检索结果-内蒙古大学图书馆

Contrastive Learning for Generating Optical Coherence Tomography Images of the Retina 1

7th International Workshop on Simulation and Synthesis in Medical Imaging (SASHIMI)

作者： Kaplan, Sinan Lensu, Lasse Lappeenranta Lahti Univ Technol LUT Dept Computat Engn POB 20 Lappeenranta 53850 Finland

ISBN: (数字)9783031169809

ISBN: (纸本)9783031169809;9783031169793

As a self-supervised learning technique, contrastive learning is an effective way to learn rich and discriminative representations from data. In this study, we propose a variational autoencoder (VAE) based approach to apply contrastive learning for the generation of optical coherence tomography (OCT) images of the retina. The approach first learns embedding representation from data by contrastive learning. Secondly, the learnt embeddings are used to synthesize disease-specific OCT images using VAEs. Our results reveal that the diseases are separated well in the embedding space and the proposed approach is able to generate high-quality images with fine-grained spatial details. The source code of the experiments in this paper can be found on Github (https://***/kaplansinan/OCTRetImageGen_CLcVAE).

关键词： Optical coherence tomography Contrastive learning variational autoencoder Deep generative model Deep learning Artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

Continual Learning of Medical Image Classification Based on Feature Replay 16

Continual Learning of Medical Image Classification Based on ...

引用

16th IEEE International Conference on Signal Processing (ICSP)

作者： Li, Xiaojie Li, Haifeng Ma, Lin Harbin Inst Technol Fac Comp Harbin Peoples R China

ISBN: (数字)9781665460569

ISBN: (纸本)9781665460569

The wide variety of diseases in clinical diagnosis makes it impractical to develop specific detection algorithms for each disease. Models with continual learning capabilities learn to detect new disease as needed and can eventually detect all diseases learned before. However, there are few researches on continual learning of medical image classification. In this paper, we design two kinds of continual learning tasks of medical image classification and evaluate continual learning methods in the literature. We propose a novel continual learning method based on feature replay. Our method also utilizes multiple conditional generators to improve quality of replayed samples. Comparison with other methods shows that our method achieves higher average accuracy and lower average forgetting. Inception Score and Frechet Inception Distance show that our method generates better samples which help to overcome catastrophic forgetting significantly.

关键词： Continual learning Generator Generative replay Classification variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Constructing a personalized recommender system for life insurance products with machine-learning techniques

引用

INTELLIGENT SYSTEMS IN ACCOUNTING FINANCE & MANAGEMENT 2022年第4期29卷 242-253页

作者： Kong, Hyeongwoo Yun, Wonje Joo, Weonyoung Kim, Ju-Hyun Kim, Kyoung-Kuk Moon, Il-Chul Kim, Woo Chang Korea Adv Inst Sci & Technol Ind & Syst Engn Daehakro 291 Daejeon South Korea Univ Chicago Social Sci Div Chicago IL 60637 USA Ewha Womans Univ Dept Stat Seoul South Korea Korea Adv Inst Sci & Technol Sch Management Engn Seoul South Korea

The collaborative filtering (CF) recommendation algorithm predicts the purchases of specific users based on their characteristics and purchase history. This study empirically analyzes the possibility of applying CF to the insurance industry using real customer data from South Korea. Using three different CF models, we examined the relevance of applying the CF model to insurance products under various situations by comparing them with logistic-regression-based recommendation models. Through experiments, we empirically show that CF models apply to the insurance industry, especially when customer purchase information is added to the model.

关键词： collaborative filtering customer choice prediction life insurance product recommender system variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Latent space improved masked reconstruction model for human skeleton-based action recognition (vol 19, 1482281, 2025)

引用

FRONTIERS IN NEUROROBOTICS 2025年 19卷 1482281页

作者： Chen, E. Wang, X. Guo, X. Zhu, Y. Li, D. School of Electrical and Information Engineering Zhengzhou University Zhengzhou China. State Grid Henan Electric Power Company Information and Communication Branch Zhengzhou China.

