Accountable Error Characterization

TL;DR

This paper introduces AEC, a method for understanding and characterizing errors in black-box machine learning models using human-understandable features, improving error analysis and data sampling.

Contribution

The paper presents AEC, a novel error characterization approach that identifies error sources and informs data sampling using linguistic features in black-box models.

Findings

AEC effectively categorizes errors into understandable groups.

AEC improves error sample selection over uncertainty-based methods.

Case study on sentiment analysis demonstrates practical utility.

Abstract

Customers of machine learning systems demand accountability from the companies employing these algorithms for various prediction tasks. Accountability requires understanding of system limit and condition of erroneous predictions, as customers are often interested in understanding the incorrect predictions, and model developers are absorbed in finding methods that can be used to get incremental improvements to an existing system. Therefore, we propose an accountable error characterization method, AEC, to understand when and where errors occur within the existing black-box models. AEC, as constructed with human-understandable linguistic features, allows the model developers to automatically identify the main sources of errors for a given classification system. It can also be used to sample for the set of most informative input points for a next round of training. We perform error detection for a sentiment analysis task using AEC as a case study. Our results on the sample sentiment task show that AEC is able to characterize erroneous predictions into human understandable categories and also achieves promising results on selecting erroneous samples when compared with the uncertainty-based sampling.