Scalable Set Encoding with Universal Mini-Batch Consistency and Unbiased Full Set Gradient Approximation

TL;DR

This paper introduces a universal mini-batch consistency framework for set functions that allows scalable, unbiased full set gradient approximation, enabling efficient training on large sets across various applications.

Contribution

The authors propose a new class of set functions called UMBC that maintains mini-batch consistency while supporting arbitrary components, along with an unbiased gradient approximation algorithm with constant memory overhead.

Findings

UMBC enables wider function class usage in MBC settings.

The proposed algorithm provides unbiased full set gradient estimates.

Experiments demonstrate efficiency and effectiveness across diverse tasks.

Abstract

Recent work on mini-batch consistency (MBC) for set functions has brought attention to the need for sequentially processing and aggregating chunks of a partitioned set while guaranteeing the same output for all partitions. However, existing constraints on MBC architectures lead to models with limited expressive power. Additionally, prior work has not addressed how to deal with large sets during training when the full set gradient is required. To address these issues, we propose a Universally MBC (UMBC) class of set functions which can be used in conjunction with arbitrary non-MBC components while still satisfying MBC, enabling a wider range of function classes to be used in MBC settings. Furthermore, we propose an efficient MBC training algorithm which gives an unbiased approximation of the full set gradient and has a constant memory overhead for any set size for both train- and test-time. We conduct extensive experiments including image completion, text classification, unsupervised clustering, and cancer detection on high-resolution images to verify the efficiency and efficacy of our scalable set encoding framework. Our code is available at github.com/jeffwillette/umbc