检索结果-内蒙古大学图书馆

Geometric prior guided hybrid deep neural network for facial beauty analysis

CAAI Transactions on Intelligence Technology 2024年第2期9卷 467-480页

作者： Tianhao Peng Mu Li Fangmei Chen Yong Xu David Zhang The School of Computer Science and Technology Guizhou UniversityGuiyangChina Department of Automation Moutai InstituteRenhuaiGuizhouChina The School of Computer Science and Technology Harbin Institute of TechnologyShenzhenShenzhenChina The Information and Communication Engineering Department Dalian Minzu UniversityDalianChina The School of Data Science The Chinese University of Hong KongShenzhenShenzhenChina

Facial beauty analysis is an important topic in human *** may be used as a guidance for face beautification applications such as cosmetic *** neural networks(DNNs)have recently been adopted for facial beauty analysis and have achieved remarkable ***,most existing DNN-based models regard facial beauty analysis as a normal classification *** ignore important prior knowledge in traditional machine learning models which illustrate the significant contribution of the geometric features in facial beauty *** be specific,landmarks of the whole face and facial organs are introduced to extract geometric features to make the *** by this,we introduce a novel dual-branch network for facial beauty analysis:one branch takes the Swin Transformer as the backbone to model the full face and global patterns,and another branch focuses on the masked facial organs with the residual network to model the local patterns of certain facial ***,the designed multi-scale feature fusion module can further facilitate our network to learn complementary semantic information between the two *** model optimisation,we propose a hybrid loss function,where especially geometric regulation is introduced by regressing the facial landmarks and it can force the extracted features to convey facial geometric *** performed on the SCUT-FBP5500 dataset and the SCUT-FBP dataset demonstrate that our model outperforms the state-of-the-art convolutional neural networks models,which proves the effectiveness of the proposed geometric regularisation and dual-branch structure with the hybrid *** the best of our knowledge,this is the first study to introduce a Vision Transformer into the facial beauty analysis task.

关键词： deep neural networks face analysis face biometrics image analysis

来源：评论

学校读者我要写书评

暂无评论

p-Mean Regret for Stochastic Bandits 39

p-Mean Regret for Stochastic Bandits

引用

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Krishna, Anand John, Philips George Barik, Adarsh Tan, Vincent Y.F. Department of Electrical and Computer Engineering NUS Singapore CNRS-CREATE Department of Computer Science NUS Singapore Institute of Data Science NUS Singapore Department of Mathematics NUS Singapore

ISBN: (纸本)157735897X

In this work, we extend the concept of the p-mean welfare objective from social choice theory to study p-mean regret in stochastic multi-armed bandit problems. The p-mean regret, defined as the difference between the optimal mean among the arms and the p-mean of the expected rewards, offers a flexible framework for evaluating bandit algorithms, enabling algorithm designers to balance fairness and efficiency by adjusting the parameter p. Our framework encompasses both average cumulative regret and Nash regret as special cases. We introduce a simple, unified UCB-based algorithm (EXPLORE-THEN-UCB) that achieves novel p-mean regret bounds. Our algorithm consists of two phases: a carefully calibrated uniform exploration phase to initialize sample means, followed by the UCB1 algorithm of Auer et al. Under mild assumptions, we prove that our algorithm achieves a p-mean regret bound of O∼(rT2k|1p|) for all p ≤ −1, where k represents the number of arms and T the time horizon. When −1 ∼(k43 T−41). For the range 0 ∼(qTk), which matches the previously established lower bound up to logarithmic factors. This result stems from the fact that the p-mean regret of any algorithm is at least its average cumulative regret for p ≤ 1. In the case of Nash regret (the limit as p approaches zero), our unified approach differs from prior work of Barman et al., which requires a new Nash Confidence Bound algorithm. Notably, we achieve the same regret bound up to constant factors using our more general method. Copyright © 2025, Association for the Advancement of Artificial Intelligence.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prediction of Alzheimer’s Disease Using Adaptive Fine-Tuned Deep Resnet-50 with Attention Mechanism

