LSEBMCL: A Latent Space Energy-Based Model for Continual Learning

2025-01-09

AI Summary

To address catastrophic forgetting in continual learning of language models, this paper proposes a replay-free and regularization-free method based on an Energy-Based Model (EBM) operating in the latent space. Specifically, the EBM is embedded into the model’s hidden representation space as an auxiliary generator; during training on new tasks, it enables efficient contrastive sampling to reconstruct salient samples from previous tasks, thereby implicitly preserving historical knowledge. The approach avoids storing raw data or introducing task-specific parameters, substantially reducing memory footprint and computational overhead. Evaluated on multiple NLP continual learning benchmarks—including WikiSQL and SQuAD—it achieves state-of-the-art performance, effectively mitigating forgetting while maintaining strong adaptation to new tasks. This work represents the first application of latent-space EBM modeling to NLP continual learning, offering a novel paradigm for replay-free continual adaptation.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Life-Long and Continual Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Continual learning has become essential in many practical applications such as online news summaries and product classification. The primary challenge is known as catastrophic forgetting, a phenomenon where a model inadvertently discards previously learned knowledge when it is trained on new tasks. Existing solutions involve storing exemplars from previous classes, regularizing parameters during the fine-tuning process, or assigning different model parameters to each task. The proposed solution LSEBMCL (Latent Space Energy-Based Model for Continual Learning) in this work is to use energy-based models (EBMs) to prevent catastrophic forgetting by sampling data points from previous tasks when training on new ones. The EBM is a machine learning model that associates an energy value with each input data point. The proposed method uses an EBM layer as an outer-generator in the continual learning framework for NLP tasks. The study demonstrates the efficacy of EBM in NLP tasks, achieving state-of-the-art results in all experiments.

Problem

• Catastrophic Forgetting
• Continuous Learning
• Language Tasks

Innovation

• LSEBMCL
• Continual Learning
• Disaster Prevention in Language Tasks