Document Image Cleaning using Budget-Aware Black-Box Approximation

TL;DR

This paper introduces budget-aware sample selection algorithms for training document preprocessors that approximate OCR engines, significantly reducing query costs and training time while maintaining high accuracy.

Contribution

It presents novel sample selection methods that enable efficient training of OCR preprocessors with minimal queries, reducing costs and computational resources.

Findings

Achieved over 60% reduction in training time.

Reduced OCR engine queries to less than 10% of original.

Improved word-level accuracy by 4% with minimal queries.

Abstract

Recent work has shown that by approximating the behaviour of a non-differentiable black-box function using a neural network, the black-box can be integrated into a differentiable training pipeline for end-to-end training. This methodology is termed "differentiable bypass,'' and a successful application of this method involves training a document preprocessor to improve the performance of a black-box OCR engine. However, a good approximation of an OCR engine requires querying it for all samples throughout the training process, which can be computationally and financially expensive. Several zeroth-order optimization (ZO) algorithms have been proposed in black-box attack literature to find adversarial examples for a black-box model by computing its gradient in a query-efficient manner. However, the query complexity and convergence rate of such algorithms makes them infeasible for our problem. In this work, we propose two sample selection algorithms to train an OCR preprocessor with less than 10% of the original system's OCR engine queries, resulting in more than 60% reduction of the total training time without significant loss of accuracy. We also show an improvement of 4% in the word-level accuracy of a commercial OCR engine with only 2.5% of the total queries and a 32x reduction in monetary cost. Further, we propose a simple ranking technique to prune 30% of the document images from the training dataset without affecting the system's performance.