relationships, employing linear fusion strategies that limit segmentation accuracy. To tackle this challenge, this study introduces an innovative lightweight network design intended to serve as the core structure for the encoder. This method aims to decrease both the parameter count and the computational demands, thereby substantially boosting the network’s efficiency in real-time applications. Additionally, a new Multi-Feature Extraction Module (MFM) is introduced before the decoder to capture multi-scale object information from images, thereby enhancing contextual understanding and improving segmentation accuracy and robustness. Furthermore, to further boost network performance, we extract high-frequency features from images and form a new stream. These features are then fused with existing feature maps through a High-Frequency Feature Fusion (HFF) module. This strategy captures finer details, thereby enhancing segmentation precision. Extensive experiments conducted on the publicly available Cityscapes and CamVid datasets using an NVIDIA 4090 GPU demonstrate average Intersection over Union (IoU) scores of 75.8% and 70.3%, respectively. Our approach not only surpasses existing network architectures in terms of accuracy but also significantly reduces computational resource consumption. © (2025), (international Association of Engineers). 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

Enhancing Emotional Support Conversations: A Framework for Dynamic Knowledge Filtering and Persona Extraction 31

Enhancing Emotional Support Conversations: A Framework for D...

引用

31st international Conference on computational Linguistics, COLING 2025

作者： Hao, Jiawang Kong, Fang Laboratory for Natural Language Processing China Soochow University School of Computer Science and Technology Suzhou China TAL Education Group Beijing China

ISBN: (纸本)9798891761964

With the growing need for accessible emotional support, conversational agents are being used more frequently to provide empathetic and meaningful interactions. However, many existing dialogue models struggle to interpret user context accurately due to irrelevant or misclassified knowledge, limiting their effectiveness in real-world scenarios. To address this, we propose a new framework that dynamically filters relevant commonsense knowledge and extracts personalized information to improve empathetic dialogue generation. We evaluate our framework on the ESConv dataset using extensive automatic and human experiments. The results show that our approach outperforms other models in metrics, demonstrating better coherence, emotional understanding, and response relevance. © 2025 Association for computational Linguistics.

关键词： Chatbots

来源：评论

学校读者我要写书评

暂无评论

Exploring the Impact of Conditional Deep Convolutional Generative Adversarial Networks in Brain Tumor Image Classification: A Novel Approach

Exploring the Impact of Conditional Deep Convolutional Gener...

引用

2025 international Conference on Pervasive computational Technologies, ICPCT 2025

作者： Upadhyay, Deepak Upadhyay, Abhay Sharma, Kuj Bihari Dhondiyal, Shiv Ashish Venu, Nookala Graphic Era Hill University Dept. of Computer Science and Engineering Dehradun India Graphic Era Deemed to Be University Dept. of Computer Science and Engineering Dehradun India Centre for Internet of Things Madhav Institute of Technology & Science Deemed University Gwalior India

ISBN: (纸本)9798331508685

In this paper, we explore the impact of conditional Deep Convolutional Generative Adversarial Networks (cDCGANs) with brain tumor image classification and introduce a new way to improve diagnosis accuracy. Shortage of Data/Data Imbalance: In Medical Imaging datasets, it helps in surging the data scarcity and class imbalance with mimicking generation images using cDCGAN. The detailed investigation demonstrates that synthetic data effectively improves classifier performance by acting as a form of augmentation, providing additional training examples for classifiers and serving as ground truth. From the computational efficiency analysis, we see that there is also scalability in training time with respect to dataset size. Consequently, this study presents cDCGANs as an important asset with which to improve diagnostic accuracy through brain tumor categorization in clinical contexts. © 2025 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

DNNT (Deep Neural Network for Telugu): a framework for speech recognition of Telugu language with parallel computing approach

引用

international Journal of Speech technology 2025年 1-9页

作者： Reddy, S. Satheeswara Mnoj, Sheik Khadar Ahmad Rao, Karu Prasada Department of MCA Dr. Lankapalli Bullayya PG College Andhra Pradesh Visakhapatnam India Department of Computer Science and Engineering GST GITAM Deemed to be University Andhra Pradesh Visakhapatnam India

The revolutionary growth in smart devices and increase in social media led to growth of online voice-based applications and online speech communication. The rapid advancement of artificial intelligence and machine learning has significantly impacted speech processing applications, enabling seamless human-computer interaction. This paper presents a Telugu Speech Recognition and Synthesis System employing a hybrid approach integrating Hidden Markov Models (HMM) and Deep Neural Networks (DNN). The system performs Speech-to-Text-to-Speech (STTS) conversion, ensuring accurate speech recognition and synthesis. The proposed model preprocesses Telugu speech data through noise reduction, silence removal, and resampling techniques. Mel-Frequency Cepstral Coefficients (MFCCs) are extracted as feature representations, followed by HMM-based phoneme modelling and DNN-based classification for robust recognition. Experimental results have shown performance in accuracy, computational efficiency, and naturalness of generated speech. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Speech synthesis

来源：评论

学校读者我要写书评

暂无评论

A comparative study of machine learning in diabetes classification: Decision trees, logistic regression, and SVM 2

A comparative study of machine learning in diabetes classifi...

