Deep Neural Linear Bandits: Overcoming Catastrophic Forgetting through Likelihood Matching
This addresses a specific issue in sequential decision-making for applications with high-dimensional data, but it is incremental as it builds on existing neural-linear bandit frameworks.
The paper tackled the problem of catastrophic forgetting in neural-linear bandits, where information loss occurs during representation learning, and proposed a limited memory method that achieved superior performance on real-world datasets like regression, classification, and sentiment analysis.
We study the neural-linear bandit model for solving sequential decision-making problems with high dimensional side information. Neural-linear bandits leverage the representation power of deep neural networks and combine it with efficient exploration mechanisms, designed for linear contextual bandits, on top of the last hidden layer. Since the representation is being optimized during learning, information regarding exploration with "old" features is lost. Here, we propose the first limited memory neural-linear bandit that is resilient to this phenomenon, which we term catastrophic forgetting. We evaluate our method on a variety of real-world data sets, including regression, classification, and sentiment analysis, and observe that our algorithm is resilient to catastrophic forgetting and achieves superior performance.