Confidence estimation in Deep Neural networks via density modelling
This addresses the issue of unreliable confidence estimates in deep neural networks for applications requiring robust decision-making, though it is incremental as it builds on existing density modeling approaches.
The paper tackles the problem of accurately estimating predictive confidence in deep neural networks, particularly when they are fooled by adversarial noise, by proposing a novel confidence measure based on density modeling that shows reduced confidence scores for distorted images, unlike traditional softmax methods.
State-of-the-art Deep Neural Networks can be easily fooled into providing incorrect high-confidence predictions for images with small amounts of adversarial noise. Does this expose a flaw with deep neural networks, or do we simply need a better way to estimate confidence? In this paper we consider the problem of accurately estimating predictive confidence. We formulate this problem as that of density modelling, and show how traditional methods such as softmax produce poor estimates. To address this issue, we propose a novel confidence measure based on density modelling approaches. We test these measures on images distorted by blur, JPEG compression, random noise and adversarial noise. Experiments show that our confidence measure consistently shows reduced confidence scores in the presence of such distortions - a property which softmax often lacks.