Sequential decision making is a typical problem in reinforcement learning with plenty of algorithms to solve it. However, only a few of them can work effectively with a very small number of observations. In this report, we introduce the progress to learn the policy for Malaria Control as a Reinforcement Learning problem in the KDD Cup Challenge 2019 and propose diverse solutions to deal with the limited observations problem. We apply the Genetic Algorithm, Bayesian Optimization, Q-learning with sequence breaking to find the optimal policy for five years in a row with only 20 episodes/100 evaluations. We evaluate those algorithms and compare their performance with Random Search as a baseline. Among these algorithms, Q-Learning with sequence breaking has been submitted to the challenge and got ranked 7th in KDD Cup.
Aiming to produce sufficient and diverse training samples, data augmentation has been demonstrated for its effectiveness in training deep models. Regarding that the criterion of the best augmentation is challenging to define, we in this paper present a novel learning-based augmentation method termed as DeepAugNet, which formulates the final augmented data as a collection of several sequentially augmented subsets. Specifically, the current augmented subset is required to maximize the performance improvement compared with the last augmented subset by learning the deterministic augmentation policy using deep reinforcement learning. By introducing an unified optimization goal, DeepAugNet intends to combine the data augmentation and the deep model training in an end-to-end training manner which is realized by simultaneously training a hybrid architecture of dueling deep Q-learning algorithm and a surrogate deep model. We extensively evaluated our proposed DeepAugNet on various benchmark datasets including Fashion MNIST, CUB, CIFAR-100 and WebCaricature. Compared with the current state-of-the-arts, our method can achieve a significant improvement in small-scale datasets, and a comparable performance in large-scale datasets. Code will be available soon.
We present an efficient algorithm for model-free episodic reinforcement learning on large (potentially continuous) state-action spaces. Our algorithm is based on a novel $Q$-learning policy with adaptive data-driven discretization. The central idea is to maintain a finer partition of the state-action space in regions which are frequently visited in historical trajectories, and have higher payoff estimates. We demonstrate how our adaptive partitions take advantage of the shape of the optimal $Q$-function and the joint space, without sacrificing the worst-case performance. In particular, we recover the regret guarantees of prior algorithms for continuous state-action spaces, which additionally require either an optimal discretization as input, and/or access to a simulation oracle. Moreover, experiments demonstrate how our algorithm automatically adapts to the underlying structure of the problem, resulting in much better performance compared both to heuristics and $Q$-learning with uniform discretization.
We present a modern scalable reinforcement learning agent called SEED (Scalable, Efficient Deep-RL). By effectively utilizing modern accelerators, we show that it is not only possible to train on millions of frames per second but also to lower the cost of experiments compared to current methods. We achieve this with a simple architecture that features centralized inference and an optimized communication layer. SEED adopts two state of the art distributed algorithms, IMPALA/V-trace (policy gradients) and R2D2 (Q-learning), and is evaluated on Atari-57, DeepMind Lab and Google Research Football. We improve the state of the art on Football and are able to reach state of the art on Atari-57 three times faster in wall-time. For the scenarios we consider, a 40% to 80% cost reduction for running experiments is achieved. The implementation along with experiments is open-sourced so results can be reproduced and novel ideas tried out.
This paper introduces Meta-Q-Learning (MQL), a new off-policy algorithm for meta-Reinforcement Learning (meta-RL). MQL builds upon three simple ideas. First, we show that Q-learning is competitive with state-of-the-art meta-RL algorithms if given access to a context variable that is a representation of the past trajectory. Second, a multi-task objective to maximize the average reward across the training tasks is an effective method to meta-train RL policies. Third, past data from the meta-training replay buffer can be recycled to adapt the policy on a new task using off-policy updates. MQL draws upon ideas in propensity estimation to do so and thereby amplifies the amount of available data for adaptation. Experiments on standard continuous-control benchmarks suggest that MQL compares favorably with the state of the art in meta-RL.
This paper introduces the deep coordination graph (DCG) for collaborative multi-agent reinforcement learning. DCG strikes a flexible trade-off between representational capacity and generalization by factoring the joint value function of all agents according to a coordination graph into payoffs between pairs of agents. The value can be maximized by local message passing along the graph, which allows training of the value function end-to-end with Q-learning. Payoff functions are approximated with deep neural networks that employ parameter sharing and low-rank approximations to significantly improve sample efficiency. We show that DCG can solve predator-prey tasks that highlight the relative overgeneralization pathology, as well as challenging StarCraft II micromanagement tasks.
