Synthesizing Action Sequences for Modifying Model Decisions

TL;DR

This paper introduces a method to generate realistic, actionable sequences of steps that individuals can take to change unfavorable decisions made by AI models, using program synthesis combined with adversarial attacks.

Contribution

It presents a novel approach that integrates search-based program synthesis with adversarial attacks to produce minimal action sequences for decision modification.

Findings

Effective in generating realistic action sequences

Successfully applied to various deep neural networks

Outperforms baseline methods in minimality and realism

Abstract

When a model makes a consequential decision, e.g., denying someone a loan, it needs to additionally generate actionable, realistic feedback on what the person can do to favorably change the decision. We cast this problem through the lens of program synthesis, in which our goal is to synthesize an optimal (realistically cheapest or simplest) sequence of actions that if a person executes successfully can change their classification. We present a novel and general approach that combines search-based program synthesis and test-time adversarial attacks to construct action sequences over a domain-specific set of actions. We demonstrate the effectiveness of our approach on a number of deep neural networks.