Self-supervised Learning in Computer Vision: A Review

The availability of huge annotated datasets and deep learning have led to astounding Natural Language Processing (NLP) progress. However, there is a plethora of low-resource languages, lack the labelled corpora to train the models in the supervision. This kind of digital divide does not allow fair NLP development across language groups. Self-supervised learning (SSL) is a new paradigm in recent years that leverages large amounts of untagged text to learn powerful representations with little or no manual annotation. This paper is a review of self-supervised deep learning models in low-resource NLP tasks. It begins with the definition of the principles of the SSL and the difference between this approach and supervised and unsupervised approaches. We describe why methods such as masked language modeling, contrastive learning, and autoregressive modeling underlie modern pre-trained transformers such as BERT, GPT, and mBERT. A particular interest is paid to multilingual and cross lingual SSL schemes which allow knowledge transfer between high-resource and low resource languages. Low-resource tasks, including machine translation, sentiment analysis, speech-to-text, and information retrieval are reviewed. As per benchmark studies, even in small labelled samples, the SSL models can achieve big gains in accuracy and generalization. Computation cost, representational bias and morphologically rich and under-documented language evaluation, are however, problematic. The researchers observe that the study suffers certain limitations, such as over-reliance on high-resource pretraining information, inequity between linguistic groups, and the difficulty of deploying large-scale SSL models in resource-constrained circumstances. The future directions include lightweight multilingual models, federation of learning in NLP and symbolic linguistic knowledge and combination with SSL. Self-supervised deep learning bridges an essential gap between the high-resource and low-resource languages, and would be a highly valuable step toward inclusive, global NLP innovation.

Review on self-supervised image recognition using deep neural networks

Spatiotemporal self-supervised predictive learning for atmospheric variable prediction via multi-group multi-attention

Computer vision systems are increasingly expanding their application in visual data analysis. Model training methods are undergoing the greatest development and improvement as the results of this stage significantly affect the final classification of objects and the interpretation of input information. Typically, computer vision systems use convolutional neural networks for training (Convolution Neural Network, CNN). The disadvantages of such systems are significant limitations in cross-modal learning, multimodality implementation, labeling of large amounts of data, etc. One of the ways to overcome these problems is to use Vision Transformers (ViT), which, compared to classical CNNs, have higher performance due to reduced inductive biases and high parallel computing efficiency. Introducing Self-Supervised Learning (SSL) technologies can significantly reduce the dependence on manually labeled data, contributing to the formation of generalized representations of images. Cross-Modal Learning (CML) expands the possibilities of processing them by combining data of different types. The development of the new approach, combined with the capabilities of cross-modal learning and self-learning in ViT in a single architecture, will ensure adaptability, efficiency, and system scalability in various applications. The research aims to provide a detailed overview of ViTs, approaches to their architecture, and methods for improving their efficiency. The mathematical foundations of the key concepts of ViT, cross-modal learning and self-learning, the main modifications of ViT, and their integration with SSL and CML technologies are considered. A comparison of methods using characteristics, performance, and efficiency is provided. The key challenges and prospects facing researchers and developers while creating universal models in computer vision are outlined. ViTs change computer vision by capturing global dependencies on images. Despite some challenges, ViTs provide excellent scalability and performance for large datasets. The active search for methods to overcome their limitations makes ViTs a key tool for improving image classification, object detection, and other computer vision tasks.

The application of deep learning has resulted in significant advancements in computer vision systems; nonetheless, extensive labelled datasets and processing resources are required. In a lot of different domains, particularly emerging countries and edge-based deployments, there is a lack of labelled data, annotation is expensive, and computational resources is limited. An innovative Self-Supervised Learning (SSL) architecture has been developed as a result of this research. It is designed for low-resource computer vision applications that have limited labelled data and computational resources. Using unlabelled data, pretext tasks, representation consistency learning, and lightweight feature encoders, the proposed method is able to learn strong visual representations without the need for manual annotation. Under low-label regimes, the suggested method achieves better results than the usual supervised and transfer learning baselines when applied to benchmark datasets. The use of self-supervised learning makes it possible to create reliable computer vision systems in situations where resources are limited, and it is also scalable.

