Incremental Classifier Learning with Generative Adversarial Networks

Abstract

In this paper, we address the incremental classifier learning problem, which suffers from catastrophic forgetting. The main reason for catastrophic forgetting is that the past data are not available during learning. Typical approaches keep some exemplars for the past classes and use distillation regularization to retain the classification capability on the past classes and balance the past and new classes. However, there are four main problems with these approaches. First, the loss function is not efficient for classification. Second, there is unbalance problem between the past and new classes. Third, the size of pre-decided exemplars is usually limited and they might not be distinguishable from unseen new classes. Forth, the exemplars may not be allowed to be kept for a long time due to privacy regulations. To address these problems, we propose (a) a new loss function to combine the cross-entropy loss and distillation loss, (b) a simple way to estimate and remove the unbalance between the old and new classes , and (c) using Generative Adversarial Networks (GANs) to generate historical data and select representative exemplars during generation. We believe that the data generated by GANs have much less privacy issues than real images because GANs do not directly copy any real image patches. We evaluate the proposed method on CIFAR-100, Flower-102, and MS-Celeb-1M-Base datasets and extensive experiments demonstrate the effectiveness of our method.

arXiv comment10 pages, 6 figures

References in corpus (10)

• Distilling the Knowledge in a Neural Network
• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Conditional Generative Adversarial Nets
• Overcoming catastrophic forgetting in neural networks
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Conditional Image Synthesis With Auxiliary Classifier GANs
• Gradient Episodic Memory for Continual Learning
• Less-forgetting Learning in Deep Neural Networks
• MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition
• One-shot Face Recognition by Promoting Underrepresented Classes

Cited by in corpus (22)

• Three scenarios for continual learning
• A Comprehensive Study of Class Incremental Learning Algorithms for Visual Tasks
• Generative replay with feedback connections as a general strategy for continual learning
• Incremental Learning for Semantic Segmentation of Large-Scale Remote Sensing Data
• Compacting, Picking and Growing for Unforgetting Continual Learning
• i-Mix: A Domain-Agnostic Strategy for Contrastive Representation Learning
• EEC: Learning to Encode and Regenerate Images for Continual Learning
• Distillation Techniques for Pseudo-rehearsal Based Incremental Learning
• ContCap: A scalable framework for continual image captioning
• Autoencoder-Based Incremental Class Learning without Retraining on Old Data
• Storing Encoded Episodes as Concepts for Continual Learning
• OpenLORIS-Object: A Robotic Vision Dataset and Benchmark for Lifelong Deep Learning
• Continual Learning: Tackling Catastrophic Forgetting in Deep Neural Networks with Replay Processes
• Memory Efficient Class-Incremental Learning for Image Classification
• Lifelong Learning with Searchable Extension Units
• Accretionary Learning with Deep Neural Networks
• Incremental Concept Learning via Online Generative Memory Recall
• Energy Aligning for Biased Models
• LIRA: Lifelong Image Restoration from Unknown Blended Distortions
• BooVAE: Boosting Approach for Continual Learning of VAE
• Incremental Classifier Learning Based on PEDCC-Loss and Cosine Distance
• Do Not Forget to Attend to Uncertainty while Mitigating Catastrophic Forgetting