The recent advances in information technology and artificial intelligence have fueled a rapid expansion of the data center (DC) industry worldwide, accompanied by an immense appetite for electricity to power the DCs. In a typical DC, around 30~40% of the energy is spent on the cooling system rather than on computer servers, posing a pressing need for developing new energy-saving optimization technologies for DC cooling systems. However, optimizing such real-world industrial systems faces numerous challenges, including but not limited to a lack of reliable simulation environments, limited historical data, and stringent safety and control robustness requirements. In this work, we present a novel physics-informed offline reinforcement learning (RL) framework for energy efficiency optimization of DC cooling systems. The proposed framework models the complex dynamical patterns and physical dependencies inside a server room using a purposely designed graph neural network architecture that is compliant with the fundamental time-reversal symmetry. Because of its well-behaved and generalizable state-action representations, the model enables sample-efficient and robust latent space offline policy learning using limited real-world operational data. Our framework has been successfully deployed and verified in a large-scale production DC for closed-loop control of its air-cooling units (ACUs). We conducted a total of 2000 hours of short and long-term experiments in the production DC environment. The results show that our method achieves 14~21% energy savings in the DC cooling system, without any violation of the safety or operational constraints. Our results have demonstrated the significant potential of offline RL in solving a broad range of data-limited, safety-critical real-world industrial control problems.

Meta-reinforcement learning (Meta-RL) facilitates rapid adaptation to unseen tasks but faces challenges in long-horizon environments. Skill-based approaches tackle this by decomposing state-action sequences into reusable skills and employing hierarchical decision-making. However, these methods are highly susceptible to noisy offline demonstrations, resulting in unstable skill learning and degraded performance. To overcome this, we propose Prioritized Refinement for Skill-Based Meta-RL (PRISM), a robust framework that integrates exploration near noisy data to generate online trajectories and combines them with offline data. Through prioritization, PRISM extracts high-quality data to learn task-relevant skills effectively. By addressing the impact of noise, our method ensures stable skill learning and achieves superior performance in long-horizon tasks, even with noisy and sub-optimal data.

This paper analyzes multi-step temporal difference (TD)-learning algorithms within the ``deadly triad'' scenario, characterized by linear function approximation, off-policy learning, and bootstrapping. In particular, we prove that $n$-step TD-learning algorithms converge to a solution as the sampling horizon $n$ increases sufficiently. The paper is divided into two parts. In the first part, we comprehensively examine the fundamental properties of their model-based deterministic counterparts, including projected value iteration, gradient descent algorithms, which can be viewed as prototype deterministic algorithms whose analysis plays a pivotal role in understanding and developing their model-free reinforcement learning counterparts. In particular, we prove that these algorithms converge to meaningful solutions when $n$ is sufficiently large. Based on these findings, in the second part, two $n$-step TD-learning algorithms are proposed and analyzed, which can be seen as the model-free reinforcement learning counterparts of the model-based deterministic algorithms.

Reasoning LLMs such as OpenAI o1, o3 and DeepSeek R1 have made significant progress in mathematics and coding, yet find challenging advanced tasks such as International Mathematical Olympiad (IMO) combinatorics problems, Abstraction and Reasoning Corpus (ARC) puzzles, and Humanity's Last Exam (HLE) questions. We use a diverse inference approach that combines multiple models and methods at test time. We find that verifying mathematics and code problems, and rejection sampling on other problems is simple and effective. We automatically verify correctness of solutions to IMO problems by Lean, and ARC puzzles by code, and find that best-of-N effectively answers HLE questions. Our approach increases answer accuracy on IMO combinatorics problems from 33.3% to 77.8%, accuracy on HLE questions from 8% to 37%, and solves 80% of ARC puzzles that 948 humans could not and 26.5% of ARC puzzles that o3 high compute does not. Test-time simulations, reinforcement learning, and meta-learning with inference feedback improve generalization by adapting agent graph representations and varying prompts, code, and datasets. Our approach is reliable, robust, and scalable, and in the spirit of reproducible research, we will make it publicly available upon publication.

This work addresses the problem of multi-robot coordination under unknown robot transition models, ensuring that tasks specified by Time Window Temporal Logic are satisfied with user-defined probability thresholds. We present a bi-level framework that integrates (i) high-level task allocation, where tasks are assigned based on the robots' estimated task completion probabilities and expected rewards, and (ii) low-level distributed policy learning and execution, where robots independently optimize auxiliary rewards while fulfilling their assigned tasks. To handle uncertainty in robot dynamics, our approach leverages real-time task execution data to iteratively refine expected task completion probabilities and rewards, enabling adaptive task allocation without explicit robot transition models. We theoretically validate the proposed algorithm, demonstrating that the task assignments meet the desired probability thresholds with high confidence. Finally, we demonstrate the effectiveness of our framework through comprehensive simulations.

