Alignment is crucial for training large language models. The predominant strategy is Reinforcement Learning from Human Feedback (RLHF), with Proximal Policy Optimization (PPO) as the de-facto algorithm. Yet, PPO is known to struggle with computational inefficiency, a challenge that this paper aims to address. We identify three important properties of RLHF tasks: fast simulation, deterministic transitions, and trajectory-level rewards, which are not leveraged in PPO. Based on these properties, we develop ReMax, a new algorithm tailored for RLHF. The design of ReMax builds on the celebrated algorithm REINFORCE but is enhanced with a new variance-reduction technique. ReMax offers threefold advantages over PPO: first, it is simple to implement with just 6 lines of code. It further eliminates more than 4 hyper-parameters in PPO, which are laborious to tune. Second, ReMax reduces memory usage by about 50%. To illustrate, PPO runs out of memory when fine-tuning a Llama2-7B model on A100-80GB GPUs, whereas ReMax can support the training. Even though memory-efficient techniques (e.g., ZeRO and offload) are employed for PPO to afford training, ReMax can utilize a larger batch size to increase throughput. Third, in terms of wall-clock time, PPO is about twice as slow as ReMax per iteration. Importantly, these improvements do not sacrifice task performance. We hypothesize that these advantages can be maintained in larger-scale models.

We derive a concentration bound of the type `for all $n \geq n_0$ for some $n_0$' for TD(0) with linear function approximation. We work with online TD learning with samples from a single sample path of the underlying Markov chain. This makes our analysis significantly different from offline TD learning or TD learning with access to independent samples from the stationary distribution of the Markov chain. We treat TD(0) as a contractive stochastic approximation algorithm, with both martingale and Markov noises. Markov noise is handled using the Poisson equation and the lack of almost sure guarantees on boundedness of iterates is handled using the concept of relaxed concentration inequalities.

The prevalence of multi-agent applications pervades various interconnected systems in our everyday lives. Despite their ubiquity, the integration and development of intelligent decision-making agents in a shared environment pose challenges to their effective implementation. This survey delves into the domain of multi-agent systems (MAS), placing a specific emphasis on unraveling the intricacies of learning optimal control within the MAS framework, commonly known as multi-agent reinforcement learning (MARL). The objective of this survey is to provide comprehensive insights into various dimensions of MAS, shedding light on myriad opportunities while highlighting the inherent challenges that accompany multi-agent applications. We hope not only to contribute to a deeper understanding of the MAS landscape but also to provide valuable perspectives for both researchers and practitioners. By doing so, we aim to facilitate informed exploration and foster development within the dynamic realm of MAS, recognizing the need for adaptive strategies and continuous evolution in addressing emerging complexities in MARL.

Reinforcement learning (RL) is a powerful tool for optimal control that has found great success in Atari games, the game of Go, robotic control, and building optimization. RL is also very brittle; agents often overfit to their training environment and fail to generalize to new settings. Unsupervised environment design (UED) has been proposed as a solution to this problem, in which the agent trains in environments that have been specially selected to help it learn. Previous UED algorithms focus on trying to train an RL agent that generalizes across a large distribution of environments. This is not necessarily desirable when we wish to prioritize performance in one environment over others. In this work, we will be examining the setting of robust RL building control, where we wish to train an RL agent that prioritizes performing well in normal weather while still being robust to extreme weather conditions. We demonstrate a novel UED algorithm, ActivePLR, that uses uncertainty-aware neural network architectures to generate new training environments at the limit of the RL agent's ability while being able to prioritize performance in a desired base environment. We show that ActivePLR is able to outperform state-of-the-art UED algorithms in minimizing energy usage while maximizing occupant comfort in the setting of building control.

Despite remarkable achievements in artificial intelligence, the deployability of learning-enabled systems in high-stakes real-world environments still faces persistent challenges. For example, in safety-critical domains like autonomous driving, robotic manipulation, and healthcare, it is crucial not only to achieve high performance but also to comply with given constraints. Furthermore, adaptability becomes paramount in non-stationary domains, where environmental parameters are subject to change. While safety and adaptability are recognized as key qualities for the new generation of AI, current approaches have not demonstrated effective adaptable performance in constrained settings. Hence, this paper breaks new ground by studying the unique challenges of ensuring safety in non-stationary environments by solving constrained problems through the lens of the meta-learning approach (learning-to-learn). While unconstrained meta-learning al-ready encounters complexities in end-to-end differentiation of the loss due to the bi-level nature, its constrained counterpart introduces an additional layer of difficulty, since the constraints imposed on task-level updates complicate the differentiation process. To address the issue, we first employ successive convex-constrained policy updates across multiple tasks with differentiable convexprogramming, which allows meta-learning in constrained scenarios by enabling end-to-end differentiation. This approach empowers the agent to rapidly adapt to new tasks under non-stationarity while ensuring compliance with safety constraints.

