检索结果-内蒙古大学图书馆

automatic data augmentation for medical image segmentation using Adaptive Sequence-length based Deep Reinforcement Learning

引用

COMPUTERS IN BIOLOGY AND MEDICINE 2024年 169卷 107877-107877页

作者： Xu, Zhenghua Wang, Shengxin Xu, Gang Liu, Yunxin Yu, Miao Zhang, Hongwei Lukasiewicz, Thomas Gu, Junhua Hebei Univ Technol Sch Hlth Sci & Biomed Engn State Key Lab Reliabil & Intelligence Elect Equipm Tianjin Peoples R China Hebei Univ Technol Sch Artificial Intelligence Tianjin Peoples R China Tianjin Univ Technol Sch Comp Sci & Engn Tianjin Peoples R China Vienna Univ Technol Inst L & Computat Vienna Austria Univ Oxford Dept Comp Sci Oxford England

Although existing deep reinforcement learning-based approaches have achieved some success in image augmentation tasks, their effectiveness and adequacy for data augmentation in intelligent medical image analysis are still unsatisfactory. Therefore, we propose a novel Adaptive Sequence-length based Deep Rein-forcement Learning (ASDRL) model for automatic data augmentation (AutoAug) in intelligent medical image analysis. The improvements of ASDRL-AutoAug are two-fold: (i) To remedy the problem of some augmented images being invalid, we construct a more accurate reward function based on different variations of the augmentation trajectories. This reward function assesses the validity of each augmentation transformation more accurately by introducing different information about the validity of the augmented images. (ii) Then, to alleviate the problem of insufficient augmentation, we further propose a more intelligent automatic stopping mechanism (ASM). ASM feeds a stop signal to the agent automatically by judging the adequacy of image augmentation. This ensures that each transformation before stopping the augmentation can smoothly improve the model performance. Extensive experimental results on three medical image segmentation datasets show that (i) ASDRL-AutoAug greatly outperforms the state-of-the-art data augmentation methods in medical image segmentation tasks, (ii) the proposed improvements are both effective and essential for ASDRL-AutoAug to achieve superior performance, and the new reward evaluates the transformations more accurately than existing reward functions, and (iii) we also demonstrate that ASDRL-AutoAug is adaptive for different images in terms of sequence length, as well as generalizable across different segmentation models.

关键词： Adaptive Sequence-length Deep Reinforcement Learning automatic data augmentation Reward function Medical image segmentation

来源：评论

学校读者我要写书评

暂无评论

automatic data augmentation Method with Improved Interpretability for Image Classification in Computer Vision Applications

Automatic Data Augmentation Method with Improved Interpretab...

引用

14th Annual Summit and Conference of the Asia-Pacific-Signal-and-Information-Processing-Association (APSIPA ASC)

作者： Ungarbayev, Dair Demirel, Osman Akhtar, Muhammad Tahir Nazarbayev Univ Dept Elect & Comp Engn Sch Engn & Digital Sci Kabanbay Batyr Ave 53 Nur Sultan Kazakhstan

ISBN: (数字)9786165904773

ISBN: (纸本)9786165904773

This paper presents an interpretable automatic data augmentation method. While the mechanism of the existing automatic augmentation methods is not easily understandable, the proposed method seeks to make the process of finding an augmentation policy more clear while maintaining simplicity. The proposed method has been tested on the Fashion-MNIST with a simple convnet and on the CIFAR10 dataset with ResNet50 network. Moreover, it was compared with the RandAugment, Augmix and TrivialAugment methods using benchmark datasets Fashion MNIST-C and CIFAR-10C. The developed method exhibits comparable results to that of existing approaches with the key advantage of interpretability.

关键词： automatic data augmentation Robustness Image classification

来源：评论

学校读者我要写书评

暂无评论

automatic data augmentation by Upper Confidence Bounds for Deep Reinforcement Learning 21

Automatic Data Augmentation by Upper Confidence Bounds for D...

引用

21st International Conference on Control, Automation and Systems (ICCAS)

作者： Gil, Yoonhee Baek, Jongchan Park, Jonghyuk Han, Soohee POSTECH Dept Convergence IT Engn Pohang 37673 South Korea

ISBN: (纸本)9788993215212

In visual reinforcement learning (RL), various approaches succeeded to improve data efficiency. However, the approaches fail to show generalization capabilities if different colors or backgrounds are applied to its environment. The lack of generalization capabilities can hinder the use of RL in real-world environment, which contains lot of distractions and noises. In this paper, a novel automatic data augmentation method that can improve generalization capabilities of an RL agent. In the experiments, the proposed method shows better generalization capabilities than other approaches. These results provide a simple automatic data augmentation method for RL that can improve generalization capabilities.

