Learning predictive world models is essential for enhancing the planning capabilities of reinforcement learning agents. Notably, the MuZero-style algorithms, based on the value equivalence principle and Monte Carlo Tree Search (MCTS), have achieved superhuman performance in various domains. However, in environments that require capturing long-term dependencies, MuZero's performance deteriorates rapidly. We identify that this is partially due to the \textit{entanglement} of latent representations with historical information, which results in incompatibility with the auxiliary self-supervised state regularization. To overcome this limitation, we present \textit{UniZero}, a novel approach that \textit{disentangles} latent states from implicit latent history using a transformer-based latent world model. By concurrently predicting latent dynamics and decision-oriented quantities conditioned on the learned latent history, UniZero enables joint optimization of the long-horizon world model and policy, facilitating broader and more efficient planning in latent space. We demonstrate that UniZero, even with single-frame inputs, matches or surpasses the performance of MuZero-style algorithms on the Atari 100k benchmark. Furthermore, it significantly outperforms prior baselines in benchmarks that require long-term memory. Lastly, we validate the effectiveness and scalability of our design choices through extensive ablation studies, visual analyses, and multi-task learning results. The code is available at \textcolor{magenta}{https://github.com/opendilab/LightZero}.

Often times in imitation learning (IL), the environment we collect expert demonstrations in and the environment we want to deploy our learned policy in aren't exactly the same (e.g. demonstrations collected in simulation but deployment in the real world). Compared to policy-centric approaches to IL like behavioural cloning, reward-centric approaches like inverse reinforcement learning (IRL) often better replicate expert behaviour in new environments. This transfer is usually performed by optimising the recovered reward under the dynamics of the target environment. However, (a) we find that modern deep IL algorithms frequently recover rewards which induce policies far weaker than the expert, even in the same environment the demonstrations were collected in. Furthermore, (b) these rewards are often quite poorly shaped, necessitating extensive environment interaction to optimise effectively. We provide simple and scalable fixes to both of these concerns. For (a), we find that reward model ensembles combined with a slightly different training objective significantly improves re-training and transfer performance. For (b), we propose a novel evolution-strategies based method EvIL to optimise for a reward-shaping term that speeds up re-training in the target environment, closing a gap left open by the classical theory of IRL. On a suite of continuous control tasks, we are able to re-train policies in target (and source) environments more interaction-efficiently than prior work.

Autocomplete suggestions are fundamental to modern text entry systems, with applications in domains such as messaging and email composition. Typically, autocomplete suggestions are generated from a language model with a confidence threshold. However, this threshold does not directly take into account the cognitive load imposed on the user by surfacing suggestions, such as the effort to switch contexts from typing to reading the suggestion, and the time to decide whether to accept the suggestion. In this paper, we study the problem of improving inline autocomplete suggestions in text entry systems via a sequential decision-making formulation, and use reinforcement learning to learn suggestion policies through repeated interactions with a target user over time. This formulation allows us to factor cognitive load into the objective of training an autocomplete model, through a reward function based on text entry speed. We acquired theoretical and experimental evidence that, under certain objectives, the sequential decision-making formulation of the autocomplete problem provides a better suggestion policy than myopic single-step reasoning. However, aligning these objectives with real users requires further exploration. In particular, we hypothesize that the objectives under which sequential decision-making can improve autocomplete systems are not tailored solely to text entry speed, but more broadly to metrics such as user satisfaction and convenience.

While recent advances in neural radiance field enable realistic digitization for large-scale scenes, the image-capturing process is still time-consuming and labor-intensive. Previous works attempt to automate this process using the Next-Best-View (NBV) policy for active 3D reconstruction. However, the existing NBV policies heavily rely on hand-crafted criteria, limited action space, or per-scene optimized representations. These constraints limit their cross-dataset generalizability. To overcome them, we propose GenNBV, an end-to-end generalizable NBV policy. Our policy adopts a reinforcement learning (RL)-based framework and extends typical limited action space to 5D free space. It empowers our agent drone to scan from any viewpoint, and even interact with unseen geometries during training. To boost the cross-dataset generalizability, we also propose a novel multi-source state embedding, including geometric, semantic, and action representations. We establish a benchmark using the Isaac Gym simulator with the Houses3K and OmniObject3D datasets to evaluate this NBV policy. Experiments demonstrate that our policy achieves a 98.26% and 97.12% coverage ratio on unseen building-scale objects from these datasets, respectively, outperforming prior solutions.

This article presents a digital twin (DT)-enhanced reinforcement learning (RL) framework aimed at optimizing performance and reliability in network resource management, since the traditional RL methods face several unified challenges when applied to physical networks, including limited exploration efficiency, slow convergence, poor long-term performance, and safety concerns during the exploration phase. To deal with the above challenges, a comprehensive DT-based framework is proposed to enhance the convergence speed and performance for unified RL-based resource management. The proposed framework provides safe action exploration, more accurate estimates of long-term returns, faster training convergence, higher convergence performance, and real-time adaptation to varying network conditions. Then, two case studies on ultra-reliable and low-latency communication (URLLC) services and multiple unmanned aerial vehicles (UAV) network are presented, demonstrating improvements of the proposed framework in performance, convergence speed, and training cost reduction both on traditional RL and neural network based Deep RL (DRL). Finally, the article identifies and explores some of the research challenges and open issues in this rapidly evolving field.