We present Graph-$Q$-SAT, a branching heuristic for a Boolean SAT solver trained with value-based reinforcement learning (RL) using Graph Neural Networks for function approximation. Solvers using Graph-$Q$-SAT are complete SAT solvers that either provide a satisfying assignment or proof of unsatisfiability, which is required for many SAT applications. The branching heuristics commonly used in SAT solvers make poor decisions during their warm-up period, whereas Graph-$Q$-SAT is trained to examine the structure of the particular problem instance to make better decisions early in the search. Training Graph-$Q$-SAT is data efficient and does not require elaborate dataset preparation or feature engineering. We train Graph-$Q$-SAT using RL interfacing with MiniSat solver and show that Graph-$Q$-SAT can reduce the number of iterations required to solve SAT problems by 2-3X. Furthermore, it generalizes to unsatisfiable SAT instances, as well as to problems with 5X more variables than it was trained on. We show that for larger problems, reductions in the number of iterations lead to wall clock time reductions, the ultimate goal when designing heuristics. We also show positive zero-shot transfer behavior when testing Graph-$Q$-SAT on a task family different from that used for training. While more work is needed to apply Graph-$Q$-SAT to reduce wall clock time in modern SAT solving settings, it is a compelling proof-of-concept showing that RL equipped with Graph Neural Networks can learn a generalizable branching heuristic for SAT search.
Active inference is a first principle account of how autonomous agents operate in dynamic, non-stationary environments. This problem is also considered in reinforcement learning (RL), but limited work exists on comparing the two approaches on the same discrete-state environments. In this paper, we provide: 1) an accessible overview of the discrete-state formulation of active inference, highlighting natural behaviors in active inference that are generally engineered in RL; 2) an explicit discrete-state comparison between active inference and RL on an OpenAI gym baseline. We begin by providing a condensed overview of the active inference literature, in particular viewing the various natural behaviors of active inference agents through the lens of RL. We show that by operating in a pure belief-based setting, active inference agents can carry out epistemic exploration, and account for uncertainty about their environment in a Bayes-optimal fashion. Furthermore, we show that the reliance on an explicit reward signal in RL is removed in active inference, where reward can simply be treated as another observation; even in the total absence of rewards, agent behaviors are learned through preference learning. We make these properties explicit by showing two scenarios in which active inference agents can infer behaviors in reward-free environments compared to both Q-learning and Bayesian model-based RL agents; by placing zero prior preferences over rewards and by learning the prior preferences over the observations corresponding to reward. We conclude by noting that this formalism can be applied to more complex settings if appropriate generative models can be formulated. In short, we aim to demystify the behavior of active inference agents by presenting an accessible discrete state-space and time formulation, and demonstrate these behaviors in a OpenAI gym environment, alongside RL agents.
This paper presents ModelicaGym toolbox that was developed to employ Reinforcement Learning (RL) for solving optimization and control tasks in Modelica models. The developed tool allows connecting models using Functional Mock-up Interface (FMI) toOpenAI Gym toolkit in order to exploit Modelica equation-based modelling and co-simulation together with RL algorithms as a functionality of the tools correspondingly. Thus, ModelicaGym facilitates fast and convenient development of RL algorithms and their comparison when solving optimal control problem for Modelicadynamic models. Inheritance structure ofModelicaGymtoolbox's classes and the implemented methods are discussed in details. The toolbox functionality validation is performed on Cart-Pole balancing problem. This includes physical system model description and its integration using the toolbox, experiments on selection and influence of the model parameters (i.e. force magnitude, Cart-pole mass ratio, reward ratio, and simulation time step) on the learning process of Q-learning algorithm supported with the discussion of the simulation results.
In this article we explore an alternative approach to address deep exploration and we introduce the ISL algorithm, which is efficient at performing deep exploration. Similarly to maximum entropy RL, we derive the algorithm by augmenting the traditional RL objective with a novel regularization term. A distinctive feature of our approach is that, as opposed to other works that tackle the problem of deep exploration, in our derivation both the learning equations and the exploration-exploitation strategy are derived in tandem as the solution to a well-posed optimization problem whose minimization leads to the optimal value function. Empirically we show that our method exhibits state of the art performance on a range of challenging deep-exploration benchmarks.