Whether tracking patient progress for clinical decision making, or investigating novel therapies, automated analysis of Computed Tomography (CT) imaging data is essential for the future of digital radiomics. In digital radiomics, as in all medical imaging, well annotated data is scarce, whereas unlabelled images are relatively plentiful. In other fields of image processing and medical imaging, the application of self-supervised learning (SSL) to large quantities of unlabelled data has resulted in great strides forward for fast, scalable, and interpretable image analysis. In this work we present a new approach applying SSL to CT imaging data which allows for: 1) Improved performance on image classification tasks, based on 2) dramatically reduced quantity of annotated CT imaging data, whilst also 3) enabling easy exploration and interpretation of the image regions. We applied a selection of self-supervised approaches (BYOL, DINO, SimCLR, & inpainting) to CT imaging data. Because the high dimensionality of CT data prevents us from using them directly to out-of the box SSL models, we adopt a 3D patching approach to reduce the dimensionality of the neural net input, and process each patch independently. We train our self-supervised models on public datasets (DeepLesion, NSCLC), and we specialize these models for tumor classification tasks that we evaluate on AstraZeneca sponsored clinical trials. The specialization is done in two different ways: 1) we use the pre-trained SSL model as an encoder that transforms the images of the clinical study into embeddings, on which we apply supervised classification models; 2) we use transfer learning to fine-tune supervised classification models that take the patches directly as inputs. We find that self-supervised pre-training significantly improves the accuracy on tumor classification tasks compared against a supervised learning baseline. Additionally, using the SSL embeddings we build an interactive map of CT imaging data enabling quick and intuitive inspection of the relevant regions. Our findings show that SSL constitutes an important tool for medical imaging analysis. SSL results in models that generalize better, and enable improved downstream interpretability and predictions. Furthermore, well trained SSL models can be re-applied to multiple indications because they are pre-trained on broad and diverse CT imaging data. Citation Format: Leon Fedden, Zhenning Zhang, Khan Baykaner, Qin Li, Lucas Bordeaux. DIME-CT: Self-supervised learning for medical image analysis using patch-based embeddings [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1937.

Anomaly classification based on self-supervised learning and its application

The implementation of computer vision generally focuses on resolving a problem of one of two types of image data, whether it was natural image datasets or medical image datasets. Self-supervised learning (SSL), as the most recent approach in computer vision, is typically used to solved tasks related only to natural image datasets such as ImageNet, CIFAR-10, and PASCAL. This study focuses on analyzing the involvement of natural images in the SSL implementation when classifying medical images. We used the ResNet50 architecture comparing the results of human embryo image classification using the conventional transfer learning method and SSL methods. We apply three SSL approaches, namely Rotation, Jigsaw Puzzle, and SimCLRv1. The results of our research, with 1226 embryo images divided into three classes, show that the SSL implementations cannot be said to be efficient enough compared to the conventional model produced by ResNet50 with pretrained weight ImageNet. However, when the models were trained using random weight initialization, the Rotation model with an average accuracy of 78.15% and the Jigsaw Puzzle model with an average accuracy of 80.30% are significantly superior to the conventional model with average accuracy is 71.07%. Moreover, related to our embryos dataset, we discuss the differences between the SSL context-context contrastive learning approach and the context instance contrastive learning approach.

