检索结果-内蒙古大学图书馆

Layer-Wise Learning Rate Optimization for Task-Dependent Fine-Tuning of Pre-Trained Models: An Evolutionary Approach

引用

ACM Transactions on Evolutionary Learning and Optimization 2024年第4期4卷 1-23页

作者： Bu, Chenyang Liu, Yuxin Huang, Manzong Shao, Jianxuan Ji, Shengwei Luo, Wenjian Wu, Xindong Key Laboratory of Knowledge Engineering with Big Data Ministry of Education and School of Computer Science and Information Engineering Hefei University of Technology Hefei China School of Artificial Intelligence and Big Data Hefei University Hefei China Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies School of Computer Science and Technology Harbin Institute of Technology Shenzhen China

The superior performance of large-scale pre-Trained models, such as Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-Trained Transformer (GPT), has received increasing attention in both academic and industrial research and has become one of the current research hotspots. A pre-Trained model refers to a model trained on large-scale unlabeled data, whose purpose is to learn general language representation or features for fine-Tuning or transfer learning in subsequent tasks. After pre-Training is complete, a small amount of labeled data can be used to fine-Tune the model for a specific task or domain. This two-stage method of "pre-Training+fine-Tuning"has achieved advanced results in natural language processing (NLP) tasks. Despite widespread adoption, existing fixed fine-Tuning schemes that adapt well to one NLP task may perform inconsistently on other NLP tasks given that different tasks have different latent semantic structures. In this article, we explore the effectiveness of automatic fine-Tuning pattern search for layer-wise learning rates from an evolutionary optimization perspective. Our goal is to use evolutionary algorithms to search for better task-dependent fine-Tuning patterns for specific NLP tasks than typical fixed fine-Tuning patterns. Experimental results on two real-world language benchmarks and three advanced pre-Training language models show the effectiveness and generality of the proposed framework. © 2024 held by the owner/author(s).

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Robust design of damping controller for power system using a combination of snake optimisation algorithm and optimal control theory

引用

International Journal of Energy Technology and Policy 2024年第1-2期19卷 171-215页

作者： Agrawal, Niharika Khan, Faheem Ahmed Gowda, Mamatha Department of Electrical and Electronics Engineering Ghousia College of Engineering Karnataka Ramanagaram District 562 159 India Department of Artificial Intelligence and Data Science BGS College of Engineering and Technology Mahalakshmi Puram Karnataka Bengaluru560 086 India

Low-frequency oscillations (LFO) are created in the power system due to various disturbances. The LFO if not controlled, grows and causes the system separation. There is a huge financial loss due to the interruption of the power supply caused by disturbances. With the increasing complexity of the modern power system, there is a need for the design of a more accurate and detailed modelling. An Advanced Heffron Phillips Model (AHPM) is developed with a higher order Synchronous Generator Model 1.1, based on ten K-Constants for stability improvement. This AHPM employs the combination of snake optimisation algorithm (SOA) and linear quadratic regulator (LQR) from optimal control theory. The highest damping ratio (99.98%) is obtained by AHPM in coordination with PSS, and TCSC based on SOA and LQR. For various parameters, the settling time ranges from 1.5 to 2.0 seconds. This AHPM is robust and capable of meeting the challenges of grid integration with renewables. © 2024 Inderscience Enterprises Ltd.

关键词： Damping

来源：评论

学校读者我要写书评

暂无评论

Next-Gen Dermatology: Revolutionizing Dermatological Diagnostics with AI and Emergency Care Solutions 3

Next-Gen Dermatology: Revolutionizing Dermatological Diagnos...