关键词： Reinforcement learning automatic data augmentation Upper confidence bounds Continuous control

来源：评论

学校读者我要写书评

暂无评论

AutoPedestrian: An automatic data augmentation and Loss Function Search Scheme for Pedestrian Detection

引用

IEEE TRANSACTIONS ON IMAGE PROCESSING 2021年 30卷 8483-8496页

作者： Tang, Yi Li, Baopu Liu, Min Chen, Boyu Wang, Yaonan Ouyang, Wanli Hunan Univ Coll Elect & Informat Engn Changsha 410082 Hunan Peoples R China Natl Engn Lab Robot Visual Percept & Control Tech Changsha 410082 Hunan Peoples R China Baidu USA Sunnyvale CA 94089 USA Univ Sydney Sch Elect & Informat Engn Sydney NSW 2006 Australia

Pedestrian detection is a challenging and hot research topic in the field of computer vision, especially for the crowded scenes where occlusion happens frequently. In this paper, we propose a novel AutoPedestrian scheme that automatically augments the pedestrian data and searches for suitable loss functions, aiming for better performance of pedestrian detection especially in crowded scenes. To our best knowledge, it is the first work to automatically search the optimal policy of data augmentation and loss function jointly for the pedestrian detection. To achieve the goal of searching the optimal augmentation scheme and loss function jointly, we first formulate the data augmentation policy and loss function as probability distributions based on different hyper-parameters. Then, we apply a double-loop scheme with importance-sampling to solve the optimization problem of data augmentation and loss function types efficiently. Comprehensive experiments on two popular benchmarks of CrowdHuman and CityPersons show the effectiveness of our proposed method. In particular, we achieve 40.58% in MR on CrowdHuman datasets and 11.3% in MR on CityPersons reasonable subset, yielding new state-of-the-art results on these two datasets.

关键词： Search problems data models Computer vision Task analysis Optimization Training Object detection Pedestrian detection automatic data augmentation loss function search

来源：评论

学校读者我要写书评

暂无评论

Improved wafer map defect pattern classification using automatic data augmentation based lightweight encoder network in contrastive learning

引用

JOURNAL OF INTELLIGENT MANUFACTURING 2024年 1-13页

作者： Sheng, Yi Yan, Jinda Piao, Minghao Soochow Univ Sch Comp Sci & Technol Suzhou 215006 Jiangsu Peoples R China

In recent years, supervised learning has been the predominant method for wafer map defect pattern classification (WM-DPC), requiring a substantial amount of labeled data to build effective models. Nonetheless, gathering industrial data is challenging and demands significant manual labeling efforts, making it both expensive and time-consuming. To overcome these obstacles, we introduced a contrastive learning framework for WM-DPC based on automatic data augmentation. This innovative augmentation approach takes account of the regional defect density characteristic of various defect types, addressing the limitations of traditional fixed data augmentation and improving the model's generalization capacity. The framework operates in two phases. At first, a lightweight encoder extracts rich representative features from unlabeled data. Then, the classification network is fine-tuned with a limited labeled data set. Experimental outcomes using the public WM-811K dataset showed that the proposed automatic data augmentation and lightweight encoder effectively captured detailed representative features from unlabeled data, and achieved an average accuracy close to 91% after fine-tuning with minimal labeled data.

关键词： automatic data augmentation Contrastive learning Lightweight encoder network Regional defect density Defect pattern recognition

来源：评论

学校读者我要写书评

暂无评论

ALADA: A lite automatic data augmentation framework for industrial defect detection

引用

ADVANCED ENGINEERING INFORMATICS 2023年 58卷

作者： Wang, Yuxuan Chung, Sai-Ho Khan, Waqar Ahmed Wang, Tianteng Xu, David Jingjun Hong Kong Polytech Univ Dept Ind & Syst Engn Hung Hom Hong Kong Peoples R China Univ Sharjah Coll Engn Dept Ind Engn & Engn Management Sharjah U Arab Emirates City Univ Hong Kong Dept Informat Syst Hong Kong Peoples R China