This paper makes one step forward towards characterizing a new family of \textit{model-free} Deep Reinforcement Learning (DRL) algorithms. The aim of these algorithms is to jointly learn an approximation of the state-value function ($V$), alongside an approximation of the state-action value function ($Q$). Our analysis starts with a thorough study of the Deep Quality-Value Learning (DQV) algorithm, a DRL algorithm which has been shown to outperform popular techniques such as Deep-Q-Learning (DQN) and Double-Deep-Q-Learning (DDQN) \cite{sabatelli2018deep}. Intending to investigate why DQV's learning dynamics allow this algorithm to perform so well, we formulate a set of research questions which help us characterize a new family of DRL algorithms. Among our results, we present some specific cases in which DQV's performance can get harmed and introduce a novel \textit{off-policy} DRL algorithm, called DQV-Max, which can outperform DQV. We then study the behavior of the $V$ and $Q$ functions that are learned by DQV and DQV-Max and show that both algorithms might perform so well on several DRL test-beds because they are less prone to suffer from the overestimation bias of the $Q$ function.
Since the recent advent of deep reinforcement learning for game play and simulated robotic control, a multitude of new algorithms have flourished. Most are model-free algorithms which can be categorized into three families: deep Q-learning, policy gradients, and Q-value policy gradients. These have developed along separate lines of research, such that few, if any, code bases incorporate all three kinds. Yet these algorithms share a great depth of common deep reinforcement learning machinery. We are pleased to share rlpyt, which implements all three algorithm families on top of a shared, optimized infrastructure, in a single repository. It contains modular implementations of many common deep RL algorithms in Python using PyTorch, a leading deep learning library. rlpyt is designed as a high-throughput code base for small- to medium-scale research in deep RL. This white paper summarizes its features, algorithms implemented, and relation to prior work, and concludes with detailed implementation and usage notes. rlpyt is available at https://github.com/astooke/rlpyt.
Currently, many applications in Machine Learning are based on define new models to extract more information about data, In this case Deep Reinforcement Learning with the most common application in video games like Atari, Mario, and others causes an impact in how to computers can learning by himself with only information called rewards obtained from any action. There is a lot of algorithms modeled and implemented based on Deep Recurrent Q-Learning proposed by DeepMind used in AlphaZero and Go. In this document, We proposed Deep Recurrent Double Q-Learning that is an implementation of Deep Reinforcement Learning using Double Q-Learning algorithms and Recurrent Networks like LSTM and DRQN.
We consider the task of learning to play families of text-based computer adventure games, i.e., fully textual environments with a common theme (e.g. cooking) and goal (e.g. prepare a meal from a recipe) but with different specifics; new instances of such games are relatively straightforward for humans to master after a brief exposure to the genre but have been curiously difficult for computer agents to learn. We find that the deep Q-learning strategies that have been successfully leveraged for superhuman performance in single-instance action video games can be applied to learn families of text video games when adopting simple strategies that correlate with human-like learning behavior. Specifically, we build agents that learn to tackle simple scenarios before more complex ones using curriculum learning, that familiarize themselves in an unfamiliar environment by navigating before acting, and that explore uncertain environments more thoroughly using contextual multi-armed bandit decision policies. We demonstrate improved task completion rates over reasonable baselines when evaluating on never-before-seen games of that theme.
Modern control theories such as systems engineering approaches try to solve nonlinear system problems by revelation of causal relationship or co-relationship among the components; most of those approaches focus on control of sophisticatedly modeled white-boxed systems. We suggest an application of actor-critic reinforcement learning approach to control a nonlinear, complex and black-boxed system. We demonstrated this approach on artificial green-house environment simulator all of whose control inputs have several side effects so human cannot figure out how to control this system easily. Our approach succeeded to maintain the circumstance at least 20 times longer than PID and Deep Q Learning.
Deep reinforcement learning has over the past few years shown great potential in learning near-optimal control in complex simulated environments with little visible information. Rainbow (Q-Learning) and PPO (Policy Optimisation) have shown outstanding performance in a variety of tasks, including Atari 2600, MuJoCo, and Roboschool test suite. While these algorithms are fundamentally different, both suffer from high variance, low sample efficiency, and hyperparameter sensitivity that in practice, make these algorithms a no-go for critical operations in the industry. On the other hand, model-based reinforcement learning focuses on learning the transition dynamics between states in an environment. If these environment dynamics are adequately learned, a model-based approach is perhaps the most sample efficient method for learning agents to act in an environment optimally. The traits of model-based reinforcement are ideal for real-world environments where sampling is slow and for mission-critical operations. In the warehouse industry, there is an increasing motivation to minimise time and to maximise production. Currently, autonomous agents act suboptimally using handcrafted policies for significant portions of the state-space. In this paper, we present The Dreaming Variational Autoencoder v2 (DVAE-2), a model-based reinforcement learning algorithm that increases sample efficiency, hence enable algorithms with low sample efficiency function better in real-world environments. We introduce Deep Warehouse, a simulated environment for industry-near testing of autonomous agents in grid-based warehouses. Finally, we illustrate that DVAE-2 improves the sample efficiency for the Deep Warehouse compared to model-free methods.