Human skeleton-based action recognition is an important task in the field of computer vision. In recent years, masked autoencoder (MAE) has been used in various fields due to its powerful self supervised learning ability and has achieved good results in masked data reconstruction tasks. However, in visual classification tasks such as action recognition, the limited ability of the encoder to learn features in the autoencoder structure results in poor classification performance. We propose to enhance the encoder's feature extraction ability in classification tasks by leveraging the latent space of variational autoencoder (VAE) and further replace it with the latent space of vector quantized variational autoencoder (VQVAE). The constructed models are called SkeletonMVAE and SkeletonMVQVAE, respectively. In SkeletonMVAE, we constrain the latent variables to represent features in the form of distributions. And in SkeletonMVQVAE, we discretize the latent variables. These help the encoder learn deeper data structures and more discriminative and generalized feature representations. The experiment results on the NTU-60 and NTU-120 datasets demonstrate that our proposed method can effectively improve the classification accuracy of the encoder in classification tasks and its generalization ability in the case of few labeled data. SkeletonMVAE exhibits stronger classification ability, while SkeletonMVQVAE exhibits stronger generalization in situations with fewer labeled data.

关键词： human skeleton-based action recognition variational autoencoder vector quantized variational autoencoder masked reconstruction model self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Fuel injection prediction for a heavy-duty diesel engine based on deep self-encoder Gaussian process regression

引用

FUEL 2025年 399卷

作者： Pan, Mingzhang Qiu, Yunlong Cao, Sheng Sang, Hailang Zhou, Shengkai Ye, Yu Liu, Xing Guan, Wei Guangxi Univ Coll Mech Engn Nanning 530004 Peoples R China Guangxi Yuchai Machinery Co Ltd Tech Res Inst Nanning Peoples R China

In this paper, a fuel injection prediction model based on variational autoencoder (VAE) with Gaussian process regression (GPR) is proposed for a heavy-duty diesel engine to meet stringent China VI emission regulations. The GPR prediction performance is enhanced by using VAE to extract nonlinear features from diesel engine operating conditions and map them to a low-dimensional latent space. The experimental results show that the model achieves an R2 value of 0.999 and a root mean square error (rMSE) of 0.018 in the interpolation task;in the extrapolation task, the model exhibits a good generalization ability (R2 value of 0.976), although the prediction accuracy is slightly reduced. Compared with traditional methods, the VAE-GPR model not only possesses higher prediction accuracy, but also better captures the nonlinear relationship of the data. By comparing the interpolation and extrapolation tasks, this study verifies the superiority of VAE-GPR in diesel injection prediction, especially in modeling complex nonlinear relationships and extracting data features. It provides important theoretical and technical support for China VI diesel engine performance and emission optimization.

关键词： Heavy-duty diesel engine Machine learning variational autoencoder Prediction accuracy Injection stability Fuel consumption

来源：评论

学校读者我要写书评

暂无评论

Multi-Domain Level Generation and Blending with Sketches via Example-Driven BSP and variational autoencoders 20

Multi-Domain Level Generation and Blending with Sketches via...

引用

15th International Conference on the Foundations of Digital Games

作者： Snodgrass, Sam Sarkar, Anurag Modlai Copenhagen Denmark Northeastern Univ Boston MA USA

ISBN: (纸本)9781450388078

Procedural content generation via machine learning (PCGML) has demonstrated its usefulness as a content and game creation approach, and has been shown to be able to support human creativity. An important facet of creativity is combinational creativity or the recombination, adaptation, and reuse of ideas and concepts between and across domains. In this paper, we present a PCGML approach for level generation that is able to recombine, adapt, and reuse structural patterns from several domains to approximate unseen domains. We extend prior work involving example-driven Binary Space Partitioning for recombining and reusing patterns in multiple domains, and incorporate variational autoencoders (VAEs) for generating unseen structures. We evaluate our approach by blending across 7 domains and subsets of those domains. We show that our approach is able to blend domains together while retaining structural components. Additionally, by using different groups of training domains our approach is able to generate both 1) levels that reproduce and capture features of a target domain, and 2) levels that have vastly different properties from the input domain.

