Value Iteration Networks
This work addresses the challenge of integrating planning into neural networks for reinforcement learning, offering a novel approach that is incremental in combining existing concepts.
The authors tackled the problem of enabling neural networks to perform planning-based reasoning by introducing the Value Iteration Network (VIN), a fully differentiable network with an embedded planning module, which showed improved generalization to unseen domains in path-planning and search tasks.
We introduce the value iteration network (VIN): a fully differentiable neural network with a `planning module' embedded within. VINs can learn to plan, and are suitable for predicting outcomes that involve planning-based reasoning, such as policies for reinforcement learning. Key to our approach is a novel differentiable approximation of the value-iteration algorithm, which can be represented as a convolutional neural network, and trained end-to-end using standard backpropagation. We evaluate VIN based policies on discrete and continuous path-planning domains, and on a natural-language based search task. We show that by learning an explicit planning computation, VIN policies generalize better to new, unseen domains.