Imagine if AI decision-support tools not only complemented our ability to make accurate decisions, but also improved our skills, boosted collaboration, and elevated the joy we derive from our tasks. Despite the potential to optimize a broad spectrum of such human-centric objectives, the design of current AI tools remains focused on decision accuracy alone. We propose offline reinforcement learning (RL) as a general approach for modeling human-AI decision-making to optimize human-AI interaction for diverse objectives. RL can optimize such objectives by tailoring decision support, providing the right type of assistance to the right person at the right time. We instantiated our approach with two objectives: human-AI accuracy on the decision-making task and human learning about the task and learned decision support policies from previous human-AI interaction data. We compared the optimized policies against several baselines in AI-assisted decision-making. Across two experiments (N=316 and N=964), our results demonstrated that people interacting with policies optimized for accuracy achieve significantly better accuracy -- and even human-AI complementarity -- compared to those interacting with any other type of AI support. Our results further indicated that human learning was more difficult to optimize than accuracy, with participants who interacted with learning-optimized policies showing significant learning improvement only at times. Our research (1) demonstrates offline RL to be a promising approach to model human-AI decision-making, leading to policies that may optimize human-centric objectives and provide novel insights about the AI-assisted decision-making space, and (2) emphasizes the importance of considering human-centric objectives beyond decision accuracy in AI-assisted decision-making, opening up the novel research challenge of optimizing human-AI interaction for such objectives.

To improve the efficiency of reinforcement learning, we propose a novel asynchronous federated reinforcement learning framework termed AFedPG, which constructs a global model through collaboration among $N$ agents using policy gradient (PG) updates. To handle the challenge of lagged policies in asynchronous settings, we design delay-adaptive lookahead and normalized update techniques that can effectively handle the heterogeneous arrival times of policy gradients. We analyze the theoretical global convergence bound of AFedPG, and characterize the advantage of the proposed algorithm in terms of both the sample complexity and time complexity. Specifically, our AFedPG method achieves $\mathcal{O}(\frac{{\epsilon}^{-2.5}}{N})$ sample complexity at each agent on average. Compared to the single agent setting with $\mathcal{O}(\epsilon^{-2.5})$ sample complexity, it enjoys a linear speedup with respect to the number of agents. Moreover, compared to synchronous FedPG, AFedPG improves the time complexity from $\mathcal{O}(\frac{t_{\max}}{N})$ to $\mathcal{O}(\frac{1}{\sum_{i=1}^{N} \frac{1}{t_{i}}})$, where $t_{i}$ denotes the time consumption in each iteration at the agent $i$, and $t_{\max}$ is the largest one. The latter complexity $\mathcal{O}(\frac{1}{\sum_{i=1}^{N} \frac{1}{t_{i}}})$ is always smaller than the former one, and this improvement becomes significant in large-scale federated settings with heterogeneous computing powers ($t_{\max}\gg t_{\min}$). Finally, we empirically verify the improved performances of AFedPG in three MuJoCo environments with varying numbers of agents. We also demonstrate the improvements with different computing heterogeneity.

We investigate multi-agent imitation learning (IL) within the framework of mean field games (MFGs), considering the presence of time-varying correlated signals. Existing MFG IL algorithms assume demonstrations are sampled from Mean Field Nash Equilibria (MFNE), limiting their adaptability to real-world scenarios. For example, in the traffic network equilibrium influenced by public routing recommendations, recommendations introduce time-varying correlated signals into the game, not captured by MFNE and other existing correlated equilibrium concepts. To address this gap, we propose Adaptive Mean Field Correlated Equilibrium (AMFCE), a general equilibrium incorporating time-varying correlated signals. We establish the existence of AMFCE under mild conditions and prove that MFNE is a subclass of AMFCE. We further propose Correlated Mean Field Imitation Learning (CMFIL), a novel IL framework designed to recover the AMFCE, accompanied by a theoretical guarantee on the quality of the recovered policy. Experimental results, including a real-world traffic flow prediction problem, demonstrate the superiority of CMFIL over state-of-the-art IL baselines, highlighting the potential of CMFIL in understanding large population behavior under correlated signals.

For multi-agent reinforcement learning systems (MARLS), the problem formulation generally involves investing massive reward engineering effort specific to a given problem. However, this effort often cannot be translated to other problems; worse, it gets wasted when system dynamics change drastically. This problem is further exacerbated in sparse reward scenarios, where a meaningful heuristic can assist in the policy convergence task. We propose GOVerned Reward Engineering Kernels (GOV-REK), which dynamically assign reward distributions to agents in MARLS during its learning stage. We also introduce governance kernels, which exploit the underlying structure in either state or joint action space for assigning meaningful agent reward distributions. During the agent learning stage, it iteratively explores different reward distribution configurations with a Hyperband-like algorithm to learn ideal agent reward models in a problem-agnostic manner. Our experiments demonstrate that our meaningful reward priors robustly jumpstart the learning process for effectively learning different MARL problems.

This research focuses on developing reinforcement learning approaches for the locomotion generation of small-size quadruped robots. The rat robot NeRmo is employed as the experimental platform. Due to the constrained volume, small-size quadruped robots typically possess fewer and weaker sensors, resulting in difficulty in accurately perceiving and responding to environmental changes. In this context, insufficient and imprecise feedback data from sensors makes it difficult to generate adaptive locomotion based on reinforcement learning. To overcome these challenges, this paper proposes a novel reinforcement learning approach that focuses on extracting effective perceptual information to enhance the environmental adaptability of small-size quadruped robots. According to the frequency of a robot's gait stride, key information of sensor data is analyzed utilizing sinusoidal functions derived from Fourier transform results. Additionally, a multifunctional reward mechanism is proposed to generate adaptive locomotion in different tasks. Extensive simulations are conducted to assess the effectiveness of the proposed reinforcement learning approach in generating rat robot locomotion in various environments. The experiment results illustrate the capability of the proposed approach to maintain stable locomotion of a rat robot across different terrains, including ramps, stairs, and spiral stairs.