Industrial defect detection is a critical and challenging task in the quality control of manufacturing production. Competent in feature extraction and pattern recognition, deep learning shows great power for classifying and locating defects in industrial products. However, insufficient data records and diverse categories of defects restrict the detection accuracy of data-driven neural networks. One direction to solve such a problem is data augmentation, which aims at generating synthetic copies from existing data and improving the generalizability of detection models. However, confirming a suitable augmentation policy involves either human experience or substantial experiments for augmentation parameters. To address these challenges, in this work, a lite automatic data augmentation (ALADA) framework is proposed to jointly optimize the data augmentation policies and the neural network for industrial defect detection. First, a lite search space is formulated to efficiently sample augmentation policies and generate augmented images for joint optimization. To reduce the hyperparameter tuning efforts for retraining with searched policies, a three-step bi-level optimization scheme is proposed to replace the retraining strategy and update the model and augmentation parameter alternately. To solve the non-differentiable problem in the joint optimization scheme, policy gradient sampling is implemented to estimate the gradient flow efficiently. Experimental results on three industrial defect detection datasets, namely, Tianchi-TILE, GC10-DET, and NEU-DET, reveal that our proposed automatic augmentation framework outperforms the state-of-the-art augmentation methods and effectively improves the accuracy of the baseline defect detection model. The proposed ALADA scheme also alleviates the missed detection of defects in four practical industrial circumstances: textured background, uneven brightness, low contrast, and intraclass difference.

关键词： Industrial defect detection automatic data augmentation Deep learning Optimization

来源：评论

学校读者我要写书评

暂无评论

Impartial Differentiable automatic data augmentation Based on Finite Difference Approximation for Pedestrian Detection

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2022年 71卷 1页

作者： Zhou, Shirui Tang, Yi Liu, Min Wang, Yaonan Wen, He Hunan Univ Coll Elect & Informat Engn Natl Engn Res Ctr Robot Visual Percept & Control Changsha 410082 Peoples R China

Pedestrian detection is a challenging task in practical scenarios. data augmentation is an effective way to deal with this problem, but designing a suitable data augmentation policy requires a great deal of manual experience and multiple adjustments of parameters. To address this problem, we propose an impartial differentiable automatic data augmentation (IDADA) that automatically searches a suitable augmentation policy for pedestrian detection in a more practical scenario. First, an impartiality loss is proposed to solve the bias of the operations probabilities distribution, so as to improve the impartiality for each operation. Second, the impartiality loss and the original validation loss of the detection model are combined as a target function to update the augmentation policy parameters. Since the target function is not differentiable, the finite difference approximation is applied to provide accurate gradient estimation for the augmentation policy optimization process. Comprehensive experiments performed on two typical datasets of CrowdHuman and CityPersons verify the effectiveness of the proposed approach. Specifically, our approach achieves 40.99% in miss rate (MR) on CrowdHuman, yielding the new state-of-the-art result on this dataset.

关键词： Optimization Training Estimation Task analysis Finite difference methods Search problems Encoding automatic data augmentation differentiable optimization pedestrian detection

来源：评论

学校读者我要写书评

暂无评论

An online adaptive augmentation strategy for cervical cytopathology image recognition

引用

PATTERN RECOGNITION LETTERS 2025年 192卷 93-98页

作者： Tang, Hongmei Liu, Xinyi Cheng, Shenghua Liu, Xiuli Huazhong Univ Sci & Technol Britton Chance Ctr Wuhan Peoples R China Huazhong Univ Sci & Technol MoE Key Lab Biomed Photon Wuhan Natl Lab Optoelect Wuhan Peoples R China Southern Med Univ Sch Biomed Engn Guangzhou Peoples R China Southern Med Univ Guangdong Prov Key Lab Med Image Proc Guangzhou Peoples R China

Cervical cancer has become one of the malignant tumors among women, posing a significant threat to women's health worldwide. Efficient computer-aided screening techniques are extremely significant for popularizing cervical cancer screening and reducing the mortality rate. However, in practical, applications, variations in staining styles and image quality of multi-center digital pathological smears pose challenges for current cervical cell recognition deep learning-based methods. In this paper, we propose an online adaptive data augmentation strategy : a simple, search-free, and adaptive method designed to improve model generalization. We transform the process of selecting the optimal augmentation strategy into the determination of probabilities assigned to each operation. The core of the method lies in directly optimize the probability distribution of the entire augmentation space based on the model's performance on the validation set. Extensive experiments on the multi-center cervical cytopathology image datasets demonstrate that our method outperforms the state-of-theart automatic augmentation methods in model generalization evaluation. We achieve an average accuracy of 87.80% on five external test sets, with only a 6.85% difference from internal test accuracy. Our work contributes to enhancing generalization of cervical cell recognition methods in multi-center scenarios.

