Compacting, Picking and Growing for Unforgetting Continual Learning

2019/10/15 by Steven C. Y. Hung, Hung, Steven C. Y., Cheng-Hao Tu +11 · 31 citations
Computer Science · Mathematics · #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Human Pose and Action Recognition #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Multimodal Machine Learning Applications #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.1910.06562

To appear in NeurIPS 2019

openalex publication_date 2019/10/15 · arxiv created 2019/10/30 · arxiv updated 2019/10/31 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Continual lifelong learning is essential to many applications. In this paper, we propose a simple but effective approach to continual deep learning. Our approach leverages the principles of deep model compression, critical weights selection, and progressive networks expansion. By enforcing their integration in an iterative manner, we introduce an incremental learning method that is scalable to the number of sequential tasks in a continual learning process. Our approach is easy to implement and owns several favorable characteristics. First, it can avoid forgetting (i.e., learn new tasks while remembering all previous tasks). Second, it allows model expansion but can maintain the model compactness when handling sequential tasks. Besides, through our compaction and selection/expansion mechanism, we show that the knowledge accumulated through learning previous tasks is helpful to build a better model for the new tasks compared to training the models independently with tasks. Experimental results show that our approach can incrementally learn a deep model tackling multiple tasks without forgetting, while the model compactness is maintained with the performance more satisfiable than individual task training.

Citations

• Large-scale Classification of Fine-Art Paintings: Learning The Right Metric on The Right Feature
• Learning without Forgetting
• Overcoming Catastrophic Forgetting by Incremental Moment Matching
• HOUDINI: Lifelong Learning as Program Synthesis
• On the role of neurogenesis in overcoming catastrophic forgetting
• Learning to Learn without Forgetting by Maximizing Transfer and Minimizing Interference
• Learning Multiple Layers of Features from Tiny Images

Cited by

• Dynamic Feedback Engines: Layer-Wise Control for Self-Regulating Continual Learning
• Sequencing to Mitigate Catastrophic Forgetting in Continual Learning
• Multi-Modal Continual Learning via Cross-Modality Adapters and Representation Alignment with Knowledge Preservation
• Sharing the Learned Knowledge-base to Estimate Convolutional Filter Parameters for Continual Image Restoration
• COLA: Continual Learning via Autoencoder Retrieval of Adapters
• KORE: Enhancing Knowledge Injection for Large Multimodal Models via Knowledge-Oriented Controls
• Understanding Catastrophic Interference: On the Identifibility of Latent Representations
• Efficient Reinforcement Learning by Reducing Forgetting with Elephant Activation Functions
• Unsupervised Video Continual Learning via Non-Parametric Deep Embedded Clustering
• Empirical Evidences for the Effects of Feature Diversity in Open Set Recognition and Continual Learning
• Index-Aligned Query Distillation for Transformer-based Incremental Object Detection
• Harnessing Textual Semantic Priors for Knowledge Transfer and Refinement in CLIP-Driven Continual Learning
• RainbowPrompt: Diversity-Enhanced Prompt-Evolving for Continual Learning
• CKAA: Cross-subspace Knowledge Alignment and Aggregation for Robust Continual Learning
• Branch, or Layer? Zeroth-Order Optimization for Continual Learning of Vision-Language Models
• Self-Composing Policies for Scalable Continual Reinforcement Learning
• Onboard Optimization and Learning: A Survey
• Rethinking Continual Learning with Progressive Neural Collapse
• SplitLoRA: Balancing Stability and Plasticity in Continual Learning Through Gradient Space Splitting
• Continual Learning Beyond Experience Rehearsal and Full Model Surrogates
• Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning
• Dynamic Dual Buffer with Divide-and-Conquer Strategy for Online Continual Learning
• Continual Learning for Task-oriented Dialogue System with Iterative Network Pruning, Expanding and Masking
• Gated Integration of Low-Rank Adaptation for Continual Learning of Large Language Models
• AdaHAT: Adaptive Hard Attention to the Task in Task-Incremental Learning
• Plasticity of Growing and Elastic Neural Networks in Online Continual Learning
• Parameter Efficient Continual Learning with Dynamic Low-Rank Adaptation
• Federated Continual Learning with Weighted Inter-client Transfer
• Generalizing MLPs With Dropouts, Batch Normalization, and Skip Connections
• Noise-Tolerant Coreset-Based Class Incremental Continual Learning
• Continual Learning of Multiple Cognitive Functions with Brain-inspired Temporal Development Mechanism

Related

• GROWN: GRow Only When Necessary for Continual Learning
• Self-Net: Lifelong Learning via Continual Self-Modeling
• Reinforced Continual Learning
• Dissecting Continual Learning a Structural and Data Analysis
• Avoiding Forgetting and Allowing Forward Transfer in Continual Learning via Sparse Networks