A Survey on Semi-, Self- and Unsupervised Learning for Image Classification

Abstract

While deep learning strategies achieve outstanding results in computer vision tasks, one issue remains: The current strategies rely heavily on a huge amount of labeled data. In many real-world problems, it is not feasible to create such an amount of labeled training data. Therefore, it is common to incorporate unlabeled data into the training process to reach equal results with fewer labels. Due to a lot of concurrent research, it is difficult to keep track of recent developments. In this survey, we provide an overview of often used ideas and methods in image classification with fewer labels. We compare 34 methods in detail based on their performance and their commonly used ideas rather than a fine-grained taxonomy. In our analysis, we identify three major trends that lead to future research opportunities. 1. State-of-the-art methods are scaleable to real-world applications in theory but issues like class imbalance, robustness, or fuzzy labels are not considered. 2. The degree of supervision which is needed to achieve comparable results to the usage of all labels is decreasing and therefore methods need to be extended to settings with a variable number of classes. 3. All methods share some common ideas but we identify clusters of methods that do not share many ideas. We show that combining ideas from different clusters can lead to better performance.

Uncertainty-Aware Deep Co-training for Semi-supervised Medical Image Segmentation

“…They trained different models with corresponding subsets of the annotated data and leveraged non-annotated images to exchange internal information among sub-models. The difference between these semisupervised methods is how they utilize unlabeled data and the way they relate to supervised algorithms [5]. Most of the approaches utilize predictions of unlabeled data for consistency training [21].…”

“…It is time-consuming and extremely relies on experienced experts. Fortunately, semi-supervised methods leverage not only labeled data but also unlabeled data [5], freeing the researchers from labeling work. Therefore, semi-supervised learning has attracted great attention [6], [7], [8].…”

Self-Supervised Joint Learning Fault Diagnosis Method Based on Three-Channel Vibration Images

“…To address the problem of data annotation and improve the learning of unlabeled data, numerous self-supervised learning methods have been recently proposed and summarized [13][14][15][16]. The basis of self-supervised learning is the acquisition of pseudo-tags with unlabeled data via the setting of auxiliary tasks.…”

“…The basis of self-supervised learning is the acquisition of pseudo-tags with unlabeled data via the setting of auxiliary tasks. These pseudo-tags do not artificial annotation and can be generated using the image or video attributes; these false tags are then used to learn the characteristics of untagged data [13,16]. Self-supervised learning can be divided into many forms according to the research goal of the auxiliary tasks [14], including generative grammar, contrastive grammar, and generative grammar.…”

Clustering COVID-19 Lung Scans

“…Clustering is an essential data analysis task where objects are placed into groups based on similarity of features. Popular supervised classification models would likely yield more accurate results, but implementing supervised learning requires labeled data in every application of a given method-a potential limiting factor for novel data [12,20]. Creating an unsupervised model with high accuracy would be a significant accomplishment because of the greater range of real-world applicability the model would have.…”