Reinforcement learning (RL) and Deep Reinforcement Learning (DRL), in particular, have the potential to disrupt and are already changing the way we interact with the world. One of the key indicators of their applicability is their ability to scale and work in real-world scenarios, that is in large-scale problems. This scale can be achieved via a combination of factors, the algorithm's ability to make use of large amounts of data and computational resources and the efficient exploration of the environment for viable solutions (i.e. policies). In this work, we investigate and motivate some theoretical foundations for deep reinforcement learning. We start with exact dynamic programming and work our way up to stochastic approximations and stochastic approximations for a model-free scenario, which forms the theoretical basis of modern reinforcement learning. We present an overview of this highly varied and rapidly changing field from the perspective of Approximate Dynamic Programming. We then focus our study on the short-comings with respect to exploration of the cornerstone approaches (i.e. DQN, DDQN, A2C) in deep reinforcement learning. On the theory side, our main contribution is the proposal of a novel Bayesian actor-critic algorithm. On the empirical side, we evaluate Bayesian exploration as well as actor-critic algorithms on standard benchmarks as well as state-of-the-art evaluation suites and show the benefits of both of these approaches over current state-of-the-art deep RL methods. We release all the implementations and provide a full python library that is easy to install and hopefully will serve the reinforcement learning community in a meaningful way, and provide a strong foundation for future work.

Combinatorial assembly uses standardized unit primitives to build objects that satisfy user specifications. Lego is a widely used platform for combinatorial assembly, in which people use unit primitives (ie Lego bricks) to build highly customizable 3D objects. This paper studies sequence planning for physical combinatorial assembly using Lego. Given the shape of the desired object, we want to find a sequence of actions for placing Lego bricks to build the target object. In particular, we aim to ensure the planned assembly sequence is physically executable. However, assembly sequence planning (ASP) for combinatorial assembly is particularly challenging due to its combinatorial nature, ie the vast number of possible combinations and complex constraints. To address the challenges, we employ deep reinforcement learning to learn a construction policy for placing unit primitives sequentially to build the desired object. Specifically, we design an online physics-aware action mask that efficiently filters out invalid actions and guides policy learning. In the end, we demonstrate that the proposed method successfully plans physically valid assembly sequences for constructing different Lego structures. The generated construction plan can be executed in real.

There has been a growing interest in multi-agent simulators in the domain of economic modeling. However, contemporary research often involves developing reinforcement learning (RL) based models that focus solely on a single type of agents, such as households, firms, or the government. Such an approach overlooks the adaptation of interacting agents thereby failing to capture the complexity of real-world economic systems. In this work, we consider a multi-agent simulator comprised of RL agents of numerous types, including heterogeneous households, firm, central bank and government. In particular, we focus on the crucial role of the government in distributing tax credits to households. We conduct two broad categories of comprehensive experiments dealing with the impact of tax credits on 1) households with varied degrees of myopia (short-sightedness in spending and saving decisions), and 2) households with diverse liquidity profiles. The first category of experiments examines the impact of the frequency of tax credits (e.g. annual vs quarterly) on consumption patterns of myopic households. The second category of experiments focuses on the impact of varying tax credit distribution strategies on households with differing liquidities. We validate our simulation model by reproducing trends observed in real households upon receipt of unforeseen, uniform tax credits, as documented in a JPMorgan Chase report. Based on the results of the latter, we propose an innovative tax credit distribution strategy for the government to reduce inequality among households. We demonstrate the efficacy of this strategy in improving social welfare in our simulation results.

The restless multi-armed bandit (RMAB) framework is a popular model with applications across a wide variety of fields. However, its solution is hindered by the exponentially growing state space (with respect to the number of arms) and the combinatorial action space, making traditional reinforcement learning methods infeasible for large-scale instances. In this paper, we propose GINO-Q, a three-timescale stochastic approximation algorithm designed to learn an asymptotically optimal index policy for RMABs. GINO-Q mitigates the curse of dimensionality by decomposing the RMAB into a series of subproblems, each with the same dimension as a single arm, ensuring that complexity increases linearly with the number of arms. Unlike recently developed Whittle-index-based algorithms, GINO-Q does not require RMABs to be indexable, enhancing its flexibility and applicability. Our experimental results demonstrate that GINO-Q consistently learns near-optimal policies, even for non-indexable RMABs where Whittle-index-based algorithms perform poorly, and it converges significantly faster than existing baselines.

As urban residents demand higher travel quality, vehicle dispatch has become a critical component of online ride-hailing services. However, current vehicle dispatch systems struggle to navigate the complexities of urban traffic dynamics, including unpredictable traffic conditions, diverse driver behaviors, and fluctuating supply and demand patterns. These challenges have resulted in travel difficulties for passengers in certain areas, while many drivers in other areas are unable to secure orders, leading to a decline in the overall quality of urban transportation services. To address these issues, this paper introduces GARLIC: a framework of GPT-Augmented Reinforcement Learning with Intelligent Control for vehicle dispatching. GARLIC utilizes multiview graphs to capture hierarchical traffic states, and learns a dynamic reward function that accounts for individual driving behaviors. The framework further integrates a GPT model trained with a custom loss function to enable high-precision predictions and optimize dispatching policies in real-world scenarios. Experiments conducted on two real-world datasets demonstrate that GARLIC effectively aligns with driver behaviors while reducing the empty load rate of vehicles.