关键词： procedural content generation level blending level generation binary space partitioning variational autoencoder PCGML

来源：评论

学校读者我要写书评

暂无评论

Novel method of performance-optimized metastructure design for electromagnetic wave absorption in specific band using deep learning

引用

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2024年第PartB期137卷

作者： On, Han-Ik Jeong, Leekyo Seo, Tae-Moon Jo, Yehrin Choi, Wonwoo Kang, Dong-Joong Park, Jun-Hyub Lee, Hak-Joo Pusan Natl Univ Sch Mech Engn 2 Busandaehak Ro 63beon Gil Busan 46241 South Korea Ctr Adv Meta Mat 156 Gajeongbuk Ro Daejeon 34103 South Korea Kyungsung Univ Dept Mech & Automot Engn 309 Suyeong Ro Busan 48434 South Korea Korea Adv Inst Sci & Technol Cho Chun Shik Grad Sch Mobil 193 Munji Ro Daejeon South Korea

In this paper, we propose a new method that utilizes deep learning techniques to design a metastructure for an electromagnetic absorber. This method enables the effective design of a metastructure with the desired performance (spectrum of S11<- 10 dB) in the frequency band specified by the designer, within a wideband range from 2 to 40 GHz. The proposed absorber consists of two dielectric layers with varied conductive patterns and a back reflector. Critical to the absorber's microwave performance is the binary pattern configuration, organized in a 20-pixel square, along with the sheet resistance and layer thickness of each layer, contributing to a significant design freedom exceeding 1037 degrees of freedom. Our model for performance-optimized design involves three steps: Initially, with limited data from 26,000 sets, a variational autoencoder (VAE) was trained to map S11 spectra and arrange a latent space linked to metastructure. Subsequently, we developed a spectrum prediction network to correlate patterns with S11 spectra, leveraging a pre-trained decoder from the auxiliary VAE in the first step. The final step trains a network for designing a metastructure with broadband absorption. To verify the performance of a metastructure designed by the developed method, we compared their performances with those obtained through Finite Difference Time Domain (FDTD) simulation and the developed network. And also to further validate our approach experimentally, the designed metastructures were fabricated by silkscreen printing using carbon paste ink, and some bands (1-18 GHz, 26.5-40 GHz) were measured to compare with the performance predicted by the VAE network.

关键词： Broadband electromagnetic metastructure absorber Finite difference time domain Deep learning variational autoencoder Performance-driven design Covariance matrix adaptation evolution strategy

来源：评论

学校读者我要写书评

暂无评论

Enhanced Dual-Channel Model-Based with Improved Unet plus plus Network for Landslide Monitoring and Region Extraction in Remote Sensing Images

引用

REMOTE SENSING 2024年第16期16卷 2990页

作者： Wang, Junxin Zhang, Qintong Xie, Hao Chen, Yingying Sun, Rui Beijing Normal Univ Fac Arts & Sci Zhuhai 519087 Peoples R China Beijing Normal Univ Fac Geog Sci State Key Lab Remote Sensing Sci Beijing 100875 Peoples R China

Landslide disasters pose significant threats to human life and property;therefore, accurate and effective detection and area extraction methods are crucial in environmental monitoring and disaster management. In our study, we address the critical tasks of landslide detection and area extraction in remote sensing images using advanced deep learning techniques. For landslide detection, we propose an enhanced dual-channel model that leverages EfficientNetB7 for feature extraction and incorporates spatial attention mechanisms (SAMs) to enhance important features. Additionally, we utilize a deep separable convolutional neural network with a Transformers module for feature extraction from digital elevation data (DEM). The extracted features are then fused using a variational autoencoder (VAE) to mine potential features and produce final classification results. Experimental results demonstrate impressive accuracy rates of 98.92% on the Bijie City landslide dataset and 94.70% on the Landslide4Sense dataset. For landslide area extraction, we enhance the traditional Unet++ architecture by incorporating Dilated Convolution to expand the receptive field and enable multi-scale feature extraction. We further integrate the Transformer and Convolutional Block Attention Module to enhance feature focus and introduce multi-task learning, including segmentation and edge detection tasks, to efficiently extract and refine landslide areas. Additionally, conditional random fields (CRFs) are applied for post-processing to refine segmentation boundaries. Comparative analysis demonstrates the superior performance of our proposed model over traditional segmentation models such as Unet, Fully Convolutional Network (FCN), and Segnet, as evidenced by improved metrics: IoU of 0.8631, Dice coefficient of 0.9265, overall accuracy (OA) of 91.53%, and Cohen's kappa coefficient of 0.9185 on the Bijie City landslide dataset;and IoU of 0.8217, Dice coefficient of 0.9021, overall accuracy (OA) of 96.68%,

