Category Archives: Reinforcement Learning In Ai

Relation between optimization and reinforcement learning

Megumi Miyashita, Shiro Yano, Toshiyuki Kondo Mirror descent search and its acceleration, Robotics and Autonomous Systems, Volume 106, 2018, Pages 107-116 DOI: 10.1016/j.robot.2018.04.009.

In recent years, attention has been focused on the relationship between black-box optimization problem and reinforcement learning problem. In this research, we propose the Mirror Descent Search (MDS) algorithm which is applicable both for black box optimization problems and reinforcement learning problems. Our method is based on the mirror descent method, which is a general optimization algorithm. The contribution of this research is roughly twofold. We propose two essential algorithms, called MDS and Accelerated Mirror Descent Search (AMDS), and two more approximate algorithms: Gaussian Mirror Descent Search (G-MDS) and Gaussian Accelerated Mirror Descent Search (G-AMDS). This research shows that the advanced methods developed in the context of the mirror descent research can be applied to reinforcement learning problem. We also clarify the relationship between an existing reinforcement learning algorithm and our method. With two evaluation experiments, we show our proposed algorithms converge faster than some state-of-the-art methods.

Adapting inverse reinforcement learning for including the risk-aversion of the agent

Sumeet Singh, Jonathan Lacotte, Anirudha Majumdar, and Marco Pavone, Risk-sensitive inverse reinforcement learning via semi- and non-parametric methods , The International Journal of Robotics Research First Published May 22, 2018 DOI: 10.1177/0278364918772017.

The literature on inverse reinforcement learning (IRL) typically assumes that humans take actions to minimize the expected value of a cost function, i.e., that humans are risk neutral. Yet, in practice, humans are often far from being risk neutral. To fill this gap, the objective of this paper is to devise a framework for risk-sensitive (RS) IRL to explicitly account for a human’s risk sensitivity. To this end, we propose a flexible class of models based on coherent risk measures, which allow us to capture an entire spectrum of risk preferences from risk neutral to worst case. We propose efficient non-parametric algorithms based on linear programming and semi-parametric algorithms based on maximum likelihood for inferring a human’s underlying risk measure and cost function for a rich class of static and dynamic decision-making settings. The resulting approach is demonstrated on a simulated driving game with 10 human participants. Our method is able to infer and mimic a wide range of qualitatively different driving styles from highly risk averse to risk neutral in a data-efficient manner. Moreover, comparisons of the RS-IRL approach with a risk-neutral model show that the RS-IRL framework more accurately captures observed participant behavior both qualitatively and quantitatively, especially in scenarios where catastrophic outcomes such as collisions can occur.

Multi-agent reinfocerment learning for working with high-dimensional spaces

David L. Leottau, Javier Ruiz-del-Solar, Robert Babuška, Decentralized Reinforcement Learning of Robot Behaviors, Artificial Intelligence, Volume 256, 2018, Pages 130-159, DOI: 10.1016/j.artint.2017.12.001.

A multi-agent methodology is proposed for Decentralized Reinforcement Learning (DRL) of individual behaviors in problems where multi-dimensional action spaces are involved. When using this methodology, sub-tasks are learned in parallel by individual agents working toward a common goal. In addition to proposing this methodology, three specific multi agent DRL approaches are considered: DRL-Independent, DRL Cooperative-Adaptive (CA), and DRL-Lenient. These approaches are validated and analyzed with an extensive empirical study using four different problems: 3D Mountain Car, SCARA Real-Time Trajectory Generation, Ball-Dribbling in humanoid soccer robotics, and Ball-Pushing using differential drive robots. The experimental validation provides evidence that DRL implementations show better performances and faster learning times than their centralized counterparts, while using less computational resources. DRL-Lenient and DRL-CA algorithms achieve the best final performances for the four tested problems, outperforming their DRL-Independent counterparts. Furthermore, the benefits of the DRL-Lenient and DRL-CA are more noticeable when the problem complexity increases and the centralized scheme becomes intractable given the available computational resources and training time.

Using interactive reinforcement learning with the advisor being another reinforcement learning agent

Francisco Cruz, Sven Magg, Yukie Nagai & Stefan Wermter, Improving interactive reinforcement learning: What makes a good teacher?, Connection Science, DOI: 10.1080/09540091.2018.1443318.

Interactive reinforcement learning (IRL) has become an important apprenticeship approach to speed up convergence in classic reinforcement learning (RL) problems. In this regard, a variant of IRL is policy shaping which uses a parent-like trainer to propose the next action to be performed and by doing so reduces the search space by advice. On some occasions, the trainer may be another artificial agent which in turn was trained using RL methods to afterward becoming an advisor for other learner-agents. In this work, we analyse internal representations and characteristics of artificial agents to determine which agent may outperform others to become a better trainer-agent. Using a polymath agent, as compared to a specialist agent, an advisor leads to a larger reward and faster convergence of the reward signal and also to a more stable behaviour in terms of the state visit frequency of the learner-agents. Moreover, we analyse system interaction parameters in order to determine how influential they are in the apprenticeship process, where the consistency of feedback is much more relevant when dealing with different learner obedience parameters.

