In this paper we propose a novel gradient algorithm to learn a policy from an expert's observed behavior assuming that the expert behaves optimally with respect to some unknown reward function of a Markovian Decision Problem. The algorithm's aim is to find a reward function such that the resulting optimal policy matches well the expert's observed behavior. The main difficulty is that the mapping from the parameters to policies is both nonsmooth and highly redundant. Resorting to subdifferentials solves the first difficulty, while the second one is over- come by computing natural gradients. We tested the proposed method in two artificial domains and found it to be more reliable and efficient than some previous methods.

Oriental ink painting, called Sumi-e, is one of the most appealing painting styles that has attracted artists around the world. Major challenges in computer-based Sumi-e simulation are to abstract complex scene information and draw smooth and natural brush strokes. To automatically find such strokes, we propose to model the brush as a reinforcement learning agent, and learn desired brush-trajectories by maximizing the sum of rewards in the policy search framework. We also provide elaborate design of actions, states, and rewards tailored for a Sumi-e agent. The effectiveness of our proposed approach is demonstrated through simulated Sumi-e experiments.

Inverse optimal control, also known as inverse reinforcement learning, is the problem of recovering an unknown reward function in a Markov decision process from expert demonstrations of the optimal policy. We introduce a probabilistic inverse optimal control algorithm that scales gracefully with task dimensionality, and is suitable for large, continuous domains where even computing a full policy is impractical. By using a local approximation of the reward function, our method can also drop the assumption that the demonstrations are globally optimal, requiring only local optimality. This allows it to learn from examples that are unsuitable for prior methods.

Continuous state spaces and stochastic, switching dynamics characterize a number of rich, realworld domains, such as robot navigation across varying terrain. We describe a reinforcementlearning algorithm for learning in these domains and prove for certain environments the algorithm is probably approximately correct with a sample complexity that scales polynomially with the state-space dimension. Unfortunately, no optimal planning techniques exist in general for such problems; instead we use fitted value iteration to solve the learned MDP, and include the error due to approximate planning in our bounds. Finally, we report an experiment using a robotic car driving over varying terrain to demonstrate that these dynamics representations adequately capture real-world dynamics and that our algorithm can be used to efficiently solve such problems.

Recent research leverages results from the continuous-armed bandit literature to create a reinforcement-learning algorithm for continuous state and action spaces. Initially proposed in a theoretical setting, we provide the first examination of the empirical properties of the algorithm. Through experimentation, we demonstrate the effectiveness of this planning method when coupled with exploration and model learning and show that, in addition to its formal guarantees, the approach is very competitive with other continuous-action reinforcement learners.

This paper has been withdrawn by the author. This draft is withdrawn for its poor quality in english, unfortunately produced by the author when he was just starting his science route. Look at the ICML version instead: http://icml2008.cs.helsinki.fi/papers/111.pdf

In reinforcement learning (RL), rewards have been considered the most important feedback in understanding the environment. However, recently there have been interesting forays into other modes such as using sporadic supervisory inputs. This brings into the learning process richer information about the world of interest. In this paper, we model these supervisory inputs as specific types of instructions that provide information in the form of an expert's control decision and certain structural regularities in the state space. We further provide a mathematical formulation for the same and propose a framework to incorporate them into the learning process.

One of the challenges for artificial agents is managing the complexity of their environment as they learn tasks especially if they are grounded in the physical world. A scalable solution to address the state explosion problem is thus a prerequisite of physically grounded, agentbased systems. This paper presents a framework for developing grounded, symbolic representations aimed at scaling subsequent learning as well as forming a basis for symbolic reasoning. These symbols partition the environment so the agent need only consider an abstract view of the original space when learning new tasks and allows it to apply acquired symbols to novel situations.

In reinforcement learning (RL), rewards have been considered the most important feedback in understanding the environment. However, recently there have been interesting forays into other modes such as using sporadic supervisory inputs. This brings into the learning process richer information about the world of interest. In this paper, we model these supervisory inputs as specific types of instructions that provide information in the form of an expert's control decision and certain structural regularities in the state space. We further provide a mathematical formulation for the same and propose a framework to incorporate them into the learning process.

The free energy functional has recently been proposed as a variational principle for bounded rational decision-making, since it instantiates a natural trade-off between utility gains and information processing costs that can be axiomatically derived. Here we apply the free energy principle to general decision trees that include both adversarial and stochastic environments. We derive generalized sequential optimality equations that not only include the Bellman optimality equations as a limit case, but also lead to well-known decision-rules such as Expectimax, Minimax and Expectiminimax. We show how these decision-rules can be derived from a single free energy principle that assigns a resource parameter to each node in the decision tree. These resource parameters express a concrete computational cost that can be measured as the amount of samples that are needed from the distribution that belongs to each node. The free energy principle therefore provides the normative basis for generalized optimality equations that account for both adversarial and stochastic environments.