引用

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

作者： Yashas, D. Kashyap, M. Shivani Charan, P.R. Hemanth Kumar, G.P. Abhishek, B.K. Deekshith, C. New Horizon College of Engineering Dept of Artificial Intelligence and Machine Learning Bengaluru India New Horizon College of Engineering Dept of Information Science and Engineering Bengaluru India Bnm Institute of Technology Dept of Artificial Intelligence and Machine Learning Bengaluru India Siddaganga Institute of Technology Dept of Artificial Intelligence and Data Science Tumkur India Vidyavardhaka College of Engineering Dept of Computer Science and Engineering Mysore India

ISBN: (纸本)9798331515911

Skin health conditions should be diagnosed and treated on time so as to prevent complications and promote better health results. This paper presents an AI-based system that makes use of Convolutional Neural Networks with the InceptionV3 Architecture for predicting skin diseases at higher levels. The model, which is based on the large DermaNet dataset, classifies skin disorders fast and effectively. The system is coupled with a simple Telegram based application for the users to get a complete solution. A user can upload a photo of the affected skin area and get a diagnosis within seconds, an explanation of the disease and how severe it is in percentage. The bot also has a location and contact sharing feature that assists users in finding nearby doctors and clinics that can provide further assistance. It further provides on the spot appropriate cures for the particular condition diagnosed. Moreover, this makes it possible for the customer to fill the healthcare void in particular the neglected ones by making it easy for them to access real time high quality services. By equipping users with relevant information and appropriate actions required at the right time, this project has showcased how AI has the ability to transform the dermatological field. © 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

Normalized AlexNet Deep Learning based Edibility Classification of Sporocarp 2

Normalized AlexNet Deep Learning based Edibility Classificat...

引用

2nd International Conference on Advancement in Computation and Computer Technologies, InCACCT 2024

作者： Shyamala Devi, M. Irene, B. Kanagavalli, A. Kousika, A. Joy Praisy, E. Humaira, S. Panimalar Engineering College Department of Computer Science and Engineering Tamilnadu Chennai India Panimalar Engineering College Department of Artificial Intelligence and Data Science Tamilnadu Chennai India

ISBN: (纸本)9798350371314

The manufacturing and dissemination of spores is the main purpose of sporocarps which is the specialized type of configuration found in freshwater plants of the Salviniales family. Though sporocarp are mainly used for reproduction, some of them are consumable by the animals and birds for the extraction of rich nutrients. Sporocarp categorization is an extremely difficult task as various species of sporocarp share significant appearance. Some sporocarp may be poisonous that lead to brain damage and infection to the animals. This paper proposes Normalized AlexNet that classifies the edible and poisonous sporocarps with high efficiency. For the sake of this investigation, 1800 sporocarp images from the Sporocarp edible dataset were used. The best CNN model was chosen by applying the Sporocarp edible dataset to the current CNN models. With an accuracy of 89.67%, AlexNet performs better than the other CNN models. AlexNet has been chosen to fine-tune its single dense layer at this level. The AlexNet typically consists of two blocks with a single convolution including a max pooling layer, one block with three convolutions along with a max pooling layer, and one last block with three dense layers. The proposed Normalized AlexNet was now suggested by modifying the second dense layer with a single block of batch normalization, relu optimization, and drop out with a vector size of 4096. The suggested Normalized AlexNet and the current CNN have both been fitted with the Sporocarp Edible dataset. The demonstration illustrates that the Normalized AlexNet model performs better in the edibility categorization of sporocarp, with a high accuracy of 98.76%. © 2024 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Emotion Recognition Using Focal Network and Resnet

Emotion Recognition Using Focal Network and Resnet

引用

2024 IEEE International Conference on Information Technology, Electronics and Intelligent Communication Systems, ICITEICS 2024

作者： Meena, T. Balasai, Sigireddy Anandh Duggireddy Kumar, P. Phaneendhra Velagapudi Ramakrishna Siddartha Engineering College Department of Computer Science and Engineering India Velagapudi Ramakrishna Siddartha Engineering College Department of Artificial Intelligence and Data Science India