Current Reinforcement Learning (RL) methods often suffer from sample-inefficiency, resulting from blind exploration strategies that neglect causal relationships among states, actions, and rewards. Although recent causal approaches aim to address this problem, they lack grounded modeling of reward-guided causal understanding of states and actions for goal-orientation, thus impairing learning efficiency. To tackle this issue, we propose a novel method named Causal Information Prioritization (CIP) that improves sample efficiency by leveraging factored MDPs to infer causal relationships between different dimensions of states and actions with respect to rewards, enabling the prioritization of causal information. Specifically, CIP identifies and leverages causal relationships between states and rewards to execute counterfactual data augmentation to prioritize high-impact state features under the causal understanding of the environments. Moreover, CIP integrates a causality-aware empowerment learning objective, which significantly enhances the agent's execution of reward-guided actions for more efficient exploration in complex environments. To fully assess the effectiveness of CIP, we conduct extensive experiments across 39 tasks in 5 diverse continuous control environments, encompassing both locomotion and manipulation skills learning with pixel-based and sparse reward settings. Experimental results demonstrate that CIP consistently outperforms existing RL methods across a wide range of scenarios.

A key objective of embodied intelligence is enabling agents to perform long-horizon tasks in dynamic environments while maintaining robust decision-making and adaptability. To achieve this goal, we propose the Spatio-Temporal Memory Agent (STMA), a novel framework designed to enhance task planning and execution by integrating spatio-temporal memory. STMA is built upon three critical components: (1) a spatio-temporal memory module that captures historical and environmental changes in real time, (2) a dynamic knowledge graph that facilitates adaptive spatial reasoning, and (3) a planner-critic mechanism that iteratively refines task strategies. We evaluate STMA in the TextWorld environment on 32 tasks, involving multi-step planning and exploration under varying levels of complexity. Experimental results demonstrate that STMA achieves a 31.25% improvement in success rate and a 24.7% increase in average score compared to the state-of-the-art model. The results highlight the effectiveness of spatio-temporal memory in advancing the memory capabilities of embodied agents.

Dynamic Reinforcement Learning (Dynamic RL), proposed in this paper, directly controls system dynamics, instead of the actor (action-generating neural network) outputs at each moment, bringing about a major qualitative shift in reinforcement learning (RL) from static to dynamic. The actor is initially designed to generate chaotic dynamics through the loop with its environment, enabling the agent to perform flexible and deterministic exploration. Dynamic RL controls global system dynamics using a local index called "sensitivity," which indicates how much the input neighborhood contracts or expands into the corresponding output neighborhood through each neuron's processing. While sensitivity adjustment learning (SAL) prevents excessive convergence of the dynamics, sensitivity-controlled reinforcement learning (SRL) adjusts them -- to converge more to improve reproducibility around better state transitions with positive TD error and to diverge more to enhance exploration around worse transitions with negative TD error. Dynamic RL was applied only to the actor in an Actor-Critic RL architecture while applying it to the critic remains a challenge. It was tested on two dynamic tasks and functioned effectively without external exploration noise or backward computation through time. Moreover, it exhibited excellent adaptability to new environments, although some problems remain. Drawing parallels between 'exploration' and 'thinking,' the author hypothesizes that "exploration grows into thinking through learning" and believes this RL could be a key technique for the emergence of thinking, including inspiration that cannot be reconstructed from massive existing text data. Finally, despite being presumptuous, the author presents the argument that this research should not proceed due to its potentially fatal risks, aiming to encourage discussion.

Reinforcement Learning (RL) has been widely used in many applications, particularly in gaming, which serves as an excellent training ground for AI models. Google DeepMind has pioneered innovations in this field, employing reinforcement learning algorithms, including model-based, model-free, and deep Q-network approaches, to create advanced AI models such as AlphaGo, AlphaGo Zero, and MuZero. AlphaGo, the initial model, integrates supervised learning and reinforcement learning to master the game of Go, surpassing professional human players. AlphaGo Zero refines this approach by eliminating reliance on human gameplay data, instead utilizing self-play for enhanced learning efficiency. MuZero further extends these advancements by learning the underlying dynamics of game environments without explicit knowledge of the rules, achieving adaptability across various games, including complex Atari games. This paper reviews the significance of reinforcement learning applications in Atari and strategy-based games, analyzing these three models, their key innovations, training processes, challenges encountered, and improvements made. Additionally, we discuss advancements in the field of gaming, including MiniZero and multi-agent models, highlighting future directions and emerging AI models from Google DeepMind.

Traversing risky terrains with sparse footholds poses a significant challenge for humanoid robots, requiring precise foot placements and stable locomotion. Existing approaches designed for quadrupedal robots often fail to generalize to humanoid robots due to differences in foot geometry and unstable morphology, while learning-based approaches for humanoid locomotion still face great challenges on complex terrains due to sparse foothold reward signals and inefficient learning processes. To address these challenges, we introduce BeamDojo, a reinforcement learning (RL) framework designed for enabling agile humanoid locomotion on sparse footholds. BeamDojo begins by introducing a sampling-based foothold reward tailored for polygonal feet, along with a double critic to balancing the learning process between dense locomotion rewards and sparse foothold rewards. To encourage sufficient trail-and-error exploration, BeamDojo incorporates a two-stage RL approach: the first stage relaxes the terrain dynamics by training the humanoid on flat terrain while providing it with task terrain perceptive observations, and the second stage fine-tunes the policy on the actual task terrain. Moreover, we implement a onboard LiDAR-based elevation map to enable real-world deployment. Extensive simulation and real-world experiments demonstrate that BeamDojo achieves efficient learning in simulation and enables agile locomotion with precise foot placement on sparse footholds in the real world, maintaining a high success rate even under significant external disturbances.