Cold PAWS: Unsupervised class discovery and addressing the cold-start problem for semi-supervised learning

17 May 2023  ยท  Evelyn J. Mannix, Howard D. Bondell ยท

In many machine learning applications, labeling datasets can be an arduous and time-consuming task. Although research has shown that semi-supervised learning techniques can achieve high accuracy with very few labels within the field of computer vision, little attention has been given to how images within a dataset should be selected for labeling. In this paper, we propose a novel approach based on well-established self-supervised learning, clustering, and manifold learning techniques that address this challenge of selecting an informative image subset to label in the first instance, which is known as the cold-start or unsupervised selective labelling problem. We test our approach using several publicly available datasets, namely CIFAR10, Imagenette, DeepWeeds, and EuroSAT, and observe improved performance with both supervised and semi-supervised learning strategies when our label selection strategy is used, in comparison to random sampling. We also obtain superior performance for the datasets considered with a much simpler approach compared to other methods in the literature.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Semi-Supervised Image Classification (Cold Start) CIFAR-10, 100 Labels SimCLR-kmediods-PAWS Percentage error 6.1 # 1
Semi-Supervised Image Classification CIFAR-10, 100 Labels SimCLR-kmediods-PAWS Percentage error 6.1 # 1
Semi-Supervised Image Classification CIFAR-10, 30 Labels SimCLR-kmediods-PAWS Percentage error 6.4 # 1
Semi-Supervised Image Classification (Cold Start) CIFAR-10, 30 Labels SimCLR-kmediods-PAWS Percentage error 6.4 # 1
Semi-Supervised Image Classification (Cold Start) DeepWeeds, 99 Labels SimCLR-kmediods-finetuned Percentage error 19.6 # 1
Semi-Supervised Image Classification DeepWeeds, 99 Labels SimCLR-kmediods-finetuned Percentage error 19.6 # 1
Semi-Supervised Image Classification EuroSAT, 100 Labels SimCLR-kmediods-PAWS Percentage error 2.6 # 1
Semi-Supervised Image Classification (Cold Start) EuroSAT, 100 Labels SimCLR-kmediods-PAWS Percentage error 2.6 # 1
Semi-Supervised Image Classification EuroSAT, 20 Labels SimCLR-kmediods-PAWS Percentage error 3.8 # 1
Semi-Supervised Image Classification (Cold Start) EuroSAT, 20 Labels SimCLR-kmediods-PAWS Percentage error 3.8 # 1
Semi-Supervised Image Classification (Cold Start) Imagenette, 100 Labels SimCLR-kmediods-PAWS Percentage error 6.1 # 1
Semi-Supervised Image Classification Imagenette, 100 Labels SimCLR-kmediods-PAWS Percentage error 6.1 # 1
Semi-Supervised Image Classification Imagenette, 20 Labels SimCLR-kmediods-PAWS Percentage error 10.8 # 1
Semi-Supervised Image Classification (Cold Start) Imagenette, 20 Labels SimCLR-kmediods-PAWS Percentage error 10.8 # 1

Methods