Learning Novel Policies For Tasks
This addresses the issue of policy convergence to suboptimal solutions in reinforcement learning, particularly for deceptive tasks, though it is incremental as it builds on existing policy gradient methods.
The paper tackles the problem of finding diverse policies for a given task reward function in reinforcement learning, resulting in a collection of policies that solve tasks while producing distinct action sequences, as demonstrated on maze navigation, robot arm reaching, hopper locomotion, and deceptive tasks.
In this work, we present a reinforcement learning algorithm that can find a variety of policies (novel policies) for a task that is given by a task reward function. Our method does this by creating a second reward function that recognizes previously seen state sequences and rewards those by novelty, which is measured using autoencoders that have been trained on state sequences from previously discovered policies. We present a two-objective update technique for policy gradient algorithms in which each update of the policy is a compromise between improving the task reward and improving the novelty reward. Using this method, we end up with a collection of policies that solves a given task as well as carrying out action sequences that are distinct from one another. We demonstrate this method on maze navigation tasks, a reaching task for a simulated robot arm, and a locomotion task for a hopper. We also demonstrate the effectiveness of our approach on deceptive tasks in which policy gradient methods often get stuck.