引用

SN computer science 2024年第4期5卷 392页

作者： Venkatesh, R. Anantharajan, Shenbagarajan Gunasekaran, Shenbagalakshmi Yogaraja, C.A. Gethzi Ahila Poornima, I. Computer Science and Engineering Ramco Institute of Technology Rajapalayam Tamil Nadu Virudhunagar 626117 India Artificial Intelligence and Data Science Mepco Schlenk Engineering College Sivakasi Tamilnadu Virudhunagar 626005 India Computer Science and Engineering Mepco Schlenk Engineering College Sivakasi Tamilnadu Virudhunagar 626005 India

Alzheimer’s disease (AD) sufferers can benefit from early detection and diagnosis so that protective treatments can be taken before permanent brain impairment develops. Magnetic resonance imaging (MRI)-detected lesion sites in AD patients exhibit significant heterogeneity and these characteristics are spread throughout the imaging space, according to cross-sectional imaging studies of Alzheimer's disease. Because conventional Deep Learning (DL) methods can’t generate enough long-distance correlations in the characteristic space, adaptive fine-tuned deep resnet-50 with attention mechanism (AFDR-AM) is introduced in this research for the early identification of AD. The input MRI data is de-noised using a wavelet thresholding approach, and the contrast is enhanced using Improved Color Histogram Equalization (ICHE). Image recognition and classification characteristics are extracted at the feature extraction phase using a Gabor Filter Bank (GFB). To narrow down a big list of traits, the Artificial Bee Colony Optimization (ABCO) technique can be used to pick the most useful ones and get rid of the rest. Measures of accuracy, precision, recall, f1-score, specificity, and MSE are used to evaluate the effectiveness of the suggested strategy. Comprehensive experiments reveal that our approach achieved the best performance compared to other approaches. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd 2024.

关键词： AFDR-AM Alzheimer’s disease (AD) Artificial Bee colony optimization (ABCO) Deep learning (DL) Gabor filter bank (GFB) Improved color histogram equalization (ICHE) Magnetic resonance imaging (MRI)

来源：评论

学校读者我要写书评

暂无评论

Deformable Dual Graph Aggregation Transformer Convolutional Networks with Spider Wasp Optimizer for Ovarian Tumor Classification Using Magnetic Resonance Imaging

Biomedical Materials and Devices

引用

Biomedical Materials and Devices 2025年 1-19页

作者： Shanmugaveni, V. Jotheeswari, M. Abarnaswara, R. VijayaKumar, M. Manojanani, A. Department of Computer Science and Engineering (Cyber Security) Dr. Mahalingam College of Engineering and Technology Tamil Nadu Pollachi India Department of Computer Science and Engineering PSNA College of Engineering and Technology Tamil Nadu Dindigul India Department of Artificial Intelligence & Data Science Dr. Mahalingam College of Engineering and Technology Tamil Nadu Pollachi India Department of Computer Science and Engineering RVS Technical Campus Tamil Nadu Coimbatore India

The diagnosis of Ovarian Tumor (OT) remains a significant challenge as there is presently no practical non-invasive technique to determine true benign or malignant lesions before treatment. This study proposes a unique Deformable Dual Graph Aggregation Transformer Convolutional Network (DeDGATCNet) with Spider Wasp Optimizer (SWO) for the classification of ovarian tumors acquired from magnetic resonance imaging (MRI). In contrast to traditional deep learning models, DeDGATCNet + SWO directly integrates bending graph-based forming transformer convolutional learning and improves representation for better classification. The SWO is combined to optimize the parameters of the network, thereby improving its performance. The procedures include pre-processing the MRI data from the study of ovarian tumor TCGA-OV datasets with an Adaptive Bilateral Texture Filter (ABTF) for removing artifacts and normalization of intensity, accompanied by a segment using a Masked-Attention Mask Transformer (MAMT). Feature extraction is achieved through a One-Dimensional Quaternion Discrete Fourier Transform (1D-QDFT), which captures features from both the spatial and frequency domains. DeDGATCNet + SWO performance is way ahead and sets a new accuracy record of 99.9% for state-of-the-art models. This research introduces a new type of hybridization between deformable graph transformers, quaternion-based spectral analysis, and nature-inspired optimization techniques, setting non-invasive OT at a new height for classification and improving early diagnosis. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Adaptive bilateral texture filter Deformable dual graph aggregation transformer convolutional networks Masked-attention mask transformer One-dimensional quaternion discrete fourier transform

