CLApr 7, 2024

F-MALLOC: Feed-forward Memory Allocation for Continual Learning in Neural Machine Translation

arXiv:2404.04846v233 citationsh-index: 8NAACL
AI Analysis

This addresses the problem of maintaining translation quality and extensibility for NMT systems in continual learning scenarios, representing a novel method for a known bottleneck.

The paper tackles catastrophic forgetting in continual learning for neural machine translation by proposing F-MALLOC, a method that decomposes feed-forward layers into memory cells allocated to tasks, resulting in higher BLEU scores and near-zero forgetting.

In the evolving landscape of Neural Machine Translation (NMT), the pretrain-then-finetune paradigm has yielded impressive results. However, the persistent challenge of Catastrophic Forgetting (CF) remains a hurdle. While previous work has introduced Continual Learning (CL) methods to address CF, these approaches grapple with the delicate balance between avoiding forgetting and maintaining system extensibility. To address this, we propose a CL method, named $\textbf{F-MALLOC}$ ($\textbf{F}$eed-forward $\textbf{M}$emory $\textbf{ALLOC}ation)$. F-MALLOC is inspired by recent insights highlighting that feed-forward layers emulate neural memories and encapsulate crucial translation knowledge. It decomposes feed-forward layers into discrete memory cells and allocates these memories to different tasks. By learning to allocate and safeguard these memories, our method effectively alleviates CF while ensuring robust extendability. Besides, we propose a comprehensive assessment protocol for multi-stage CL of NMT systems. Experiments conducted following this new protocol showcase the superior performance of F-MALLOC, evidenced by higher BLEU scores and almost zero forgetting.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes