Neural Information Processing Systems http://nips.cc/

Tree Search (TS) is crucial to some of the most influential successes in reinforcement learning. Here, we tackle two major challenges with TS that limit its usability: \textit{distribution shift} and \textit{scalability}. We first discover and analyze a counter-intuitive phenomenon: action selection through TS and a pre-trained value function often leads to lower performance compared to the original pre-trained agent, even when having access to the exact state and reward in future steps. We show this is due to a distribution shift to areas where value estimates are highly inaccurate and analyze this effect using Extreme Value theory. To overcome this problem, we introduce a novel off-policy correction term that accounts for the mismatch between the pre-trained value and its corresponding TS policy by penalizing under-sampled trajectories.

Neural Information Processing Systems http://nips.cc/

Reinforcement learning has achieved great success in many applications.

Tree Search (TS) is crucial to some of the most influential successes in reinforcement learning. Here, we tackle two major challenges with TS that limit its usability: \textit{distribution shift} and \textit{scalability}. We first discover and analyze a counter-intuitive phenomenon: action selection through TS and a pre-trained value function often leads to lower performance compared to the original pre-trained agent, even when having access to the exact state and reward in future steps. We show this is due to a distribution shift to areas where value estimates are highly inaccurate and analyze this effect using Extreme Value theory. To overcome this problem, we introduce a novel off-policy correction term that accounts for the mismatch between the pre-trained value and its corresponding TS policy by penalizing under-sampled trajectories.

Compared to on-policy counterparts, off-policy model-free deep reinforcement learning can improve data efficiency by repeatedly using the previously gathered data. However, off-policy learning becomes challenging when the discrepancy between the underlying distributions of the agent's policy and collected data increases. Although the well-studied importance sampling and off-policy policy gradient techniques were proposed to compensate for this discrepancy, they usually require a collection of long trajectories and induce additional problems such as vanishing/exploding gradients or discarding many useful experiences, which eventually increases the computational complexity. Moreover, their generalization to either continuous action domains or policies approximated by deterministic deep neural networks is strictly limited. To overcome these limitations, we introduce a novel policy similarity measure to mitigate the effects of such discrepancy in continuous control. Our method offers an adequate single-step off-policy correction that is applicable to deterministic policy networks. Theoretical and empirical studies demonstrate that it can achieve a "safe" off-policy learning and substantially improve the state-of-the-art by attaining higher returns in fewer steps than the competing methods through an effective schedule of the learning rate in Q-learning and policy optimization.

Attention-based sequential recommendation methods have demonstrated promising results by accurately capturing users' dynamic interests from historical interactions. In addition to generating superior user representations, recent studies have begun integrating reinforcement learning (RL) into these models. Framing sequential recommendation as an RL problem with reward signals, unlocks developing recommender systems (RS) that consider a vital aspect-incorporating direct user feedback in the form of rewards to deliver a more personalized experience. Nonetheless, employing RL algorithms presents challenges, including off-policy training, expansive combinatorial action spaces, and the scarcity of datasets with sufficient reward signals. Contemporary approaches have attempted to combine RL and sequential modeling, incorporating contrastive-based objectives and negative sampling strategies for training the RL component. In this study, we further emphasize the efficacy of contrastive-based objectives paired with augmentation to address datasets with extended horizons. Additionally, we recognize the potential instability issues that may arise during the application of negative sampling. These challenges primarily stem from the data imbalance prevalent in real-world datasets, which is a common issue in offline RL contexts. While our established baselines attempt to mitigate this through various techniques, instability remains an issue. Therefore, we introduce an enhanced methodology aimed at providing a more effective solution to these challenges.

While multi-agent trust region algorithms have achieved great success empirically in solving coordination tasks, most of them, however, suffer from a non-stationarity problem since agents update their policies simultaneously. In contrast, a sequential scheme that updates policies agent-by-agent provides another perspective and shows strong performance. However, sample inefficiency and lack of monotonic improvement guarantees for each agent are still the two significant challenges for the sequential scheme. In this paper, we propose the \textbf{A}gent-by-\textbf{a}gent \textbf{P}olicy \textbf{O}ptimization (A2PO) algorithm to improve the sample efficiency and retain the guarantees of monotonic improvement for each agent during training. We justify the tightness of the monotonic improvement bound compared with other trust region algorithms. From the perspective of sequentially updating agents, we further consider the effect of agent updating order and extend the theory of non-stationarity into the sequential update scheme. To evaluate A2PO, we conduct a comprehensive empirical study on four benchmarks: StarCraftII, Multi-agent MuJoCo, Multi-agent Particle Environment, and Google Research Football full game scenarios. A2PO consistently outperforms strong baselines.

Reinforcement learning has achieved great success in many applications. However, sample efficiency remains a key challenge, with prominent methods requiring millions (or even billions) of environment steps to train. Recently, there has been significant progress in sample efficient image-based RL algorithms; however, consistent human-level performance on the Atari game benchmark remains an elusive goal. We propose a sample efficient model-based visual RL algorithm built on MuZero, which we name EfficientZero. Our method achieves 194.3% mean human performance and 109.0% median performance on the Atari 100k benchmark with only two hours of real-time game experience and outperforms the state SAC in some tasks on the DMControl 100k benchmark. This is the first time an algorithm achieves super-human performance on Atari games with such little data. EfficientZero's performance is also close to DQN's performance at 200 million frames while we consume 500 times less data. EfficientZero's low sample complexity and high performance can bring RL closer to real-world applicability. We implement our algorithm in an easy-to-understand manner and it is available at https://github.com/YeWR/EfficientZero. We hope it will accelerate the research of MCTS-based RL algorithms in the wider community.

One of my favorite things about deep reinforcement learning is that, unlike supervised learning, it really, really doesn't want to work. Throwing a neural net at a computer vision problem might get you 80% of the way there. Throwing a neural net at an RL problem will probably blow something up in front of your face -- and it will blow up in a different way each time you try. A lot of the biggest challenges in RL revolve around two questions: how we interact with the environment effectively (e.g. In this post, I want to explore a few recent directions in deep RL research that attempt to address these challenges, and do so with particularly elegant parallels to human cognition. This post will begin with a quick review of two canonical deep RL algorithms -- DQN and A3C -- to provide us some intuitions to refer back to, and then jump into a deep dive on a few recent papers and breakthroughs in the categories described above.