More Like this

1. What is Your Discount Factor?

   Tasdighi_Kalat, Shadi; Sankaranarayanan, Sriram; Trivedi, Ashutosh (August 2024, Springer, Cham)

   Hillston, Jane; Soudjiani, Sadegh (Ed.)

   We study the problem of inferring the discount factor of an agent optimizing a discounted reward objective in a finite state Markov Decision Process (MDP). Discounted reward objectives are common in sequential optimization, reinforcement learning, and algorithmic game theory. The discount factor is an important parameter used in formulating the discounted reward. It captures the “time value” of the reward- i.e., how much reward at hand would equal a promised reward at a future time. Knowing an agent’s discount factor can provide valuable insights into their decision-making, and help predict their preferences in previously unseen environments. However, pinpointing the exact value of the discount factor used by the agent is a challenging problem. Ad-hoc guesses are often incorrect. This paper focuses on the problem of computing the range of possible discount factors for a rational agent given their policy. A naive solution to this problem can be quite expensive. A classic result by Smallwood shows that the interval [0, 1) of possible discount factor can be partitioned into finitely many sub-intervals, such that the optimal policy remains the same for each such sub-interval. Furthermore, optimal policies for neighboring sub-intervals differ for a single state. We show how Smallwood’s result can be exploited to search for discount factor intervals for which a given policy is optimal by reducing it to polynomial root isolation. We extend the result to situations where the policy is suboptimal, but with a value function that is close to optimal. We develop numerical approaches to solve the discount factor elicitation problem and demonstrate the effectiveness of our algorithms through some case studies. 

   more » « less
2. Optimal Attack and Defense for Reinforcement Learning

   https://doi.org/10.1609/aaai.v38i13.29346  

   McMahan, Jeremy; Wu, Young; Zhu, Xiaojin; Xie, Qiaomin (March 2024, Proceedings of the AAAI Conference on Artificial Intelligence)

   To ensure the usefulness of Reinforcement Learning (RL) in real systems, it is crucial to ensure they are robust to noise and adversarial attacks. In adversarial RL, an external attacker has the power to manipulate the victim agent's interaction with the environment. We study the full class of online manipulation attacks, which include (i) state attacks, (ii) observation attacks (which are a generalization of perceived-state attacks), (iii) action attacks, and (iv) reward attacks. We show the attacker's problem of designing a stealthy attack that maximizes its own expected reward, which often corresponds to minimizing the victim's value, is captured by a Markov Decision Process (MDP) that we call a meta-MDP since it is not the true environment but a higher level environment induced by the attacked interaction. We show that the attacker can derive optimal attacks by planning in polynomial time or learning with polynomial sample complexity using standard RL techniques. We argue that the optimal defense policy for the victim can be computed as the solution to a stochastic Stackelberg game, which can be further simplified into a partially-observable turn-based stochastic game (POTBSG). Neither the attacker nor the victim would benefit from deviating from their respective optimal policies, thus such solutions are truly robust. Although the defense problem is NP-hard, we show that optimal Markovian defenses can be computed (learned) in polynomial time (sample complexity) in many scenarios. 

   more » « less
3. Online Reinforcement Learning-Based Pedagogical Planning for Narrative-Centered Learning Environments

   https://doi.org/10.1609/aaai.v38i21.30365  

   Fahid, Fahmid Morshed; Rowe, Jonathan; Kim, Yeojin; Srivastava, Shashank; Lester, James (March 2024, Proceedings of the AAAI Conference on Artificial Intelligence)