Self-supervised learning (SSL) represents a revolutionary AI paradigm which lets machines acquire significant data representations directly from unlabeled information through unsupervised learning approaches. SSL uses contrastive learning and masked data modeling and predictive learning approaches to optimize data efficiency thereby improving model generalization between multiple domains. This paper evaluates the core concepts of SSL alongside its superiority to supervised and unsupervised learning and its usage in different fields such as NLP, computer vision, speech recognition, healthcare, finance and robotics. The paper focuses on analysis of essential techniques and architectures which include SimCLR, MoCo, BERT, MAE, BYOL and approaches combining SSL with reinforcement learning and weak supervision methods. The research analyzes SSL's current challenges including operational expenses and representation degeneration as well as the assessment obstructions while proposing future uses for the method in mixed-data learning and minimal-resource contexts and artificial general intelligence (AGI). The adoption of SSL in real-world AI applications depends on effectively dealing with ethical matters that include bias issues and responsible AI practices and fairness assurance.

The artificial intelligence (AI) community has recently made tremendous progress in developing self-supervised learning (SSL) algorithms that can learn high-quality data representations from massive amounts of unlabeled data. These methods brought great results even to the fields outside of AI. Due to the joint efforts of researchers in various areas, new SSL methods come out daily. However, such a sheer number of publications make it difficult for beginners to see clearly how the subject progresses. This survey bridges this gap by carefully selecting a small portion of papers that we believe are milestones or essential work. We see these researches as the “dots” of SSL and connect them through how they evolve. Hopefully, by viewing the connections of these dots, readers will have a high-level picture of the development of SSL across multiple disciplines including natural language processing, computer vision, graph learning, audio processing, and protein learning.

This paper proposes a novel self-supervised learning framework that uses model reinforcement, REference-free LAtent map eXtraction with MOdel REinforcement (RELAX-MORE), for accelerated quantitative MRI (qMRI) reconstruction. The proposed method uses an optimization algorithm to unroll an iterative model-based qMRI reconstruction into a deep learning framework, enabling accelerated MR parameter maps that are highly accurate and robust. Unlike conventional deep learning methods which require large amounts of training data, RELAX-MORE is a subject-specific method that can be trained on single-subject data through self-supervised learning, making it accessible and practically applicable to many qMRI studies. Using quantitative mapping as an example, the proposed method was applied to the brain, knee and phantom data. The proposed method generates high-quality MR parameter maps that correct for image artifacts, removes noise, and recovers image features in regions of imperfect image conditions. Compared with other state-of-the-art conventional and deep learning methods, RELAX-MORE significantly improves efficiency, accuracy, robustness, and generalizability for rapid MR parameter mapping. This work demonstrates the feasibility of a new self-supervised learning method for rapid MR parameter mapping, that is readily adaptable to the clinical translation of qMRI.

A limited amount of training data is a well-known challenge in the application of deep learning methods. This issue is particularly relevant in biomedical signal processing, such as the analysis of electrocardiograms, due to the labor-intensive nature of data preparation, which requires the involvement of qualified specialists. Self-supervised learning methods, originally developed in such domains as natural language processing and computer vision, have emerged as a potential approach to addressing this challenge and are increasingly being explored in biomedical signal processing. However, direct adaptation of self-supervised learning techniques from other domains does not fully account for ECG-specific characteristics, such as quasi-periodicity, localized morphological features, and susceptibility to noise. This highlights the relevance of developing ECG-specific self-supervised learning methods. This study presents a novel self-supervised learning approach for pretraining neural networks on unlabeled ECG data. The proposed method is based on predicting the short consecutive signal segment using a preceding one and a learned representation vector. The representation extraction and prediction models are trained jointly on the MIMIC-ECG-IV dataset using backpropagation to minimize the mean squared error between the predicted and original signal segments. As an example of a downstream task, a linear binary classifier was trained on the PTB-XL dataset to diagnose pathological conditions using Lead I. The number of training examples for each diagnosis was limited to thirty-four samples. Firstly, the representation model was pre-trained on the unlabeled MIMIC-ECG-IV dataset, and then linear classifiers were trained on the learned representations for each selected diagnosis in PTB-XL. A comparison was also conducted with a randomly initialized representation model trained jointly with the classifier in a fully supervised manner. The proposed method was evaluated against adaptations of Contrastive Learning, Contrastive Predictive Coding, and Masked Autoencoders method. To ensure a controlled experimental setup, implementations of all considered methods were developed using a unified codebase and shared architectural components. Experimental results demonstrated a significant advantage of all self-supervised learning approaches over joint training of feature extraction and classification models. The proposed SSL method outperformed other tested approaches, particularly for diagnoses with subtle short-term morphological features, such as atrial fibrillation and flutter. These findings suggest the potential for further research in developing ECG-specific self-supervised learning methods as a promising approach to improving neural network performance in scenarios with limited labeled data.

