We address the problem of sequentially selecting and observing processes from a given set to find the anomalies among them. The decision-maker observes a subset of the processes at any given time instant and obtains a noisy binary indicator of whether or not the corresponding process is anomalous. In this setting, we develop an anomaly detection algorithm that chooses the processes to be observed at a given time instant, decides when to stop taking observations, and declares the decision on anomalous processes. The objective of the detection algorithm is to identify the anomalies with an accuracy exceeding the desired value while minimizing the delay in decision making. We devise a centralized algorithm where the processes are jointly selected by a common agent as well as a decentralized algorithm where the decision of whether to select a process is made independently for each process. Our algorithms rely on a Markov decision process defined using the marginal probability of each process being normal or anomalous, conditioned on the observations. We implement the detection algorithms using the deep actor-critic reinforcement learning framework. Unlike prior work on this topic that has exponential complexity in the number of processes, our algorithms have computational and memory requirements that are both polynomial in the number of processes. We demonstrate the efficacy of these algorithms using numerical experiments by comparing them with state-of-the-art methods.

A main research goal in various studies is to use an observational data set and provide a new set of counterfactual guidelines that can yield causal improvements. Dynamic Treatment Regimes (DTRs) are widely studied to formalize this process. However, available methods in finding optimal DTRs often rely on assumptions that are violated in real-world applications (e.g., medical decision-making or public policy), especially when (a) the existence of unobserved confounders cannot be ignored, and (b) the unobserved confounders are time-varying (e.g., affected by previous actions). When such assumptions are violated, one often faces ambiguity regarding the underlying causal model that is needed to be assumed to obtain an optimal DTR. This ambiguity is inevitable, since the dynamics of unobserved confounders and their causal impact on the observed part of the data cannot be understood from the observed data. Motivated by a case study of finding superior treatment regimes for patients who underwent transplantation in our partner hospital and faced a medical condition known as New Onset Diabetes After Transplantation (NODAT), we extend DTRs to a new class termed Ambiguous Dynamic Treatment Regimes (ADTRs), in which the casual impact of treatment regimes is evaluated based on a "cloud" of potential causal models. We then connect ADTRs to Ambiguous Partially Observable Mark Decision Processes (APOMDPs) proposed by Saghafian (2018), and develop two Reinforcement Learning methods termed Direct Augmented V-Learning (DAV-Learning) and Safe Augmented V-Learning (SAV-Learning), which enable using the observed data to efficiently learn an optimal treatment regime. We establish theoretical results for these learning methods, including (weak) consistency and asymptotic normality. We further evaluate the performance of these learning methods both in our case study and in simulation experiments.

Market makers play a key role in financial markets by providing liquidity. They usually fill order books with buy and sell limit orders in order to provide traders alternative price levels to operate. This paper focuses precisely on the study of these markets makers strategies from an agent-based perspective. In particular, we propose the application of Reinforcement Learning (RL) for the creation of intelligent market markers in simulated stock markets. This research analyzes how RL market maker agents behaves in non-competitive (only one RL market maker learning at the same time) and competitive scenarios (multiple RL market markers learning at the same time), and how they adapt their strategies in a Sim2Real scope with interesting results. Furthermore, it covers the application of policy transfer between different experiments, describing the impact of competing environments on RL agents performance. RL and deep RL techniques are proven as profitable market maker approaches, leading to a better understanding of their behavior in stock markets.

We develop a new continual meta-learning method to address challenges in sequential multi-task learning. In this setting, the agent's goal is to achieve high reward over any sequence of tasks quickly. Prior meta-reinforcement learning algorithms have demonstrated promising results in accelerating the acquisition of new tasks. However, they require access to all tasks during training. Beyond simply transferring past experience to new tasks, our goal is to devise continual reinforcement learning algorithms that learn to learn, using their experience on previous tasks to learn new tasks more quickly. We introduce a new method, continual meta-policy search (CoMPS), that removes this limitation by meta-training in an incremental fashion, over each task in a sequence, without revisiting prior tasks. CoMPS continuously repeats two subroutines: learning a new task using RL and using the experience from RL to perform completely offline meta-learning to prepare for subsequent task learning. We find that CoMPS outperforms prior continual learning and off-policy meta-reinforcement methods on several sequences of challenging continuous control tasks.

The large variety of digital payment choices available to consumers today has been a key driver of e-commerce transactions in the past decade. Unfortunately, this has also given rise to cybercriminals and fraudsters who are constantly looking for vulnerabilities in these systems by deploying increasingly sophisticated fraud attacks. A typical fraud detection system employs standard supervised learning methods where the focus is on maximizing the fraud recall rate. However, we argue that such a formulation can lead to sub-optimal solutions. The design requirements for these fraud models requires that they are robust to the high-class imbalance in the data, adaptive to changes in fraud patterns, maintain a balance between the fraud rate and the decline rate to maximize revenue, and be amenable to asynchronous feedback since usually there is a significant lag between the transaction and the fraud realization. To achieve this, we formulate fraud detection as a sequential decision-making problem by including the utility maximization within the model in the form of the reward function. The historical decline rate and fraud rate define the state of the system with a binary action space composed of approving or declining the transaction. In this study, we primarily focus on utility maximization and explore different reward functions to this end. The performance of the proposed Reinforcement Learning system has been evaluated for two publicly available fraud datasets using Deep Q-learning and compared with different classifiers. We aim to address the rest of the issues in future work.