Modelling emotions in adaptive agents through the action selection part of reinforcement learning, plus some references on the neurophysiological bases of RL and a good review of literature on emotions

Joost Broekens , Elmer Jacobs , Catholijn M. Jonker, A reinforcement learning model of joy, distress, hope and fear, Connection Science, Vol. 27, Iss. 3, 2015, DOI: 10.1080/09540091.2015.1031081.

In this paper we computationally study the relation between adaptive behaviour and emotion. Using the reinforcement learning framework, we propose that learned state utility, V(s), models fear (negative) and hope (positive) based on the fact that both signals are about anticipation of loss or gain. Further, we propose that joy/distress is a signal similar to the error signal. We present agent-based simulation experiments that show that this model replicates psychological and behavioural dynamics of emotion. This work distinguishes itself by assessing the dynamics of emotion in an adaptive agent framework – coupling it to the literature on habituation, development, extinction and hope theory. Our results support the idea that the function of emotion is to provide a complex feedback signal for an organism to adapt its behaviour. Our work is relevant for understanding the relation between emotion and adaptation in animals, as well as for human–robot interaction, in particular how emotional signals can be used to communicate between adaptive agents and humans.

Transfer learning in reinforcement learning through case-based and the use of heuristics for selecting actions

Reinaldo A.C. Bianchi, Luiz A. Celiberto Jr., Paulo E. Santos, Jackson P. Matsuura, Ramon Lopez de Mantaras, Transferring knowledge as heuristics in reinforcement learning: A case-based approach, Artificial Intelligence, Volume 226, September 2015, Pages 102-121, ISSN 0004-3702, DOI: 10.1016/j.artint.2015.05.008.

The goal of this paper is to propose and analyse a transfer learning meta-algorithm that allows the implementation of distinct methods using heuristics to accelerate a Reinforcement Learning procedure in one domain (the target) that are obtained from another (simpler) domain (the source domain). This meta-algorithm works in three stages: first, it uses a Reinforcement Learning step to learn a task on the source domain, storing the knowledge thus obtained in a case base; second, it does an unsupervised mapping of the source-domain actions to the target-domain actions; and, third, the case base obtained in the first stage is used as heuristics to speed up the learning process in the target domain.
A set of empirical evaluations were conducted in two target domains: the 3D mountain car (using a learned case base from a 2D simulation) and stability learning for a humanoid robot in the Robocup 3D Soccer Simulator (that uses knowledge learned from the Acrobot domain). The results attest that our transfer learning algorithm outperforms recent heuristically-accelerated reinforcement learning and transfer learning algorithms.

Reinforcement learning when a human is the one providing the rewards to the algorithm

W. Bradley Knox, Peter Stone, Framing reinforcement learning from human reward: Reward positivity, temporal discounting, episodicity, and performance, Artificial Intelligence, Volume 225, August 2015, Pages 24-50, ISSN 0004-3702, DOI: 10.1016/j.artint.2015.03.009.

Several studies have demonstrated that reward from a human trainer can be a powerful feedback signal for control-learning algorithms. However, the space of algorithms for learning from such human reward has hitherto not been explored systematically. Using model-based reinforcement learning from human reward, this article investigates the problem of learning from human reward through six experiments, focusing on the relationships between reward positivity, which is how generally positive a trainer’s reward values are; temporal discounting, the extent to which future reward is discounted in value; episodicity, whether task learning occurs in discrete learning episodes instead of one continuing session; and task performance, the agent’s performance on the task the trainer intends to teach. This investigation is motivated by the observation that an agent can pursue different learning objectives, leading to different resulting behaviors. We search for learning objectives that lead the agent to behave as the trainer intends.
We identify and empirically support a “positive circuits” problem with low discounting (i.e., high discount factors) for episodic, goal-based tasks that arises from an observed bias among humans towards giving positive reward, resulting in an endorsement of myopic learning for such domains. We then show that converting simple episodic tasks to be non-episodic (i.e., continuing) reduces and in some cases resolves issues present in episodic tasks with generally positive reward and—relatedly—enables highly successful learning with non-myopic valuation in multiple user studies. The primary learning algorithm introduced in this article, which we call “vi-tamer”, is the first algorithm to successfully learn non-myopically from reward generated by a human trainer; we also empirically show that such non-myopic valuation facilitates higher-level understanding of the task. Anticipating the complexity of real-world problems, we perform further studies—one with a failure state added—that compare (1) learning when states are updated asynchronously with local bias—i.e., states quickly reachable from the agent’s current state are updated more often than other states—to (2) learning with the fully synchronous sweeps across each state in the vi-tamer algorithm. With these locally biased updates, we find that the general positivity of human reward creates problems even for continuing tasks, revealing a distinct research challenge for future work.