Unbiased and Efficient Self-Supervised Incremental Contrastive Learning
Not on the map. Why?
Abstract (arXiv)
Contrastive Learning (CL) has been proved to be a powerful self-supervised approach for a wide range of domains, including computer vision and graph representation learning. However, the incremental learning issue of CL has rarely been studied, which brings the limitation in applying it to real-world applications. Contrastive learning identifies the samples with the negative ones from the noise distribution that changes in the incremental scenarios. Therefore, only fitting the change of data without noise distribution causes bias, and directly retraining results in low efficiency. To bridge this research gap, we propose a self-supervised Incremental Contrastive Learning (ICL) framework consisting of (i) a novel Incremental InfoNCE (NCE-II) loss function by estimating the change of noise distribution for old data to guarantee no bias with respect to the retraining, (ii) a meta-optimization with deep reinforced Learning Rate Learning (LRL) mechanism which can adaptively learn the learning rate according to the status of the training processes and achieve fast convergence which is critical for incremental learning. Theoretically, the proposed ICL is equivalent to retraining, which is based on solid mathematical derivation. In practice, extensive experiments in different domains demonstrate that, without retraining a new model, ICL achieves up to 16.7x training speedup and 16.8x faster convergence with competitive results.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 27 of its 48 references that are in the map. First posted on arXiv 28 Jan 2023.
Builds on
27 of 48 references.
References · 27
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 12,049
- 2016Semi-Supervised Classification with Graph Convolutional Networkscited by 11,361
- 2018Representation Learning with Contrastive Predictive Codingcited by 7,777
- 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,513
- 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
- 2015Continuous control with deep reinforcement learningcited by 3,407
- 2013Playing Atari with Deep Reinforcement Learningcited by 3,274
- 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,333
- 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,845
- 2020Graph Contrastive Learning with Augmentationscited by 1,208
- 2020TUDataset: A collection of benchmark datasets for learning with graphscited by 728
15 more
- 2020Self-supervised Learning: Generative or Contrastivecited by 694
- 2018Overcoming catastrophic forgetting with hard attention to the taskcited by 681
- 2018End-to-End Incremental Learningcited by 597
- 2020A Survey on Contrastive Self-supervised Learningcited by 554
- 2020GCC: Graph Contrastive Coding for Graph Neural Network Pre-Trainingcited by 463
- 2021Co2L: Contrastive Continual Learningcited by 244
- 2018An efficient framework for learning sentence representationscited by 198
- 2021Supervised Contrastive Replay: Revisiting the Nearest Class Mean Classifier in Online Class-Incremental Continual Learningcited by 141
- 2021SEED: Self-supervised Distillation For Visual Representationcited by 134
- 2021Graph Structure Learning with Variational Information Bottleneckcited by 81
- 2021SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanismcited by 75
- 2017Reinforcement Learning for Learning Rate Controlcited by 16
- 2021A Survey on Deep Learning Event Extraction: Approaches and Applicationscited by 12
- 2021Reinforcement Learning-based Dialogue Guided Event Extraction to Exploit Argument Relationscited by 10
- 2019RWNE: A Scalable Random-Walk-Based Network Embedding Framework with Personalized Higher-Order Proximity Preservedcited by 1
Built on by
Citing papers per year: 2023: 1.
1 paper.