引用

2nd international Conference on Big Data, computational Intelligence, and Applications, BDCIA 2024

作者： Zhu, Xinglong Liao, Huitong Zhang, Tingting Zhang, Lujuan Li, Xueri Department of Computer Science Guangdong University of Science and Technology Dongguan China

ISBN: (纸本)9781510689053

As the application of computer science in healthcare continues to expand, machine learning techniques have become an important tool for disease diagnosis. In this study, we trained and predicted diabetes datasets by preprocessing, standardising and normalising the data, and then using three machine learning models, aiming to improve the accuracy of diabetes prediction and the value of clinical application. The experimental results show that these machine learning models exhibit a high degree of accuracy and reliability in the field of diabetes diagnosis, among which the SVM performs the best on normalised data, with the highest values of accuracy, precision, recall, F1-score, and AUC evaluation metrics, showing the good performance of the support vector machine in predicting diabetes. The decision tree model performed consistently on raw and standardised data, while the logistic regression model performed best on AUC values. This study not only verifies the effectiveness of machine learning models in diabetes prediction, but also provides clinicians with accurate diagnostic assistance, which is of significant academic and practical significance. © 2025 SPIE.

关键词： Logistic regression

来源：评论

学校读者我要写书评

暂无评论

Path Planning Method for Mobile Robots in Pedestrian-Friendly Environments Using Drone Bird's-Eye View

Path Planning Method for Mobile Robots in Pedestrian-Friendl...

引用

2025 IEEE/SICE international symposium on System Integration, SII 2025

作者： Ye, Linghao Chugo, Daisuke Muramatsu, Satoshi Yokota, Sho Hashimoto, Hiroshi Kwansei Gakuin University Graduate School of Science and Technology Hyogo Sanda6691330 Japan Tokai University School of Information Science and Technology Kanagawa Hiratsuka2591292 Japan Toyo University Faculty of Science and Engineering Saitama Kawagoe3508585 Japan Advanced Institute of Industrial Technology Master Program of Innovation for Design and Engineering Shinagawa Tokyo1400011 Japan

ISBN: (纸本)9798331531614

The objective of this paper is to develop a robot system that can transmit the video taken by the drone to the computer in real time, use YOLO to recognize the image of pedestrians, and enable the ground robot to plan the path to avoid pedestrians. External sensors, such as laser range finders (LRF) and CCD cameras, are often used on mobile robots. As we all know, these sensors are susceptible to occlusion. Therefore, even if the mobile robot avoids the pedestrian in front of it, it is likely to collide with other pedestrians that cannot be detected due to occlusion. In this study, we used a drone to shoot videos at high altitudes, transferred the videos to a computer, used YOLO on the computer to perform image recognition of pedestrians around the robot, and passed the recognition results to ROS, which enabled it to plan a path to avoid pedestrians around it. As a result, showed that the videos shot at high altitudes could effectively identify pedestrians around the robot after being recognized by YOLO. © 2025 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

A comparative study of recommender systems in different domains 4th

A comparative study of recommender systems in different doma...

引用

4th international Conference on computational Methods in science and technology, ICCMST 2024

作者： Rohit Sharma, Aditi Kumar, Aditya Singh, Prabal Singh, Ashutosh Kumar Department of Computer Science and Engineering Institute of Engineering and Rural Technology Prayagraj India

ISBN: (纸本)9781032911571

Recommender systems are techniques designed to enhance user experience in various domains. They suggest relevant items to users based on their behavior and preferences (Linyuan et al. Feb 2012). These systems are being extensively utilized in domains such as e-commerce, social media, and video streaming platforms. A recommender system aims to provide personalized recommendations to users based on their past behavior, preferences, and other similar data. The recommender system receives input data from users. Several recommender systems are available nowadays, making it challenging to select one. It would be beneficial to conduct a comparative analysis of various recommendation systems. The research paper attempts to conclude the performance of various recommender systems by using a variety of evaluation metrics. Evaluation metrics provide insights into the effectiveness of the recommendation algorithms. This guides the selection of the most appropriate model for a given application (Oona et al. 2024). The comparative study aims to examine and analyze different recommender systems across various domains. We evaluate the performance of these systems in improving user experiences and increasing user interaction. By studying the performance of recommender systems in different domains, we gain insights into the strengths and limitations of each approach. © 2025 the Author(s).

关键词： collaborative content-based evaluation metric hybrid recommendation recommender

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：