Multi-intention Inverse Q-learning for Interpretable Behavior Representation

In advancing the understanding of decision-making processes, Inverse Reinforcement Learning (IRL) have proven instrumental in reconstructing animal's multiple intentions amidst complex behaviors. Given the recent development of a continuous-time multi-intention IRL framework, there has been persistent inquiry into inferring discrete time-varying rewards with IRL. To tackle the challenge, we introduce Latent (Markov) Variable Inverse Q-learning (L(M)V-IQL), a novel class of IRL algorthms tailored for accommodating discrete intrinsic reward functions. Leveraging an Expectation-Maximization approach, we cluster observed expert trajectories into distinct intentions and independently solve the IRL problem for each. Demonstrating the efficacy of L(M)V-IQL through simulated experiments and its application to different real mouse behavior datasets, our approach surpasses current benchmarks in animal behavior prediction, producing interpretable reward functions. This advancement holds promise for neuroscience and cognitive science, contributing to a deeper understanding of decision-making and uncovering underlying brain mechanisms.