Inverse Reward Design
نویسندگان
چکیده
Autonomous agents optimize the reward function we give them. What they don’t know is how hard it is for us to design a reward function that actually captures what we want. When designing the reward, we might think of some specific training scenarios, and make sure that the reward will lead to the right behavior in those scenarios. Inevitably, agents encounter new scenarios (e.g., new types of terrain) where optimizing that same reward may lead to undesired behavior. Our insight is that reward functions are merely observations about what the designer actually wants, and that they should be interpreted in the context in which they were designed. We introduce inverse reward design (IRD) as the problem of inferring the true objective based on the designed reward and the training MDP. We introduce approximate methods for solving IRD problems, and use their solution to plan risk-averse behavior in test MDPs. Empirical results suggest that this approach can help alleviate negative side effects of misspecified reward functions and mitigate reward hacking.
منابع مشابه
Compatible Reward Inverse Reinforcement Learning
PROBLEM • Inverse Reinforcement Learning (IRL) problem: recover a reward function explaining a set of expert’s demonstrations. • Advantages of IRL over Behavioral Cloning (BC): – Transferability of the reward. • Issues with some IRL methods: – How to build the features for the reward function? – How to select a reward function among all the optimal ones? – What if no access to the environment? ...
متن کاملNonlinear Inverse Reinforcement Learning with Gaussian Processes
We present a probabilistic algorithm for nonlinear inverse reinforcement learning. The goal of inverse reinforcement learning is to learn the reward function in a Markov decision process from expert demonstrations. While most prior inverse reinforcement learning algorithms represent the reward as a linear combination of a set of features, we use Gaussian processes to learn the reward as a nonli...
متن کاملInverse Optimal Control
In Reinforcement Learning, an agent learns a policy that maximizes a given reward function. However, providing a reward function for a given learning task is often non trivial. Inverse Reinforcement Learning, which is sometimes also called Inverse Optimal Control, addresses this problem by learning the reward function from expert demonstrations. The aim of this paper is to give a brief introduc...
متن کاملInverse Reinforcement Learning based on Critical State
Inverse reinforcement learning is tried to search a reward function based on Markov Decision Process. In the IRL topics, experts produce some good traces to make agents learn and adjust the reward function. But the function is difficult to set in some complicate problems. In this paper, Inverse Reinforcement Learning based on Critical State (IRLCS) is proposed to search a succinct and meaningfu...
متن کاملCompetitive Multi-agent Inverse Reinforcement Learning with Sub-optimal Demonstrations
This paper considers the problem of inverse reinforcement learning in zero-sum stochastic games when expert demonstrations are known to be not optimal. Compared to previous works that decouple agents in the game by assuming optimality in expert strategies, we introduce a new objective function that directly pits experts against Nash Equilibrium strategies, and we design an algorithm to solve fo...
متن کامل