Training corpuses for vision language models (VLMs) typically lack sufficient amounts of decision-centric data. This renders off-the-shelf VLMs sub-optimal for decision-making tasks such as in-the-wild device control through graphical user interfaces (GUIs). While training with static demonstrations has shown some promise, we show that such methods fall short for controlling real GUIs due to their failure to deal with real-world stochasticity and non-stationarity not captured in static observational data. This paper introduces a novel autonomous RL approach, called DigiRL, for training in-the-wild device control agents through fine-tuning a pre-trained VLM in two stages: offline RL to initialize the model, followed by offline-to-online RL. To do this, we build a scalable and parallelizable Android learning environment equipped with a VLM-based evaluator and develop a simple yet effective RL approach for learning in this domain. Our approach runs advantage-weighted RL with advantage estimators enhanced to account for stochasticity along with an automatic curriculum for deriving maximal learning signal. We demonstrate the effectiveness of DigiRL using the Android-in-the-Wild (AitW) dataset, where our 1.3B VLM trained with RL achieves a 49.5% absolute improvement -- from 17.7 to 67.2% success rate -- over supervised fine-tuning with static human demonstration data. These results significantly surpass not only the prior best agents, including AppAgent with GPT-4V (8.3% success rate) and the 17B CogAgent trained with AitW data (38.5%), but also the prior best autonomous RL approach based on filtered behavior cloning (57.8%), thereby establishing a new state-of-the-art for digital agents for in-the-wild device control.

Neural network supported tree-search has shown strong results in a variety of perfect information multi-agent tasks. However, the performance of these methods on partial information games has generally been below competing approaches. Here we study the class of simultaneous-move games, which are a subclass of partial information games which are most similar to perfect information games: both agents know the game state with the exception of the opponent's move, which is revealed only after each agent makes its own move. Simultaneous move games include popular benchmarks such as Google Research Football and Starcraft. In this study we answer the question: can we take tree search algorithms trained through self-play from perfect information settings and adapt them to simultaneous move games without significant loss of performance? We answer this question by deriving a practical method that attempts to approximate a coarse correlated equilibrium as a subroutine within a tree search. Our algorithm works on cooperative, competitive, and mixed tasks. Our results are better than the current best MARL algorithms on a wide range of accepted baseline environments.

Sparse matrix-matrix multiplication (SpGEMM) is a critical operation in numerous fields, including scientific computing, graph analytics, and deep learning. These applications exploit the sparsity of matrices to reduce storage and computational demands. However, the irregular structure of sparse matrices poses significant challenges for performance optimization. Traditional hardware accelerators are tailored for specific sparsity patterns with fixed dataflow schemes - inner, outer, and row-wise but often perform suboptimally when the actual sparsity deviates from these predetermined patterns. As the use of SpGEMM expands across various domains, each with distinct sparsity characteristics, the demand for hardware accelerators that can efficiently handle a range of sparsity patterns is increasing. This paper presents a machine learning based approach for adaptively selecting the most appropriate dataflow scheme for SpGEMM tasks with diverse sparsity patterns. By employing decision trees and deep reinforcement learning, we explore the potential of these techniques to surpass heuristic-based methods in identifying optimal dataflow schemes. We evaluate our models by comparing their performance with that of a heuristic, highlighting the strengths and weaknesses of each approach. Our findings suggest that using machine learning for dynamic dataflow selection in hardware accelerators can provide upto 28 times gains.

One of the key arguments for building robots that have similar form factors to human beings is that we can leverage the massive human data for training. Yet, doing so has remained challenging in practice due to the complexities in humanoid perception and control, lingering physical gaps between humanoids and humans in morphologies and actuation, and lack of a data pipeline for humanoids to learn autonomous skills from egocentric vision. In this paper, we introduce a full-stack system for humanoids to learn motion and autonomous skills from human data. We first train a low-level policy in simulation via reinforcement learning using existing 40-hour human motion datasets. This policy transfers to the real world and allows humanoid robots to follow human body and hand motion in real time using only a RGB camera, i.e. shadowing. Through shadowing, human operators can teleoperate humanoids to collect whole-body data for learning different tasks in the real world. Using the data collected, we then perform supervised behavior cloning to train skill policies using egocentric vision, allowing humanoids to complete different tasks autonomously by imitating human skills. We demonstrate the system on our customized 33-DoF 180cm humanoid, autonomously completing tasks such as wearing a shoe to stand up and walk, unloading objects from warehouse racks, folding a sweatshirt, rearranging objects, typing, and greeting another robot with 60-100% success rates using up to 40 demonstrations. Project website: https://humanoid-ai.github.io/

Chronic pain significantly diminishes the quality of life for millions worldwide. While psychoeducation and therapy can improve pain outcomes, many individuals experiencing pain lack access to evidence-based treatments or fail to complete the necessary number of sessions to achieve benefit. Reinforcement learning (RL) shows potential in tailoring personalized pain management interventions according to patients' individual needs while ensuring the efficient use of scarce clinical resources. However, clinicians, patients, and healthcare decision-makers are concerned that RL solutions could exacerbate disparities associated with patient characteristics like race or gender. In this article, we study gender fairness in personalized pain care recommendations using a real-world application of reinforcement learning (Piette et al., 2022a). Here, adhering to gender fairness translates to minimal or no disparity in the utility received by subpopulations as defined by gender. We investigate whether the selection of relevant patient information (referred to as features) used to assist decision-making affects gender fairness. Our experiments, conducted using real-world data (Piette, 2022), indicate that included features can impact gender fairness. Moreover, we propose an RL solution, NestedRecommendation, that demonstrates the ability: i) to adaptively learn to select the features that optimize for utility and fairness, and ii) to accelerate feature selection and in turn, improve pain care recommendations from early on, by leveraging clinicians' domain expertise.