ISBN: (纸本)9798350382693

Image emotion recognition involves finding the emotions from visual data, usually done through convolutional neural networks (CNN) or deep neural networks (DNN). The existing methodologies are often high complex or time complex and it is not giving the good considerable results and computationally expensive due to high number of parameters. In this paper we try to accomplish a better output by using focal network which stood among other transformers architecture due to the unique property of the dynamic attention Recent studies have shown the efficacy of transformer architectures in various deep learning tasks and challenged the convolution neural networks on various tasks. but, It has a tradeoff between reduced parameter complexity and the need for extensive training data. Drawing inspiration from recent publications, the study adopts a focal network for training, demonstrating its effectiveness in enhancing the model's attention and optimizing emotional feature extraction. The proposed methodology aims to do the process of image emotion recognition by using the advantages of transformers while mitigating the computational problem associated with the deep neural networks. The incorporation of a focal network addresses attention-related challenges, and it contributes to the development of more efficient emotion recognition systems. This research offers insights into the evolving landscape of image emotion recognition methodologies and underscores the potential of transformer architectures in advancing this field. © 2024 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Early Diagnosis of Schizophrenia in Patients Using Deep Learning Techniques 10

Early Diagnosis of Schizophrenia in Patients Using Deep Lear...

引用

10th International Conference on Advanced Computing and Communication Systems, ICACCS 2024

作者： Priyanka, G. Nivaashini, M. Pavithra, S. Srimaha, T.S. Kanishgaa, P.R. Sri Ramakrishna Engineering College Department of Computer Science and Engineering Coimbatore India Sri Eshwar College of Engineering Department of Artificial Intelligence and Data Science Coimbatore India

ISBN: (数字)9798350384369

ISBN: (纸本)9798350384369

While Brain, emotion, and behaviours abnormalities are hallmarks of schizophrenia, a complex and long-lasting mental illness. Its symptoms are divided into three categories: positive, negative, and cognitive. Positive symptoms include delusions, disordered thinking, abnormal gait, and hallucinations. Diminished ability to express emotions, alogia, anhedonia, avolition, and social disengagement are detrimental symptoms. Deficiencies in working memory, problems paying attention, and trouble making decisions are examples of cognitive symptoms. Environmental variables, genetic predisposition, and neurochemical imbalances interact intricately in an etiology of schizophrenia. Typically, antipsychotic drugs, psychotherapy, and social support are used in conjunction for treatment. The management of symptoms and enhancement of the general health of persons diagnosed with schizophrenia need early intervention and customized treatment strategies. The CNN, BiLSTM, GRU, and Transformer deep learning models are implemented in this study, and their respective performances are assessed. The transformer and GRU models outperform the other models when hyperparameter adjustment is carried out to increase the models' efficacy. © 2024 IEEE.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

VaeDiff-DocRE: End-to-end data Augmentation Framework for Document-level Relation Extraction 31

VaeDiff-DocRE: End-to-end Data Augmentation Framework for Do...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Tran, Khai Phan Hua, Wen Li, Xue School of Electrical Engineering and Computer Science The University of Queensland Australia Department of Data Science and Artificial Intelligence The Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)9798891761964

Document-level Relation Extraction (DocRE) aims to identify relationships between entity pairs within a document. However, most existing methods assume a uniform label distribution, resulting in suboptimal performance on real-world, imbalanced datasets. To tackle this challenge, we propose a novel data augmentation approach using generative models to enhance data from the embedding space. Our method leverages the Variational Autoencoder (VAE) architecture to capture all relation-wise distributions formed by entity pair representations and augment data for underrepresented relations. To better capture the multi-label nature of DocRE, we parameterize the VAE's latent space with a Diffusion Model. Additionally, we introduce a hierarchical training framework to integrate the proposed VAE-based augmentation module into DocRE systems. Experiments on two benchmark datasets demonstrate that our method outperforms state-of-the-art models, effectively addressing the long-tail distribution problem in DocRE. Our code is released at: https://***/khaitran22/VaeDiff-DocRE. © 2025 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Parallelism in Cloud Architecture using Implicit Partition Clustering Framework 4

Parallelism in Cloud Architecture using Implicit Partition C...

