Learning to Rehearse in Long Sequence Memorization

TL;DR

This paper introduces Rehearsal Memory, a self-supervised memory system that improves long sequence reasoning by selectively rehearsing important information, addressing forgetting and information prioritization issues in memory-augmented neural networks.

Contribution

It proposes a novel rehearsal memory with a history sampler and self-supervised training to better retain crucial information in long sequence tasks.

Findings

Enhanced memory retention of early sequence information.

Improved performance on long sequence reasoning tasks.

Effective in text/video question answering and recommendation tasks.

Abstract

Existing reasoning tasks often have an important assumption that the input contents can be always accessed while reasoning, requiring unlimited storage resources and suffering from severe time delay on long sequences. To achieve efficient reasoning on long sequences with limited storage resources, memory augmented neural networks introduce a human-like write-read memory to compress and memorize the long input sequence in one pass, trying to answer subsequent queries only based on the memory. But they have two serious drawbacks: 1) they continually update the memory from current information and inevitably forget the early contents; 2) they do not distinguish what information is important and treat all contents equally. In this paper, we propose the Rehearsal Memory (RM) to enhance long-sequence memorization by self-supervised rehearsal with a history sampler. To alleviate the gradual forgetting of early information, we design self-supervised rehearsal training with recollection and familiarity tasks. Further, we design a history sampler to select informative fragments for rehearsal training, making the memory focus on the crucial information. We evaluate the performance of our rehearsal memory by the synthetic bAbI task and several downstream tasks, including text/video question answering and recommendation on long sequences.