Reinforcement learning (RL) agents can leverage batches of previously collected data to extract a reasonable control policy. An emerging issue in this offline RL setting, however, is that the bootstrapping update underlying many of our methods suffers from insufficient action-coverage: standard max operator may select a maximal action that has not been seen in the dataset. Bootstrapping from these inaccurate values can lead to overestimation and even divergence. There are a growing number of methods that attempt to approximate an in-sample max, that only uses actions well-covered by the dataset. We highlight a simple fact: it is more straightforward to approximate an in-sample softmax using only actions in the dataset. We show that policy iteration based on the in-sample softmax converges, and that for decreasing temperatures it approaches the in-sample max. We derive an In-Sample Actor-Critic (AC), using this in-sample softmax, and show that it is consistently better or comparable to existing offline RL methods, and is also wellsuited to fine-tuning. We release the code at github.com/hwang-ua/inac A common goal in reinforcement learning (RL) is to learn a control policy from data. In the offline setting, the agent has access to a batch of previously collected data. This data could have been gathered under a near-optimal behavior policy, from a mediocre policy, or a mixture of different policies (perhaps produced by several human operators). A key challenge is to be robust to this data gathering distribution, since we often do not have control over data collection in some application settings.

A key challenge in sequential decision making is optimizing systems safely under partial information. While much of the literature has focused on the cases of either partially known states or partially known dynamics, it is further exacerbated in cases where both states and dynamics are partially known. Computing heparin doses for patients fits this paradigm since the concentration of heparin in the patient cannot be measured directly and the rates at which patients metabolize heparin vary greatly between individuals. While many proposed solutions are model free, they require complex models and have difficulty ensuring safety. However, if some of the structure of the dynamics is known, a model based approach can be leveraged to provide safe policies. In this paper we propose such a framework to address the challenge of optimizing personalized heparin doses. We use a predictive model parameterized individually by patient to predict future therapeutic effects. We then leverage this model using a scenario generation based approach that is capable of ensuring patient safety. We validate our models with numerical experiments by comparing the predictive capabilities of our model against existing machine learning techniques and demonstrating how our dosing algorithm can treat patients in a simulated ICU environment.

Resource allocation is the assignment of resources to activities that must be executed in a business process at a particular moment at run-time. While resource allocation is well-studied in other fields, such as manufacturing, there exist only a few methods in business process management. Existing methods are not suited for application in large business processes or focus on optimizing resource allocation for a single case rather than for all cases combined. To fill this gap, this paper proposes two learning-based methods for resource allocation in business processes: a deep reinforcement learning-based approach and a score-based value function approximation approach. The two methods are compared against existing heuristics in a set of scenarios that represent typical business process structures and on a complete network that represents a realistic business process. The results show that our learning-based methods outperform or are competitive with common heuristics in most scenarios and outperform heuristics in the complete network.

This paper introduces the QDQN-DPER framework to enhance the efficiency of quantum reinforcement learning (QRL) in solving sequential decision tasks. The framework incorporates prioritized experience replay and asynchronous training into the training algorithm to reduce the high sampling complexities. Numerical simulations demonstrate that QDQN-DPER outperforms the baseline distributed quantum Q learning with the same model architecture. The proposed framework holds potential for more complex tasks while maintaining training efficiency.

This paper proposes an exploration technique for multi-agent reinforcement learning (MARL) with graph-based communication among agents. We assume the individual rewards received by the agents are independent of the actions by the other agents, while their policies are coupled. In the proposed framework, neighbouring agents collaborate to estimate the uncertainty about the state-action space in order to execute more efficient explorative behaviour. Different from existing works, the proposed algorithm does not require counting mechanisms and can be applied to continuous-state environments without requiring complex conversion techniques. Moreover, the proposed scheme allows agents to communicate in a fully decentralized manner with minimal information exchange. And for continuous-state scenarios, each agent needs to exchange only a single parameter vector. The performance of the algorithm is verified with theoretical results for discrete-state scenarios and with experiments for continuous ones.