关键词： Cervical cytopathology image recognition automatic data augmentation Model generalization Multi-center cytopathology images

来源：评论

学校读者我要写书评

暂无评论

STACoRe: Spatio-temporal and action-based contrastive representations for reinforcement learning in Atari

引用

NEURAL NETWORKS 2023年第1期160卷 1-11页

作者： Lee, Young Jae Kim, Jaehoon Kwak, Mingu Park, Young Joon Kim, Seoung Bum Korea Univ Sch Ind & Management Engn Seoul South Korea Georgia Inst Technol Sch Ind & Syst Engn Atlanta GA USA LG AI Res Seoul South Korea

With the development of deep learning technology, deep reinforcement learning (DRL) has successfully built intelligent agents in sequential decision-making problems through interaction with image-based environments. However, learning from unlimited interaction is impractical and sample inefficient because training an agent requires many trial and error and numerous samples. One response to this problem is sample-efficient DRL, a research area that encourages learning effective state representations in limited interactions with image-based environments. Previous methods could effectively surpass human performance by training an RL agent using self-supervised learning and data augmentation to learn good state representations from a given interaction. However, most of the existing methods only consider similarity of image observations so that they are hard to capture semantic representations. To address these challenges, we propose spatio-temporal and action-based contrastive representation (STACoRe) learning for sample-efficient DRL. STACoRe performs two contrastive learning to learn proper state representations. One uses the agent's actions as pseudo labels, and the other uses spatio-temporal information. In particular, when performing the action-based contrastive learning, we propose a method that automatically selects data augmentation techniques suitable for each environment for stable model training. We train the model by simultaneously optimizing an action-based contrastive loss function and spatio-temporal contrastive loss functions in an end-to-end manner. This leads to improving sample efficiency for DRL. We use 26 benchmark games in Atari 2600 whose environment interaction is limited to only 100k steps. The experimental results confirm that our method is more sample efficient than existing methods. The code is available at https://***/dudwojae/STACoRe. (c) 2022 Elsevier Ltd. All rights reserved.

关键词： Atari automatic data augmentation End-to-end learning Reinforcement learning Supervised contrastive learning Spatio-temporal contrastive learning

来源：评论

学校读者我要写书评

暂无评论

Evolutionary Neural Architecture Search for 2D and 3D Medical Image Classification 24th

Evolutionary Neural Architecture Search for 2D and 3D Medica...

引用

24th International Conference on Computational Science (ICCS)

作者： Ali, Muhammad Junaid Moalic, Laurent Essaid, Mokhtar Idoumghar, Lhassane Univ Haute Alsace IRIMAS UR 7499 F-68093 Mulhouse France

ISBN: (纸本)9783031637537;9783031637513

Designing deep learning architectures is a challenging and time-consuming task. To address this problem, Neural Architecture Search (NAS) which automatically searches for a network topology is used. While existing NAS methods mainly focus on image classification tasks, particularly 2D medical images, this study presents an evolutionary NAS approach for 2D and 3D Medical image classification. We defined two different search spaces for 2D and 3D datasets and performed a comparative study of different meta-heuristics used in different NAS studies. Moreover, zero-cost proxies have been used to evaluate the performance of deep neural networks, which helps reduce the searching cost of the overall approach. Furthermore, recognizing the importance of data augmentation (DA) in model generalization, we propose a genetic algorithm based automatic DA strategy to find the optimal DA policy. Experiments on MedMNIST benchmark and BreakHIS dataset demonstrate the effectiveness of our approach, showcasing competitive results compared to existing AutoML approaches. The source code of our proposed approach is available at https://***/Junaid199f/evo_nas_med_2d_3d.

关键词： Evolutionary Neural Architecture Search Medical Image Classification AutoML AutoDL automatic data augmentation

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：