Scrutinizing and De-Biasing Intuitive Physics with Neural Stethoscopes

TL;DR

This paper introduces neural stethoscopes to analyze and guide deep neural networks in intuitive physics tasks, improving their robustness and de-biasing capabilities.

Contribution

It presents neural stethoscopes as a versatile framework for understanding and controlling factors influencing neural network learning in physics prediction tasks.

Findings

Baseline model is misled by incorrect visual cues.

Promoting meaningful features improves accuracy from 51% to 90%.

De-biasing with adversarial stethoscopes increases performance from 66% to 88%.

Abstract

Visually predicting the stability of block towers is a popular task in the domain of intuitive physics. While previous work focusses on prediction accuracy, a one-dimensional performance measure, we provide a broader analysis of the learned physical understanding of the final model and how the learning process can be guided. To this end, we introduce neural stethoscopes as a general purpose framework for quantifying the degree of importance of specific factors of influence in deep neural networks as well as for actively promoting and suppressing information as appropriate. In doing so, we unify concepts from multitask learning as well as training with auxiliary and adversarial losses. We apply neural stethoscopes to analyse the state-of-the-art neural network for stability prediction. We show that the baseline model is susceptible to being misled by incorrect visual cues. This leads to a performance breakdown to the level of random guessing when training on scenarios where visual cues are inversely correlated with stability. Using stethoscopes to promote meaningful feature extraction increases performance from 51% to 90% prediction accuracy. Conversely, training on an easy dataset where visual cues are positively correlated with stability, the baseline model learns a bias leading to poor performance on a harder dataset. Using an adversarial stethoscope, the network is successfully de-biased, leading to a performance increase from 66% to 88%.