引用

4th International Conference on Soft Computing for Security Applications, ICSCSA 2024

作者： Sudharson, D. Diwakaran, M. Senthil Kumar, V. Sukirtha, S. Afsheen Zaahrah, A. Karthick, A. Artificial Intelligence and Data Science Kumaraguru College of Technology Tamil Nadu Coimbatore India Sri Krishna College of Eng & Tech Dept of IT Tamil Nadu Coimbatore India Kumaraguru College of Technology Dept of CSE Tamil Nadu Coimbatore India Shiv Nadar University Chennai Artificial Intelligence and Data Science Tamil Nadu Chennai India Kumaraguru College of Engineering Artificial Intelligence and Data Science Tamil Nadu Coimbatore India Kumaraguru College of Technology Artificial Intelligence and Data Science Tamil Nadu Coimbatore India

ISBN: (数字)9798331515720

ISBN: (纸本)9798331515720

To overcome the reliability issues concerning data's exponential growth, many optimization techniques are deployed in artificial intelligence. Parallelism in a distributed environment provides an effective solution to handle this expansion efficiently. Catering to real-Time and batch-Time requirements in distributed environments, Spark and Flink are two prominent, powerful, versatile, fault-Tolerant, open-source frameworks providing a unified platform for large data processing. Primarily analyzing the efficiency optimization through implicit partitioning clustering algorithms and adaptable K-means. Combining a few methods for adaptable K value enhances the robustness of the clustering solution, which effectively yields optimized performance. © 2024 IEEE.

关键词： Cloud platforms

来源：评论

学校读者我要写书评

暂无评论

Deep Masked Input UNet Framework based Lips Segmentation prediction for Gender Classification 2

Deep Masked Input UNet Framework based Lips Segmentation pre...

引用

2nd IEEE International Conference on Advances in Information Technology, ICAIT 2024

作者： Devi, M. Shyamala Hepzhibha Rachel, S. Janani, G. Irene, B. Joy Praisy, E. Humaira, S. Department of Computer Science and Engineering Panimalar Engineering College Tamilnadu Chennai India Department of Artificial Intelligence and Data Science Panimalar Engineering College Tamilnadu Chennai India

ISBN: (纸本)9798350383867

Assessments have demonstrated that the human lip and its motions provide a wealth of knowledge about the identity and substance of communication. However, due to large differences in illumination condition, head perspective, and background, obtaining strong and precise lip image segmentation in natural settings remains difficult. This paper recommends Deep Masked Input UNet that categorizes the Gender based on the lip segmentation with high precision. For implementation, 10,132 face images from the KAGGLE Lip Segmentation dataset were used. The dataset includes 5066 images of the face and 5066 segmented lip images. The proposed Deep Masked Input UNet starts by masking the original picture with a segmented lip image to create masked face images with lips segments. Utilizing the Relu Activation function, Deep Masked Input UNet with contracting and expansion was enabled. The masked lip images are fitted to the proposed Deep Masked Input UNet and traditional CNN models. The results show that the suggested Deep Masked Input UNet model performs better in lip segmentation and gender classification, with a high accuracy of 98.95%. © 2024 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Machine learning and quantum computing for 5G/6G communication networks - A survey

引用

International Journal of Intelligent Networks 2022年 3卷 197-203页

作者： M, Suriya Department of Artificial Intelligence & Data Science Sri Eshwar College of Engineering Tamil Nadu Coimbatore India

Recently fifth generation (5G) and beyond applications are evolving, which demands more computational and complex data processing. Quantum computing and quantum learning algorithms are incorporated to enhance processing capabilities and data computation compared to conventional machine learning approaches. This study presents the significance of quantum computing and quantum machine learning models and their research challenges concerning 5G and beyond applications. The researchers focus on global coverage, enhanced spectrum support, increased energy and cost efficiency, high security, and dynamic intelligence, along with big data processing that demands complex data structures and algorithms. © 2022 The Author

关键词： Learning algorithms

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：