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

Market resilience in turbulent times: a proactive approach to predicting stock market responses during geopolitical tensions

引用

Journal of Capital Markets Studies 2024年第2期8卷 173-194页

作者： Maddodi, Srivatsa Kunte, Srinivasa Rao Department of Computer Science Srinivas University Mangaluru India Department of Data Engineering United Airlines Inc Bengaluru India

Purpose: The Indian stock market can be tricky when there's trouble in the world, like wars or big conflicts. It's like trying to read a secret message. We want to figure out what makes investors nervous or happy, because their feelings often affect how they buy and sell stocks. We're building a tool to make prediction that uses both numbers and people's opinions. Design/methodology/approach: Hybrid approach leverages Twitter sentiment, market data, volatility index (VIX) and momentum indicators like moving average convergence divergence (MACD) and relative strength index (RSI) to deliver accurate market insights for informed investment decisions during uncertainty. Findings: Our study reveals that geopolitical tensions' impact on stock markets is fleeting and confined to the short term. Capitalizing on this insight, we built a ground-breaking predictive model with an impressive 98.47% accuracy in forecasting stock market values during such events. Originality/value: To the best of the authors' knowledge, this model's originality lies in its focus on short-term impact, novel data fusion and high accuracy. Focus on short-term impact: Our model uniquely identifies and quantifies the fleeting effects of geopolitical tensions on market behavior, a previously under-researched area. Novel data fusion: Combining sentiment analysis with established market indicators like VIX and momentum offers a comprehensive and dynamic approach to predicting market movements during volatile periods. Advanced predictive accuracy: Achieving the prediction accuracy (98.47%) sets this model apart from existing solutions, making it a valuable tool for informed decision-making. © 2024, Srivatsa Maddodi and Srinivasa Rao Kunte.

关键词： Geopolitical tension LSTM Momentum indicator Sentiment analysis Stock market prediction

来源：评论

学校读者我要写书评

暂无评论

G-SPAC:a more granular greedy graph partition algorithm with spatial locality and judgment-aware edge folding

引用

science China(Information sciences) 2025年

作者： Yuedan CHEN Zhijie LI Guoqing XIAO Peixin XU Xiaofei ZHANG Kenli LI Big Data Institute Central South University College of Computer Science and Electronic Engineering Hunan University National Supercomputing Center in Changsha Department of Computer Science University of Memphis

Graph partitioning is the basis of many graph computations. A good graph partitioning algorithm is related to the partition balance and the communication overhead between nodes after partitioning. This paper measures the former by the relative standard deviation(RSD). At the same time, the latter is often related to the number of replicated vertices or edge cuts. In order to balance the above two goals, the greedy strategy is adopted to improve, which is a graph edge partitioning algorithm that transforms the graph so that the vertex partitioning methods can be applied to the transformed graph. A novel graph edge partitioning algorithm named G-SPAC is proposed. Firstly, to reduce the number of duplicated vertices, the method of degree-and-spatial-locality priority is designed so that adjacent vertices are more accessible to divide into the same subgraph. Secondly, to improve the division balance, the judgment-aware folding method is designed to carry out the graph division from a more granular perspective. Experiments are conducted on eighteen graph data sets. The results show that for most of the graphs, the RSD for the number of edges of each subgraph partitioned by the G-SPAC algorithm is below 3%. In terms of the number of vertex copies, G-SPAC's performance is not inferior to the SPAC algorithm, and it can even reduce the number of copies by an order of magnitude on some graphs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Digitalizing the Dress-up Experience: An Exploration of Virtual Try-On for Traditional Chinese Costume 18

Digitalizing the Dress-up Experience: An Exploration of Virt...

引用

18th IEEE Conference on Industrial Electronics and Applications, ICIEA 2023

作者： Lu, Yifan Zhang, Maochun Huang, Wenxuan Guan, Shouqin Ningbotech University College of Computer Science and Data Engineering Ningbo China

