On Model Stability as a Function of Random Seed
This addresses model reliability issues for practitioners in machine learning, though it is incremental as it builds on existing weight averaging techniques.
The paper tackles the problem of model instability due to random seeds by quantifying its effects on performance and robustness, and proposes Aggressive Stochastic Weight Averaging (ASWA) and its extension NASWA to improve stability, reducing the standard deviation of model performance by 72% on average.
In this paper, we focus on quantifying model stability as a function of random seed by investigating the effects of the induced randomness on model performance and the robustness of the model in general. We specifically perform a controlled study on the effect of random seeds on the behaviour of attention, gradient-based and surrogate model based (LIME) interpretations. Our analysis suggests that random seeds can adversely affect the consistency of models resulting in counterfactual interpretations. We propose a technique called Aggressive Stochastic Weight Averaging (ASWA)and an extension called Norm-filtered Aggressive Stochastic Weight Averaging (NASWA) which improves the stability of models over random seeds. With our ASWA and NASWA based optimization, we are able to improve the robustness of the original model, on average reducing the standard deviation of the model's performance by 72%.