6 Hits in 5.2 sec

Self-Supervised Learning for 3D Medical Image Analysis using 3D SimCLR and Monte Carlo Dropout [article]

Yamen Ali, Aiham Taleb, Marina M. -C. Höhne, Christoph Lippert
2021 arXiv   pre-print
Self-supervised learning methods can be used to learn meaningful representations from unlabeled data that can be transferred to supervised downstream tasks to reduce the need for labeled data.  ...  In this paper, we propose a 3D self-supervised method that is based on the contrastive (SimCLR) method.  ...  Introduction As 3D medical imaging became an essential tool in medicine, the need for accurate and reliable machine learning algorithms that analyze such images has become more apparent.  ... 
arXiv:2109.14288v2 fatcat:fwp4bsfxzfcahorzgxs7xpfbym

BYOLMed3D: Self-Supervised Representation Learning of Medical Videos using Gradient Accumulation Assisted 3D BYOL Framework [article]

Siladittya Manna, Rakesh Dey, Souvik Chakraborty
2022 arXiv   pre-print
Applications on Medical Image Analysis suffer from acute shortage of large volume of data properly annotated by medical experts.  ...  In this work, we train a 3D BYOL self-supervised model using gradient accumulation technique to deal with the large number of samples in a batch generally required in a self-supervised algorithm.  ...  In the domain of medical images, [8] uses a 3D self-supervised network based on contrastive SimCLR [3] and Monte Carlo dropout to enhance results in downstream task.  ... 
arXiv:2208.00444v2 fatcat:gwpwtrkxybfnharyfnyaemyday

Frequency Dropout: Feature-Level Regularization via Randomized Filtering [article]

Mobarakol Islam, Ben Glocker
2022 arXiv   pre-print
vision and medical imaging datasets.  ...  Our training strategy is model-agnostic and can be used for any computer vision task.  ...  This project has received funding from the European Research Council (ERC under the European Union's Horizon 2020 research and innovation programme (Grant Agreement No. 757173, Project MIRA).  ... 
arXiv:2209.09844v1 fatcat:szqzg6pkbnemfgrdekazlvvhdm

Deep Active Learning for Computer Vision Tasks: Methodologies, Applications, and Challenges

Mingfei Wu, Chen Li, Zehuan Yao
2022 Applied Sciences  
Deep active learning plays a crucial role in computer vision tasks, especially in label-insensitive scenarios, such as hard-to-label tasks (medical images analysis) and time-consuming tasks (autonomous  ...  Active learning focuses on achieving the best possible performance while using as few, high-quality sample annotations as possible.  ...  A survey on active learning and humanin-the-loop deep learning for medical image analysis [6] Investigate the active learning in the medical image analysis.  ... 
doi:10.3390/app12168103 fatcat:v5iovucmkvamtbezp2bpztxynq

A Survey on Deep Semi-supervised Learning [article]

Xiangli Yang, Zixing Song, Irwin King, Zenglin Xu
2021 arXiv   pre-print
This paper provides a comprehensive survey on both fundamentals and recent advances in deep semi-supervised learning methods from perspectives of model design and unsupervised loss functions.  ...  We first present a taxonomy for deep semi-supervised learning that categorizes existing methods, including deep generative methods, consistency regularization methods, graph-based methods, pseudo-labeling  ...  For example, in medical tasks, the measurements are made with expensive machinery, and labels are drawn from a time-consuming analysis of multiple human experts.  ... 
arXiv:2103.00550v2 fatcat:lymncf5wavgkhaenbvqlyvhuaa

An Overview of Neural Network Compression [article]

James O' Neill
2020 arXiv   pre-print
Pushing state of the art on salient tasks within these domains corresponds to these models becoming larger and more difficult for machine learning practitioners to use given the increasing memory and storage  ...  We assume a basic familiarity with deep learning architectures[%s], namely, Recurrent Neural Networks , Convolutional Neural Networks [%s] and Self-Attention based networks [%s],[%s].  ...  They note that L E is intractable for noisy weights and in practice Monte Carlo integration is used.  ... 
arXiv:2006.03669v2 fatcat:u2p6gvwhobh53hfjxawzclw7fq