ISBN: (纸本)9798350312201

With the recent rapid developments in artificial intelligence (AI) technology, virtual try-on technology empowered by generative approaches has reached a level of maturity that is ripe for commercialization. However, the application of virtual try-on technology to traditional cultural costumes has not received adequate attention from research institutions and enterprises. To address this research gap, we present a novel virtual try-on system for traditional Chinese costume that incorporates diverse AI technologies, including generative models, pose estimation, human parsing, and image processing algorithms, to provide an accurate and immersive virtual try-on experience. In particular, we leverage five distinct techniques Human Keypoint Detection, Human Parsing, Densepose Parts Segmentation, Cloth Mask Extraction, and Clothing-Agnostic Representation - to realize a web service that can effectively handle the specific challenges of traditional Chinese costume, such as the need to preserve cultural authenticity and ensure accurate fit and drape. Finally, we conduct various experiments to evaluate the system's performance and have received positive feedback from users, indicating that it provides a realistic and engaging virtual try-on experience. Our work contributes to the growing body of research on virtual try-on technology and has the potential to impact the way traditional cultural costumes are designed, marketed, and worn. © 2023 IEEE.

关键词： Web services

来源：评论

学校读者我要写书评

暂无评论

Encrypt Decrypt ReLU Activated UNet Prototype based Prediction of Leaf Disease Segmentation

Encrypt Decrypt ReLU Activated UNet Prototype based Predicti...

引用

2024 International Conference on Signal Processing, Computation, Electronics, Power and Telecommunication, IConSCEPT 2024

作者： Shyamala Devi, M. Eswar, R. Hibban Rashid, M.D. Jai Harshan, J.H. Hari, P. Hemanth, 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)9798331540685

The issues of sustainable agriculture and climate change are intimately linked to the problem of effective leaf disease prevention. In practice, the primary method for identifying and detecting plant diseases is expert observation with the unaided eye. However, this necessitates for the ongoing supervision of specialists, which can be too costly for big farms. The automatic identification of leaf diseases is a significant area of research because it might be useful for monitoring vast agricultural fields and identify diseases based only on symptoms that occur on the leaves of the plants. This paper recommends ReLU Activated UNet (RAA-UNet) that categorizes the leaf diseases with high precision. The KAGGLE Leaf disease dataset that contains 2094 leaf images were used for implementation. The dataset consists of 1047 leaf image and 1047 respective leaf images that have been segmented. The proposed ReLU Activated UNet initiates by creating masked leaf images by masking the original image with a segmented leaf image. To classify the leaf disease, the masked leaf images are assigned to ReLU Activated UNet with contracting and expansion which was activated using Relu Activation function. The proposed ReLU Activated UNet and conventional CNN models are fitted to the masked leaf images. As demonstrated by the experiments, the proposed ReLU Activated UNet model works better with a high accuracy of 98.85% in the classification of leaf disease like leaf spot, blight and powdery mildew. © 2024 IEEE.

关键词： Accuracy Activation CNN ReLU Segmentation UNet

来源：评论

学校读者我要写书评

暂无评论

Corner Key Point Extracted Deep Learning Framework Based Birds Species Classification 8

Corner Key Point Extracted Deep Learning Framework Based Bir...

引用

8th IEEE International Conference on Computational System and Information Technology for Sustainable Solutions, CSITSS 2024

作者： Devi, M. Shyamala Jamal Mohammed, R. Hemanth, S. Harish Raj, B. Gajapathy, R. Kamalesh, P. 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)9798331505462

Climate change and habitat destruction have caused the extinction of many bird species. Classifying bird species is a crucial aspect of ornithology, as it aids in the comprehension of species distribution, habitat needs, and environmental changes that impact bird populations. Ornithologists can use variations in bird species distributions to measure the health of a particular environment. Expert ornithologists have relied heavily on visual inspection, which can be tedious, time-consuming, and unreliable when classifying bird species. The development of technology can be used to categorize the bird species. This paper proposes Corner Key Filtered VGG16 (CKP-VGG16) that categorizes eight bird species with high accuracy. The execution was carried out using the 4000 bird images from the KAGGLE Birds species dataset. First, the CKP-VGG16 model separates the images of birds, and then it labels each image. After data augmentation on the labeled images, 84,000 bird images are produced. The augmented images are transformed into grayscale versions, which are then used in filtering techniques to produce Corner Key Point Filtered (CKP) bird image. The performance is analyzed by applying the CKP filtered bird images to the current CNN models. The application demonstrates that the suggested CKP-VGG16 model performs significantly well in classifying bird species, with a high accuracy of 99.7%. © 2024 IEEE.

关键词： Birds

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：