Off-policy reinforcement learning (RL) using a fixed offline dataset of logged interactions is an important consideration in real world applications. This paper studies offline RL using the DQN replay dataset comprising the entire replay experience of a DQN agent on 60 Atari 2600 games. We demonstrate that recent off-policy deep RL algorithms, even when trained solely on this fixed dataset, outperform the fully trained DQN agent. To enhance generalization in the offline setting, we present Random Ensemble Mixture (REM), a robust Q-learning algorithm that enforces optimal Bellman consistency on random convex combinations of multiple Q-value estimates. Offline REM trained on the DQN replay dataset surpasses strong RL baselines. Ablation studies highlight the role of offline dataset size and diversity as well as the algorithm choice in our positive results. Overall, the results here present an optimistic view that robust RL algorithms trained on sufficiently large and diverse offline datasets can lead to high quality policies. The DQN replay dataset can serve as an offline RL benchmark and is open-sourced.
Most deep reinforcement learning (RL) systems are not able to learn effectively from off-policy data, especially if they cannot explore online in the environment. These are critical shortcomings for applying RL to real-world problems where collecting data is expensive, and models must be tested offline before being deployed to interact with the environment -- e.g. systems that learn from human interaction. Thus, we develop a novel class of off-policy batch RL algorithms, which are able to effectively learn offline, without exploring, from a fixed batch of human interaction data. We leverage models pre-trained on data as a strong prior, and use KL-control to penalize divergence from this prior during RL training. We also use dropout-based uncertainty estimates to lower bound the target Q-values as a more efficient alternative to Double Q-Learning. The algorithms are tested on the problem of open-domain dialog generation -- a challenging reinforcement learning problem with a 20,000-dimensional action space. Using our Way Off-Policy algorithm, we can extract multiple different reward functions post-hoc from collected human interaction data, and learn effectively from all of these. We test the real-world generalization of these systems by deploying them live to converse with humans in an open-domain setting, and demonstrate that our algorithm achieves significant improvements over prior methods in off-policy batch RL.
Drawing an inspiration from behavioral studies of human decision making, we propose here a general parametric framework for a reinforcement learning problem, which extends the standard Q-learning approach to incorporate a two-stream framework of reward processing with biases biologically associated with several neurological and psychiatric conditions, including Parkinson's and Alzheimer's diseases, attention-deficit/hyperactivity disorder (ADHD), addiction, and chronic pain. For AI community, the development of agents that react differently to different types of rewards can enable us to understand a wide spectrum of multi-agent interactions in complex real-world socioeconomic systems. Moreover, from the behavioral modeling perspective, our parametric framework can be viewed as a first step towards a unifying computational model capturing reward processing abnormalities across multiple mental conditions and user preferences in long-term recommendation systems.
A rise in Advanced Persistent Threats (APTs) has introduced a need for robustness against long-running, stealthy attacks which circumvent existing cryptographic security guarantees. FlipIt is a security game that models attacker-defender interactions in advanced scenarios such as APTs. Previous work analyzed extensively non-adaptive strategies in FlipIt, but adaptive strategies rise naturally in practical interactions as players receive feedback during the game. We model the FlipIt game as a Markov Decision Process and introduce QFlip, an adaptive strategy for FlipIt based on temporal difference reinforcement learning. We prove theoretical results on the convergence of our new strategy against an opponent playing with a Periodic strategy. We confirm our analysis experimentally by extensive evaluation of QFlip against specific opponents. QFlip converges to the optimal adaptive strategy for Periodic and Exponential opponents using associated state spaces. Finally, we introduce a generalized QFlip strategy with composite state space that outperforms a Greedy strategy for several distributions including Periodic and Uniform, without prior knowledge of the opponent's strategy. We also release an OpenAI Gym environment for FlipIt to facilitate future research.