Dropout Prediction Uncertainty Estimation Using Neuron Activation Strength
This work addresses the problem of high computational cost in uncertainty estimation for practitioners, offering an incremental improvement over existing dropout methods.
The paper tackles the computational expense of dropout-based uncertainty estimation by proposing a method that uses neuron activation strengths to estimate prediction uncertainty with a single inference pass, achieving comparable performance to using all layers while reducing resource usage on datasets like MovieLens, Criteo, and EMNIST.
Dropout has been commonly used to quantify prediction uncertainty, i.e, the variations of model predictions on a given input example. However, using dropout in practice can be expensive as it requires running dropout inferences many times. In this paper, we study how to estimate dropout prediction uncertainty in a resource-efficient manner. We demonstrate that we can use neuron activation strengths to estimate dropout prediction uncertainty under different dropout settings and on a variety of tasks using three large datasets, MovieLens, Criteo, and EMNIST. Our approach provides an inference-once method to estimate dropout prediction uncertainty as a cheap auxiliary task. We also demonstrate that using activation features from a subset of the neural network layers can be sufficient to achieve uncertainty estimation performance almost comparable to that of using activation features from all layers, thus reducing resources even further for uncertainty estimation.