Contextual Memory Trees
This work addresses the challenge of scalable memory integration in learning systems, offering incremental improvements for applications like classification and image captioning.
The paper tackles the problem of efficiently managing an unbounded memory store for machine learning by introducing Contextual Memory Trees (CMT), which enable logarithmic-time insertion and retrieval, and demonstrates statistical improvements when integrated into classification algorithms and computational advantages over nearest neighbors in image-captioning tasks.
We design and study a Contextual Memory Tree (CMT), a learning memory controller that inserts new memories into an experience store of unbounded size. It is designed to efficiently query for memories from that store, supporting logarithmic time insertion and retrieval operations. Hence CMT can be integrated into existing statistical learning algorithms as an augmented memory unit without substantially increasing training and inference computation. Furthermore CMT operates as a reduction to classification, allowing it to benefit from advances in representation or architecture. We demonstrate the efficacy of CMT by augmenting existing multi-class and multi-label classification algorithms with CMT and observe statistical improvement. We also test CMT learning on several image-captioning tasks to demonstrate that it performs computationally better than a simple nearest neighbors memory system while benefitting from reward learning.