Currently, there is a growing trend of outsourcing the execution of DNNs to cloud services. For service providers, managing multi-tenancy and ensuring high-quality service delivery, particularly in meeting stringent execution time constraints, assumes paramount importance, all while endeavoring to maintain cost-effectiveness. In this context, the utilization of heterogeneous multi-accelerator systems becomes increasingly relevant. This paper presents RELMAS, a low-overhead deep reinforcement learning algorithm designed for the online scheduling of DNNs in multi-tenant environments, taking into account the dataflow heterogeneity of accelerators and memory bandwidths contentions. By doing so, service providers can employ the most efficient scheduling policy for user requests, optimizing Service-Level-Agreement (SLA) satisfaction rates and enhancing hardware utilization. The application of RELMAS to a heterogeneous multi-accelerator system composed of various instances of Simba and Eyeriss sub-accelerators resulted in up to a 173% improvement in SLA satisfaction rate compared to state-of-the-art scheduling techniques across different workload scenarios, with less than a 1.5% energy overhead.

Model-based reinforcement learning is an effective approach for controlling an unknown system. It is based on a longstanding pipeline familiar to the control community in which one performs experiments on the environment to collect a dataset, uses the resulting dataset to identify a model of the system, and finally performs control synthesis using the identified model. As interacting with the system may be costly and time consuming, targeted exploration is crucial for developing an effective control-oriented model with minimal experimentation. Motivated by this challenge, recent work has begun to study finite sample data requirements and sample efficient algorithms for the problem of optimal exploration in model-based reinforcement learning. However, existing theory and algorithms are limited to model classes which are linear in the parameters. Our work instead focuses on models with nonlinear parameter dependencies, and presents the first finite sample analysis of an active learning algorithm suitable for a general class of nonlinear dynamics. In certain settings, the excess control cost of our algorithm achieves the optimal rate, up to logarithmic factors. We validate our approach in simulation, showcasing the advantage of active, control-oriented exploration for controlling nonlinear systems.

The Vehicle Routing Problem with Drones (VRPD) seeks to optimize the routing paths for both trucks and drones, where the trucks are responsible for delivering parcels to customer locations, and the drones are dispatched from these trucks for parcel delivery, subsequently being retrieved by the trucks. Given the NP-Hard complexity of VRPD, numerous heuristic approaches have been introduced. However, improving solution quality and reducing computation time remain significant challenges. In this paper, we conduct a comprehensive examination of heuristic methods designed for solving VRPD, distilling and standardizing them into core elements. We then develop a novel reinforcement learning (RL) framework that is seamlessly integrated with the heuristic solution components, establishing a set of universal principles for incorporating the RL framework with heuristic strategies in an aim to improve both the solution quality and computation speed. This integration has been applied to a state-of-the-art heuristic solution for VRPD, showcasing the substantial benefits of incorporating the RL framework. Our evaluation results demonstrated that the heuristic solution incorporated with our RL framework not only elevated the quality of solutions but also achieved rapid computation speeds, especially when dealing with extensive customer locations.

Deep reinforcement learning (DRL) has recently made remarkable progress in several application domains, such as games, finance, autonomous driving, and recommendation systems. However, the black-box nature of deep neural networks and the complex interaction among various factors raise challenges in understanding and interpreting the models' decision-making processes. This workshop brought together researchers, practitioners, and experts from both the DRL and the explainable AI communities to focus on methods, techniques, and frameworks that enhance the explainability and interpretability of DRL algorithms. The responsible language models (ReLM) workshop focused on the development, implementation, and applications of LMs aligned with responsible AI principles. Both theoretical and practical challenges regarding the design and deployment of responsible LMs were discussed, including bias identification and quantification, bias mitigation, transparency, privacy and security issues, hallucination, uncertainty quantification, and various other risks associated with LMs.

Preference based Reinforcement Learning (PbRL) removes the need to hand specify a reward function by learning a reward from preference feedback over policy behaviors. Current approaches to PbRL do not address the credit assignment problem inherent in determining which parts of a behavior most contributed to a preference, which result in data intensive approaches and subpar reward functions. We address such limitations by introducing a credit assignment strategy (Hindsight PRIOR) that uses a world model to approximate state importance within a trajectory and then guides rewards to be proportional to state importance through an auxiliary predicted return redistribution objective. Incorporating state importance into reward learning improves the speed of policy learning, overall policy performance, and reward recovery on both locomotion and manipulation tasks. For example, Hindsight PRIOR recovers on average significantly (p < 0.05) more reward on MetaWorld (20%) and DMC (15%). The performance gains and our ablations demonstrate the benefits even a simple credit assignment strategy can have on reward learning and that state importance in forward dynamics prediction is a strong proxy for a state's contribution to a preference decision. Code repository can be found at https://github.com/apple/ Preference-based reinforcement learning (PbRL) learns a policy from preference feedback removing the need to hand specify a reward function. Compared to other methods that avoid hand-specifying a reward function (e.g. Additionally, PbRL can be deployed as human-in-the-loop allowing guidance to adapt on-the-fly to sub-optimal policies, and has shown to be highly effective for complex tasks where reward specification is not feasible (e.g.