We consider the finite-horizon offline reinforcement learning (RL) setting, and are motivated by the challenge of learning the policy at any step h in dynamic programming (DP) algorithms. To learn this, it is sufficient to evaluate the treatment effect of deviating from the behavioral policy at step h after having optimized the policy for all future steps. Since the policy at any step can affect next-state distributions, the related distributional shift challenges can make this problem far more statistically hard than estimating such treatment effects in the stochastic contextual bandit setting. However, the hardness of many real-world RL instances lies between the two regimes. We develop a flexible and general method called selective uncertainty propagation for confidence interval construction that adapts to the hardness of the associated distribution shift challenges. We show benefits of our approach on toy environments and demonstrate the benefits of these techniques for offline policy learning.

Multi-objective optimization (MOO) aims to optimize multiple, possibly conflicting objectives with widespread applications. We introduce a novel interacting particle method for MOO inspired by molecular dynamics simulations. Our approach combines overdamped Langevin and birth-death dynamics, incorporating a “dominance potential” to steer particles toward global Pareto optimality. In contrast to previous methods, our method is able to relocate dominated particles, making it particularly adept at managing Pareto fronts of complicated geometries. Our method is also theoretically grounded as a Wasserstein-Fisher-Rao gradient flow with convergence guarantees. Extensive experiments confirm that our approach outperforms state-of-the-art methods on challenging synthetic and real-world datasets.

Many parallel algorithms in domains such as graph analytics and simulations rely on priority-based task scheduling. In such environments, the data structure of choice is a concurrent priority queue (PQ). Unfortunately, PQ algorithms exhibit an undesirable tradeoff. On one hand, strict PQs always dequeue the highest-priority task, and thus fail to scale because of contention at the head of the queue. On the other hand, relaxed PQs avoid contention by dequeuing tasks that are sometimes so far from the head that the resulting schedule misses the benefit of priority-based scheduling.

Long-term temporal credit assignment is an important challenge in deep reinforcement learning (RL). It refers to the ability of the agent to attribute actions to consequences that may occur after a long time interval. Existing policy-gradient and Q-learning algorithms typically rely on dense environmental rewards that provide rich short-term supervision and help with credit assignment. However, they struggle to solve tasks with delays between an action and the corresponding rewarding feedback. To make credit assignment easier, recent works have proposed algorithms to learn dense "guidance" rewards that could be used in place of the sparse or delayed environmental rewards. This paper is in the same vein -- starting with a surrogate RL objective that involves smoothing in the trajectory-space, we arrive at a new algorithm for learning guidance rewards. We show that the guidance rewards have an intuitive interpretation, and can be obtained without training any additional neural networks. Due to the ease of integration, we use the guidance rewards in a few popular algorithms (Q-learning, Actor-Critic, Distributional-RL) and present results in single-agent and multi-agent tasks that elucidate the benefit of our approach when the environmental rewards are sparse or delayed.

Deep reinforcement learning (RL) algorithms have achieved great success on a wide variety of sequential decision-making tasks. However, many of these algorithms suffer from high sample complexity when learning from scratch using environmental rewards, due to issues such as credit-assignment and high-variance gradients, among others. Transfer learning, in which knowledge gained on a source task is applied to more efficiently learn a different but related target task, is a promising approach to improve the sample complexity in RL. Prior work has considered using pre-trained teacher policies to enhance the learning of the student policy, albeit with the constraint that the teacher and the student MDPs share the state-space or the action-space. In this paper, we propose a new framework for transfer learning where the teacher and the student can have arbitrarily different state- and action-spaces. To handle this mismatch, we produce embeddings which can systematically extract knowledge from the teacher policy and value networks, and blend it into the student networks. To train the embeddings, we use a task-aligned loss and show that the representations could be enriched further by adding a mutual information loss. Using a set of challenging simulated robotic locomotion tasks involving many-legged centipedes, we demonstrate successful transfer learning in situations when the teacher and student have different state- and action-spaces.

In multicores, performance-critical synchronization is increasingly performed in a lock-free manner using atomic instructions such as CAS or LL/SC. However, when many processors synchronize on the same variable, performance can still degrade significantly. Contending writes get serialized, creating a non-scalable condition. Past proposals that build hardware queues of synchronizing processors do not fundamentally solve this problem---at best, they help to efficiently serialize the contending writes.