Meta-Learning Neural Bloom Filters

TL;DR

This paper introduces Neural Bloom Filters, a neural data structure learned via meta-learning for approximate set membership, achieving significant compression and fast adaptation in high-throughput or ephemeral data scenarios.

Contribution

It proposes a novel neural memory architecture, Neural Bloom Filter, that enables one-shot learning of set membership with superior compression compared to classical Bloom Filters.

Findings

Achieves substantial compression over classical Bloom Filters.

Enables one-shot learning of set membership.

Effective in high-throughput and ephemeral data scenarios.

Abstract

There has been a recent trend in training neural networks to replace data structures that have been crafted by hand, with an aim for faster execution, better accuracy, or greater compression. In this setting, a neural data structure is instantiated by training a network over many epochs of its inputs until convergence. In applications where inputs arrive at high throughput, or are ephemeral, training a network from scratch is not practical. This motivates the need for few-shot neural data structures. In this paper we explore the learning of approximate set membership over a set of data in one-shot via meta-learning. We propose a novel memory architecture, the Neural Bloom Filter, which is able to achieve significant compression gains over classical Bloom Filters and existing memory-augmented neural networks.