Unsupervised Continual Learning Via Pseudo Labels

By Jiangpeng He, Feng Zhu. 2021

32 citations recorded by the source corpus.

Continual learning aims to learn new tasks incrementally using less computation and memory resources instead of retraining the model from scratch whenever new task arrives. However, existing approaches are designed in supervised fashion assuming all data from new tasks have been manually annotated, which are not practical for many real-life applications. In this work, we propose to use pseudo label instead of the ground truth to make continual learning feasible in unsupervised mode. The pseudo labels of new data are obtained by applying global clustering algorithm and we propose to use the model updated from last incremental step as the feature extractor. Due to the scarcity of existing work, we introduce a new benchmark experimental protocol for unsupervised continual learning of image classification task under class-incremental setting where no class label is provided for each incremental learning step. Our method is evaluated on the CIFAR-100 and ImageNet (ILSVRC) datasets by incorporating the pseudo label with various existing supervised approaches and show promising results in unsupervised scenario.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Feng Zhu
2. Jiangpeng He
3. Deepak Pathak
4. C. Dovrolis
5. Deva Ramanan
6. Jia Shi
7. Cameron Taylor
8. Zhiqiu Lin

In Computer Vision and Pattern Recognition

1. Feng Zhu
2. Deepak Pathak
3. Luc Van Gool
4. Jiangpeng He
5. Trevor Darrell
6. Deva Ramanan
7. Jitendra Malik
8. Ming-Hsuan Yang

In Radiology, Nuclear Medicine and Imaging

1. Feng Zhu
2. Jiangpeng He
3. C. Dovrolis
4. Deva Ramanan

Citations in the corpus

• Online Continual Learning Via Candidates Voting (2021, background): “Our future work will focus on unsupervised continual learning, which is more realistic and one possible solution is to use pseudo label as recently introduced in [15].”
• Unsupervised Progressive Learning and the STAM Architecture (2019, background): “[ He and Zhu, 2021 ] investigate a similar setting where pseudo-labels are assigned to new data based on cluster assignments but assumes knowledge of the number of classes per task and class boundaries.”
• Queried Unlabeled Data Improves and Robustifies Class-Incremental Learning (2022, unknown): “…(Castro et al., 2018; He et al., 2018; Javed & Shafait, 2018; Rebuffi et al., 2017; Belouadah & Popescu, 2019; 2020) On the other hand, recent investigations (Wang et al., 2021; He & Zhu, 2021) also explore other alternative possibilities of leveraging unlabelled data to boost continual learning.”
• The CLEAR Benchmark: Continual LEArning on Real-World Imagery (2022, background): “Besides high-quality labeled data, the rest of the images per bucket in CLEAR can be viewed as large-scale free unlabeled data, which we hope will spur future research on continual semi-supervised learning [20, 45, 52, 58].”
• Online Continual Learning For Visual Food Classification (2021, background): “Knowledge distillation [13] is a popular representative technique, which makes the model mimic the output distribution for learned classes from a teacher model to mitigate forgetting during continual learning [21, 35, 4, 42, 15, 20, 11].”