With the aim of improving performance in Markov Decision Problem in an Off-Policy setting, we suggest taking inspiration from what is done in Offline Reinforcement Learning (RL). In Offline RL, it is a common practice during policy learning to maintain proximity to a reference policy to mitigate uncertainty, reduce potential policy errors, and help improve performance. We find ourselves in a different setting, yet it raises questions about whether a similar concept can be applied to enhance performance ie, whether it is possible to find a guiding policy capable of contributing to performance improvement, and how to incorporate it into our RL agent. Our attention is particularly focused on algorithms based on Monte Carlo Tree Search (MCTS) as a guide.MCTS renowned for its state-of-the-art capabilities across various domains, catches our interest due to its ability to converge to equilibrium in single-player and two-player contexts. By harnessing the power of MCTS as a guide for our RL agent, we observed a significant performance improvement, surpassing the outcomes achieved by utilizing each method in isolation. Our experiments were carried out on the Atari 100k benchmark.

Assigning orders to drivers under localized spatiotemporal context (micro-view order-dispatching) is a major task in Didi, as it influences ride-hailing service experience. Existing industrial solutions mainly follow a two-stage pattern that incorporate heuristic or learning-based algorithms with naive combinatorial methods, tackling the uncertainty of both sides' behaviors, including emerging timings, spatial relationships, and travel duration, etc. In this paper, we propose a one-stage end-to-end reinforcement learning based order-dispatching approach that solves behavior prediction and combinatorial optimization uniformly in a sequential decision-making manner. Specifically, we employ a two-layer Markov Decision Process framework to model this problem, and present \underline{D}eep \underline{D}ouble \underline{S}calable \underline{N}etwork (D2SN), an encoder-decoder structure network to generate order-driver assignments directly and stop assignments accordingly. Besides, by leveraging contextual dynamics, our approach can adapt to the behavioral patterns for better performance. Extensive experiments on Didi's real-world benchmarks justify that the proposed approach significantly outperforms competitive baselines in optimizing matching efficiency and user experience tasks. In addition, we evaluate the deployment outline and discuss the gains and experiences obtained during the deployment tests from the view of large-scale engineering implementation.

Reinforcement Learning (RL) is a potent tool for sequential decision-making and has achieved performance surpassing human capabilities across many challenging real-world tasks. As the extension of RL in the multi-agent system domain, multi-agent RL (MARL) not only need to learn the control policy but also requires consideration regarding interactions with all other agents in the environment, mutual influences among different system components, and the distribution of computational resources. This augments the complexity of algorithmic design and poses higher requirements on computational resources. Simultaneously, simulators are crucial to obtain realistic data, which is the fundamentals of RL. In this paper, we first propose a series of metrics of simulators and summarize the features of existing benchmarks. Second, to ease comprehension, we recall the foundational knowledge and then synthesize the recently advanced studies of MARL-related autonomous driving and intelligent transportation systems. Specifically, we examine their environmental modeling, state representation, perception units, and algorithm design. Conclusively, we discuss open challenges as well as prospects and opportunities. We hope this paper can help the researchers integrate MARL technologies and trigger more insightful ideas toward the intelligent and autonomous driving.

Device-to-device (D2D) spectrum sharing in wireless communications is a challenging non-convex combinatorial optimization problem, involving entangled link scheduling and power control in a large-scale network. The state-of-the-art methods, either from a model-based or a data-driven perspective, exhibit certain limitations such as the critical need for channel state information (CSI) and/or a large number of (solved) instances (e.g., network layouts) as training samples. To advance this line of research, we propose a novel hybrid model/datadriven spectrum sharing mechanism with graph reinforcement learning for link scheduling (GRLinQ), injecting information theoretical insights into machine learning models, in such a way that link scheduling and power control can be solved in an intelligent yet explainable manner. Through an extensive set of experiments, GRLinQ demonstrates superior performance to the existing model-based and data-driven link scheduling and/or power control methods, with a relaxed requirement for CSI, a substantially reduced number of unsolved instances as training samples, a possible distributed deployment, reduced online/offline computational complexity, and more remarkably excellent scalability and generalizability over different network scenarios and system configurations.

Reliable quantum computation requires systematic identification and correction of errors that occur and accumulate in quantum hardware. To diagnose and correct such errors, standard quantum error-correcting protocols utilize $\textit{global}$ error information across the system obtained by mid-circuit readout of ancillary qubits. We investigate circuit-level error-correcting protocols that are measurement-free and based on $\textit{local}$ error information. Such a local error correction (LEC) circuit consists of faulty multi-qubit gates to perform both syndrome extraction and ancilla-controlled error removal. We develop and implement a reinforcement learning framework that takes a fixed set of faulty gates as inputs and outputs an optimized LEC circuit. To evaluate this approach, we quantitatively characterize an extension of logical qubit lifetime by a noisy LEC circuit. For the 2D classical Ising model and 4D toric code, our optimized LEC circuit performs better at extending a memory lifetime compared to a conventional LEC circuit based on Toom's rule in a sub-threshold gate error regime. We further show that such circuits can be used to reduce the rate of mid-circuit readouts to preserve a 2D toric code memory. Finally, we discuss the application of the LEC protocol on dissipative preparation of quantum states with topological phases.