Online Limited Memory Neural-Linear Bandits with Likelihood Matching
This work is significant for researchers and practitioners working with online learning systems, particularly in scenarios where memory constraints lead to catastrophic forgetting in neural-linear bandits.
This paper addresses the challenge of catastrophic forgetting in neural-linear bandits when memory for storing past data is limited. They propose a likelihood matching algorithm, NeuralLinear-LiM2, which achieves performance comparable to unlimited memory approaches while being resilient to catastrophic forgetting.
We study neural-linear bandits for solving problems where {\em both} exploration and representation learning play an important role. Neural-linear bandits harnesses the representation power of Deep Neural Networks (DNNs) and combines it with efficient exploration mechanisms by leveraging uncertainty estimation of the model, designed for linear contextual bandits on top of the last hidden layer. In order to mitigate the problem of representation change during the process, new uncertainty estimations are computed using stored data from an unlimited buffer. Nevertheless, when the amount of stored data is limited, a phenomenon called catastrophic forgetting emerges. To alleviate this, we propose a likelihood matching algorithm that is resilient to catastrophic forgetting and is completely online. We applied our algorithm, Limited Memory Neural-Linear with Likelihood Matching (NeuralLinear-LiM2) on a variety of datasets and observed that our algorithm achieves comparable performance to the unlimited memory approach while exhibits resilience to catastrophic forgetting.