Experience replay \citep{lin1993reinforcement, mnih2015human} is a widely used technique to achieve efficient use of data and improved performance in RL algorithms. In experience replay, past transitions are stored in a memory buffer and re-used during learning. Various suggestions for sampling schemes from the replay buffer have been suggested in previous works, attempting to optimally choose those experiences which will most contribute to the convergence to an optimal policy. Here, we give some conditions on the replay sampling scheme that will ensure convergence, focusing on the well-known Q-learning algorithm in the tabular setting. After establishing sufficient conditions for convergence, we turn to suggest a slightly different usage for experience replay - replaying memories in a biased manner as a means to change the properties of the resulting policy. We initiate a rigorous study of experience replay as a tool to control and modify the properties of the resulting policy. In particular, we show that using an appropriate biased sampling scheme can allow us to achieve a \emph{safe} policy. We believe that using experience replay as a biasing mechanism that allows controlling the resulting policy in desirable ways is an idea with promising potential for many applications.

Q-learning is a well-known and commonly used algorithm for reinforcement learning. In recent years, a technique referred to as experience replay [9, 11] has been suggested as a mechanism to improve Q-learning by allowing the learner to access previous experiences, and use them offline as if they were examples currently sampled from the world. It has been suggested that using past experiences in such a way might allow Q-learning to better converge to the optimal Q values, by breaking the time and space correlation structure of experiences as they are sampled from the real world, allowing for policy updates not dependent on the current time and location in the markov decision process. Moreover, using experience replay improves the efficiency of data usage, since every experience is used for learning more than once. This can be useful in situations where data acquirement is costly or difficult.

Using a model of the environment and a value function, an agent can construct many estimates of a state's value, by unrolling the model for different lengths and bootstrapping with its value function. Our key insight is that one can treat this set of value estimates as a type of ensemble, which we call an \emph{implicit value ensemble} (IVE). Consequently, the discrepancy between these estimates can be used as a proxy for the agent's epistemic uncertainty; we term this signal \emph{model-value inconsistency} or \emph{self-inconsistency} for short. Unlike prior work which estimates uncertainty by training an ensemble of many models and/or value functions, this approach requires only the single model and value function which are already being learned in most model-based reinforcement learning algorithms. We provide empirical evidence in both tabular and function approximation settings from pixels that self-inconsistency is useful (i) as a signal for exploration, (ii) for acting safely under distribution shifts, and (iii) for robustifying value-based planning with a model.

Explanations play a considerable role in human learning, especially in areas that remain major challenges for AI -- forming abstractions, and learning about the relational and causal structure of the world. Here, we explore whether reinforcement learning agents might likewise benefit from explanations. We outline a family of relational tasks that involve selecting an object that is the odd one out in a set (i.e., unique along one of many possible feature dimensions). Odd-one-out tasks require agents to reason over multi-dimensional relationships among a set of objects. We show that agents do not learn these tasks well from reward alone, but achieve >90% performance when they are also trained to generate language explaining object properties or why a choice is correct or incorrect. In further experiments, we show how predicting explanations enables agents to generalize appropriately from ambiguous, causally-confounded training, and even to meta-learn to perform experimental interventions to identify causal structure. We show that explanations help overcome the tendency of agents to fixate on simple features, and explore which aspects of explanations make them most beneficial. Our results suggest that learning from explanations is a powerful principle that could offer a promising path towards training more robust and general machine learning systems.

Learning rational behaviors in open-world games like Minecraft remains to be challenging for Reinforcement Learning (RL) research due to the compound challenge of partial observability, high-dimensional visual perception and delayed reward. To address this, we propose JueWu-MC, a sample-efficient hierarchical RL approach equipped with representation learning and imitation learning to deal with perception and exploration. Specifically, our approach includes two levels of hierarchy, where the high-level controller learns a policy to control over options and the low-level workers learn to solve each sub-task. To boost the learning of sub-tasks, we propose a combination of techniques including 1) action-aware representation learning which captures underlying relations between action and representation, 2) discriminator-based self-imitation learning for efficient exploration, and 3) ensemble behavior cloning with consistency filtering for policy robustness. Extensive experiments show that JueWu-MC significantly improves sample efficiency and outperforms a set of baselines by a large margin. Notably, we won the championship of the NeurIPS MineRL 2021 research competition and achieved the highest performance score ever.