Multi-Agent Reinforcement Learning (MARL) has emerged as a foundational approach for addressing diverse, intelligent control tasks, notably in autonomous driving within the Internet of Vehicles (IoV) domain. However, the widely assumed existence of a central node for centralized, federated learning-assisted MARL might be impractical in highly dynamic environments. This can lead to excessive communication overhead, potentially overwhelming the IoV system. To address these challenges, we design a novel communication-efficient and policy collaboration algorithm for MARL under the frameworks of Soft Actor-Critic (SAC) and Decentralized Federated Learning (DFL), named RSM-MASAC, within a fully distributed architecture. In particular, RSM-MASAC enhances multi-agent collaboration and prioritizes higher communication efficiency in dynamic IoV system by incorporating the concept of segmented aggregation in DFL and augmenting multiple model replicas from received neighboring policy segments, which are subsequently employed as reconstructed referential policies for mixing. Distinctively diverging from traditional RL approaches, with derived new bounds under Maximum Entropy Reinforcement Learning (MERL), RSM-MASAC adopts a theory-guided mixture metric to regulate the selection of contributive referential policies to guarantee the soft policy improvement during communication phase. Finally, the extensive simulations in mixed-autonomy traffic control scenarios verify the effectiveness and superiority of our algorithm.

Answering complex natural language questions often necessitates multi-step reasoning and integrating external information. Several systems have combined knowledge retrieval with a large language model (LLM) to answer such questions. These systems, however, suffer from various failure cases, and we cannot directly train them end-to-end to fix such failures, as interaction with external knowledge is non-differentiable. To address these deficiencies, we define a ReAct-style LLM agent with the ability to reason and act upon external knowledge. We further refine the agent through a ReST-like method that iteratively trains on previous trajectories, employing growing-batch reinforcement learning with AI feedback for continuous self-improvement and self-distillation. Starting from a prompted large model and after just two iterations of the algorithm, we can produce a fine-tuned small model that achieves comparable performance on challenging compositional question-answering benchmarks with two orders of magnitude fewer parameters.

Unmanned aerial vehicle (UAV)-assisted sensor networks (UASNets), which play a crucial role in creating new opportunities, are experiencing significant growth in civil applications worldwide. UASNets improve disaster management through timely surveillance and advance precision agriculture with detailed crop monitoring, thereby significantly transforming the commercial economy. UASNets revolutionize the commercial sector by offering greater efficiency, safety, and cost-effectiveness, highlighting their transformative impact. A fundamental aspect of these new capabilities and changes is the collection of data from rugged and remote areas. Due to their excellent mobility and maneuverability, UAVs are employed to collect data from ground sensors in harsh environments, such as natural disaster monitoring, border surveillance, and emergency response monitoring. One major challenge in these scenarios is that the movements of UAVs affect channel conditions and result in packet loss. Fast movements of UAVs lead to poor channel conditions and rapid signal degradation, resulting in packet loss. On the other hand, slow mobility of a UAV can cause buffer overflows of the ground sensors, as newly arrived data is not promptly collected by the UAV. Our proposal to address this challenge is to minimize packet loss by jointly optimizing the velocity controls and data collection schedules of multiple UAVs.Furthermore, in UASNets, swift movements of UAVs result in poor channel conditions and fast signal attenuation, leading to an extended age of information (AoI). In contrast, slow movements of UAVs prolong flight time, thereby extending the AoI of ground sensors.To address this challenge, we propose a new mean-field flight resource allocation optimization to minimize the AoI of sensory data.

Peer learning is a novel high-level reinforcement learning framework for agents learning in groups. While standard reinforcement learning trains an individual agent in trial-and-error fashion, all on its own, peer learning addresses a related setting in which a group of agents, i.e., peers, learns to master a task simultaneously together from scratch. Peers are allowed to communicate only about their own states and actions recommended by others: "What would you do in my situation?". Our motivation is to study the learning behavior of these agents. We formalize the teacher selection process in the action advice setting as a multi-armed bandit problem and therefore highlight the need for exploration. Eventually, we analyze the learning behavior of the peers and observe their ability to rank the agents' performance within the study group and understand which agents give reliable advice. Further, we compare peer learning with single agent learning and a state-of-the-art action advice baseline. We show that peer learning is able to outperform single-agent learning and the baseline in several challenging discrete and continuous OpenAI Gym domains. Doing so, we also show that within such a framework complex policies from action recommendations beyond discrete action spaces can evolve.

We present a modular approach to \emph{reinforcement learning} (RL) in environments consisting of simpler components evolving in parallel. A monolithic view of such modular environments may be prohibitively large to learn, or may require unrealizable communication between the components in the form of a centralized controller. Our proposed approach is based on the assume-guarantee paradigm where the optimal control for the individual components is synthesized in isolation by making \emph{assumptions} about the behaviors of neighboring components, and providing \emph{guarantees} about their own behavior. We express these \emph{assume-guarantee contracts} as regular languages and provide automatic translations to scalar rewards to be used in RL. By combining local probabilities of satisfaction for each component, we provide a lower bound on the probability of satisfaction of the complete system. By solving a Markov game for each component, RL can produce a controller for each component that maximizes this lower bound. The controller utilizes the information it receives through communication, observations, and any knowledge of a coarse model of other agents. We experimentally demonstrate the efficiency of the proposed approach on a variety of case studies.