关键词： landslide detection dual-channel model feature fusion variational autoencoder improving Unet plus plus model

来源：评论

学校读者我要写书评

暂无评论

Uncovering obscured phonon dynamics from powder inelastic neutron scattering using machine learning

引用

MACHINE LEARNING-SCIENCE AND TECHNOLOGY 2024年第3期5卷 035080页

作者： Su, Yaokun Li, Chen Univ Calif Riverside Mat Sci & Engn Riverside CA 92507 USA Univ Calif Riverside Mech Engn Riverside CA 92507 USA

The study of phonon dynamics is pivotal for understanding material properties, yet it faces challenges due to the irreversible information loss inherent in powder inelastic neutron scattering spectra and the limitations of traditional analysis methods. In this study, we present a machine learning framework designed to reveal obscured phonon dynamics from powder spectra. Using a variational autoencoder, we obtain a disentangled latent representation of spectra and successfully extract force constants for reconstructing phonon dispersions. Notably, our model demonstrates effective applicability to experimental data even when trained exclusively on physics-based simulations. The fine-tuning with experimental spectra further mitigates issues arising from domain shift. Analysis of latent space underscores the model's versatility and generalizability, affirming its suitability for complex system applications. Furthermore, our framework's two-stage design is promising for developing a universal pre-trained feature extractor. This approach has the potential to revolutionize neutron measurements of phonon dynamics, offering researchers a potent tool to decipher intricate spectra and gain valuable insights into the intrinsic physics of materials.

关键词： phonon dynamics inelastic neutron scattering variational autoencoder domain adaptation

来源：评论

学校读者我要写书评

暂无评论

Dynamic behavior recognition in aerial deployment of multi-segmented foldable-wing drones using variational autoencoders

引用

CHINESE JOURNAL OF AERONAUTICS 2025年第6期38卷

作者： Dou, Yilin Zhou, Zhou Wang, Rui Northwestern Polytech Univ Sch Aeronaut Xian 710072 Peoples R China Natl Key Lab Aircraft Configurat Design Xian 710072 Peoples R China

The aerial deployment method enables Unmanned Aerial Vehicles (UAVs) to be directly positioned at the required altitude for their mission. This method typically employs folding technology to improve loading efficiency, with applications such as the gravity-only aerial deployment of high-aspect-ratio solar-powered UAVs, and aerial takeoff of fixed-wing drones in Mars research. However, the significant morphological changes during deployment are accompanied by strong nonlinear dynamic aerodynamic forces, which result in multiple degrees of freedom and an unstable character. This hinders the description and analysis of unknown dynamic behaviors, further leading to difficulties in the design of deployment strategies and flight control. To address this issue, this paper proposes an analysis method for dynamic behaviors during aerial deployment based on the variational autoencoder (VAE). Focusing on the gravity-only deployment problem of highaspect-ratio foldable-wing UAVs, the method encodes the multi-degree-of-freedom unstable motion signals into a low-dimensional feature space through a data-driven approach. By clustering in the feature space, this paper identifies and studies several dynamic behaviors during aerial deployment. The research presented in this paper offers a new method and perspective for feature extraction and analysis of complex and difficult-to-describe extreme flight dynamics, guiding the research on aerial deployment drones' design and control strategies. (c) 2025 Published by Elsevier Ltd on behalf of Chinese Society of Aeronautics and Astronautics. This is an open access article under the CC BY-NC-ND license (http://***/licenses/by-nc-nd/4.0/).

关键词： Dynamic behavior recogni tion Aerial deployment technol ogy variational autoencoder Pattern recognition Multi-rigid-body dynamics

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：