Online reinforcement learning and other adaptive sampling algorithms are increasingly used in digital intervention experiments to optimize treatment delivery for users over time. In this work, we focus on longitudinal user data collected by a large class of adaptive sampling algorithms that are designed to optimize treatment decisions online using accruing data from multiple users. Combining or "pooling" data across users allows adaptive sampling algorithms to potentially learn faster. However, by pooling, these algorithms induce dependence between the sampled user data trajectories; we show that this can cause standard variance estimators for i.i.d. data to underestimate the true variance of common estimators on this data type. We develop novel methods to perform a variety of statistical analyses on such adaptively sampled data via Z-estimation. Specifically, we introduce the \textit{adaptive} sandwich variance estimator, a corrected sandwich estimator that leads to consistent variance estimates under adaptive sampling. Additionally, to prove our results we develop novel theoretical tools for empirical processes on non-i.i.d., adaptively sampled longitudinal data which may be of independent interest. This work is motivated by our efforts in designing experiments in which online reinforcement learning algorithms optimize treatment decisions, yet statistical inference is essential for conducting analyses after experiments conclude.

Coronary artery disease is the most common cardiovascular diseases and kills millions every year [1]. Percutaneous coronary intervention (PCI) is a widely used treatment for coronary artery disease. In PCI, physicians use X-ray fluoroscopy for guidance and deliver guidewires, catheters, and other instruments to the target vessel for treatments such as stenting and drugs. Due to X-ray fluoroscopy guidance, physicians are exposed to radiation and wear heavy lead-lined garments for radiation protection, which leads to radiation-associated hazards [2] and orthopedic strain injuries [3]. Vascular robotic systems with the master-slave control mode [4, 5] have been developed to reduce the risks mentioned above. Robot-assisted intervention has shown numerous benefits in clinical trials, including X-ray exposure reduction, control precision improvement, and procedural duration decrease. In robot-assisted intervention, instruments are shaped as flexible wires. Physicians manipulate the proximal tip of instruments outside the patient body to deliver the distal tip to the target in vessels. The relationship between manipulations and distal motion is non-linear, making instrument deliveries challenging.

Domain-adaptive trajectory imitation is a skill that some predators learn for survival, by mapping dynamic information from one domain (their speed and steering direction) to a different domain (current position of the moving prey). An intelligent agent with this skill could be exploited for a diversity of tasks, including the recognition of abnormal motion in traffic once it has learned to imitate representative trajectories. Towards this direction, we propose DATI, a deep reinforcement learning agent designed for domain-adaptive trajectory imitation using a cycle-consistent generative adversarial method. Our experiments on a variety of synthetic families of reference trajectories show that DATI outperforms baseline methods for imitation learning and optimal control in this setting, keeping the same per-task hyperparameters. Its generalization to a real-world scenario is shown through the discovery of abnormal motion patterns in maritime traffic, opening the door for the use of deep reinforcement learning methods for spatially-unconstrained trajectory data mining.

This work contributes to developing an agent based on deep reinforcement learning capable of acting in a beyond visual range (BVR) air combat simulation environment. The paper presents an overview of building an agent representing a high-performance fighter aircraft that can learn and improve its role in BVR combat over time based on rewards calculated using operational metrics. Also, through self-play experiments, it expects to generate new air combat tactics never seen before. Finally, we hope to examine a real pilot's ability, using virtual simulation, to interact in the same environment with the trained agent and compare their performances. This research will contribute to the air combat training context by developing agents that can interact with real pilots to improve their performances in air defense missions.

This work investigates the formal policy synthesis of continuous-state stochastic dynamic systems given high-level specifications in linear temporal logic. To learn an optimal policy that maximizes the satisfaction probability, we take a product between a dynamic system and the translated automaton to construct a product system on which we solve an optimal planning problem. Since this product system has a hybrid product state space that results in reward sparsity, we introduce a generalized optimal backup order, in reverse to the topological order, to guide the value backups and accelerate the learning process. We provide the optimality proof for using the generalized optimal backup order in this optimal planning problem. Further, this paper presents an actor-critic reinforcement learning algorithm when topological order applies. This algorithm leverages advanced mathematical techniques and enjoys the property of hyperparameter self-tuning. We provide proof of the optimality and convergence of our proposed reinforcement learning algorithm. We use neural networks to approximate the value function and policy function for hybrid product state space. Furthermore, we observe that assigning integer numbers to automaton states can rank the value or policy function approximated by neural networks. To break the ordinal relationship, we use an individual neural network for each automaton state's value (policy) function, termed modular learning. We conduct two experiments. First, to show the efficacy of our reinforcement learning algorithm, we compare it with baselines on a classic control task, CartPole. Second, we demonstrate the empirical performance of our formal policy synthesis framework on motion planning of a Dubins car with a temporal specification.