Recent studies in self-supervised learning have shown encouraging results that fill the gap between supervised and self-supervised learning in computer vision. These methods are evaluated using linear evaluation protocol, which is a standard benchmark to measure the quality of representations inferred from self-supervised models. On the other hand, visual similarity search without supervision is very much less studied. With the amount of unlabeled data grows every day on the Internet (e.g., e-commerce platforms), it is only natural to embrace self-supervised methods for image retrieval task. In this paper, we present our evaluation of self-supervised methods for visual similarity search on e-commerce dataset. Our experiments show that while self-supervised learning have not yet matched the performance of supervised learning, it is capable to extract necessary features from images to perform similarity search. We present the results on both linear evaluation and semi-supervised settings. We find that the use of contrastive loss function during self-supervised pretraining creates better quality of representations than other methods.

Recent advances in self-supervised learning (SSL) in computer vision are primarily comparative, whose goal is to preserve invariant and discriminative semantics in latent representations by comparing siamese image views. However, the preserved high-level semantics do not contain enough local information, which is vital in medical image analysis (e.g., image-based diagnosis and tumor segmentation). To mitigate the locality problem of comparative SSL, we propose to incorporate the task of pixel restoration for explicitly encoding more pixel-level information into high-level semantics. We also address the preservation of scale information, a powerful tool in aiding image understanding but has not drawn much attention in SSL. The resulting framework can be formulated as a multi-task optimization problem on the feature pyramid. Specifically, we conduct multi-scale pixel restoration and siamese feature comparison in the pyramid. In addition, we propose non-skip U-Net to build the feature pyramid and develop sub-crop to replace multi-crop in 3D medical imaging. The proposed unified SSL framework (PCRLv2) surpasses its self-supervised counterparts on various tasks, including brain tumor segmentation (BraTS 2018), chest pathology identification (ChestX-ray, CheXpert), pulmonary nodule detection (LUNA), and abdominal organ segmentation (LiTS), sometimes outperforming them by large margins with limited annotations. Codes and models are available at https://github.com/RL4M/PCRLv2.

To make Graph Neural Networks (GNNs) meet the requirements of the Web, the universality and the generalization become two important research directions. On one hand, many universal GNNs are presented for semi-supervised tasks on both homophilic and non-homophilic graphs by distinguishing homophilic and heterophilic edges with the help of labels. On the other hand, self-supervised learning (SSL) algorithms on graphs are presented by leveraging the self-supervised learning schemes from computer vision and natural language processing. Unfortunately, graph universal self-supervised learning remains resolved. Most existing SSL methods on graphs, which often employ two-layer GCN as the encoder and train the mapping functions, can't alter the low-passing filtering characteristic of GCN. Therefore, to be universal, SSL must becustomized for the graph, i.e., learning the graph. However, learning the graph via universal GNNs is disabled in SSL, since their distinguishability on homophilic and heterophilic edges disappears without the labels. To overcome this difficulty, this paper proposes novel GrAph-customized Universal Self-Supervised Learning (GAUSS) by exploiting local attribute distribution. The main idea is to replace the global parameters with locally learnable propagation. To make the propagation matrix demonstrate the affinity between the nodes, the self-representative learning framework is employed with k-block diagonal regularization. Extensive experiments on synthetic and real-world datasets demonstrate its effectiveness, universality and robustness to noises.