i-Algebra: Towards Interactive Interpretability of Deep Neural Networks

TL;DR

i-Algebra introduces an interactive framework for interpreting deep neural networks, enabling users to perform flexible, composable analyses that improve understanding and usability of DNNs in critical domains.

Contribution

This paper presents i-Algebra, the first interactive interpretability framework for DNNs using composable operators and a declarative query language.

Findings

User studies show improved interpretability and usability.

Effective analysis of adversarial inputs and data cleansing.

Demonstrated flexibility in various interpretability tasks.

Abstract

Providing explanations for deep neural networks (DNNs) is essential for their use in domains wherein the interpretability of decisions is a critical prerequisite. Despite the plethora of work on interpreting DNNs, most existing solutions offer interpretability in an ad hoc, one-shot, and static manner, without accounting for the perception, understanding, or response of end-users, resulting in their poor usability in practice. In this paper, we argue that DNN interpretability should be implemented as the interactions between users and models. We present i-Algebra, a first-of-its-kind interactive framework for interpreting DNNs. At its core is a library of atomic, composable operators, which explain model behaviors at varying input granularity, during different inference stages, and from distinct interpretation perspectives. Leveraging a declarative query language, users are enabled to build various analysis tools (e.g., "drill-down", "comparative", "what-if" analysis) via flexibly composing such operators. We prototype i-Algebra and conduct user studies in a set of representative analysis tasks, including inspecting adversarial inputs, resolving model inconsistency, and cleansing contaminated data, all demonstrating its promising usability.