   Pedagogical planners can provide adaptive support to students in narrative-centered learning environments by dynamically scaffolding student learning and tailoring problem scenarios. Reinforcement learning (RL) is frequently used for pedagogical planning in narrative-centered learning environments. However, RL-based pedagogical planning raises significant challenges due to the scarcity of data for training RL policies. Most prior work has relied on limited-size datasets and offline RL techniques for policy learning. Unfortunately, offline RL techniques do not support on-demand exploration and evaluation, which can adversely impact the quality of induced policies. To address the limitation of data scarcity and offline RL, we propose INSIGHT, an online RL framework for training data-driven pedagogical policies that optimize student learning in narrative-centered learning environments. The INSIGHT framework consists of three components: a narrative-centered learning environment simulator, a simulated student agent, and an RL-based pedagogical planner agent, which uses a reward metric that is associated with effective student learning processes. The framework enables the generation of synthetic data for on-demand exploration and evaluation of RL-based pedagogical planning. We have implemented INSIGHT with OpenAI Gym for a narrative-centered learning environment testbed with rule-based simulated student agents and a deep Q-learning-based pedagogical planner. Our results show that online deep RL algorithms can induce near-optimal pedagogical policies in the INSIGHT framework, while offline deep RL algorithms only find suboptimal policies even with large amounts of data. 

   more » « less
4. Inverse Reinforcement Learning of Interaction Dynamics from Demonstrations

   Mostafa Hussein, Momotaz Begum (January 2019, IEEE International Conference on Robotics and Automation)

   This paper presents a framework to learn the reward function underlying high-level sequential tasks from demonstrations. The purpose of reward learning, in the context of learning from demonstration (LfD), is to generate policies that mimic the demonstrator’s policies, thereby enabling imitation learning. We focus on a human-robot interaction(HRI) domain where the goal is to learn and model structured interactions between a human and a robot. Such interactions can be modeled as a partially observable Markov decision process (POMDP) where the partial observability is caused by uncertainties associated with the ways humans respond to different stimuli. The key challenge in finding a good policy in such a POMDP is determining the reward function that was observed by the demonstrator. Existing inverse reinforcement learning(IRL) methods for POMDPs are computationally very expensive and the problem is not well understood. In comparison, IRL algorithms for Markov decision process (MDP) are well defined and computationally efficient. We propose an approach of reward function learning for high-level sequential tasks from human demonstrations where the core idea is to reduce the underlying POMDP to an MDP and apply any efficient MDP-IRL algorithm. Our extensive experiments suggest that the reward function learned this way generates POMDP policies that mimic the policies of the demonstrator well. 

   more » « less
5. Feasible Q-Learning for Average Reward Reinforcement Learning

   Ying, Jin; Ramki, Gummadi; Zhengyuan, Zhou; Blanchet, Jose (May 2024, Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR)

   Dasgupta, Sanjoy; Mandt, Stephan; Li, Yingzhen (Ed.)

   Average reward reinforcement learning (RL) provides a suitable framework for capturing the objective (i.e. long-run average reward) for continuing tasks, where there is often no natural way to identify a discount factor. However, existing average reward RL algorithms with sample complexity guarantees are not feasible, as they take as input the (unknown) mixing time of the Markov decision process (MDP). In this paper, we make initial progress towards addressing this open problem. We design a feasible average-reward $$Q$$-learning framework that requires no knowledge of any problem parameter as input. Our framework is based on discounted $$Q$$-learning, while we dynamically adapt the discount factor (and hence the effective horizon) to progressively approximate the average reward. In the synchronous setting, we solve three tasks: (i) learn a policy that is $$\epsilon$$-close to optimal, (ii) estimate optimal average reward with $$\epsilon$$-accuracy, and (iii) estimate the bias function (similar to $$Q$$-function in discounted case) with $$\epsilon$$-accuracy. We show that with carefully designed adaptation schemes, (i) can be achieved with $$\tilde{O}(\frac{SA t_{\mathrm{mix}}^{8}}{\epsilon^{8}})$$ samples, (ii) with $$\tilde{O}(\frac{SA t_{\mathrm{mix}}^5}{\epsilon^5})$$ samples, and (iii) with $$\tilde{O}(\frac{SA B}{\epsilon^9})$$ samples, where $$t_\mathrm{mix}$$ is the mixing time, and $B > 0$ is an MDP-dependent constant. To our knowledge, we provide the first finite-sample guarantees that are polynomial in $$S, A, t_{\mathrm{mix}}, \epsilon$$ for a feasible variant of $$Q$$-learning. That said, the sample complexity bounds have tremendous room for improvement, which we leave for the community’s best minds. Preliminary simulations verify that our framework is effective without prior knowledge of parameters as input. 

   more » « less