This paper proposes a first order gradient reinforcement learning algorithm, which can be seen as a variant for Trust Region Policy Optimization(TRPO). This method, which we call policy optimization with penalized point probability distance (POP3D), keeps almost all positive spheres of proximal policy optimization (PPO) such as easy implementation, fast learning and high score capability. As PPO, we also use a single surrogate objective without constraints, where a penalized item based on point probability distance is included to prevent update step from growing too large. Experiments verify that POP3D is state-of-the-art within 40 million frame steps on 49 Atari games based on two common metrics, which can be a competitive alternative to PPO. Moreover, comparison experiments regarding PPO based on Mujoco environment verify that POP3D is also competitive in continuous domain. In addition, we release the code on github https://github.com/cxxgtxy/POP3D.git.

Deep reinforcement learning has led to several recent breakthroughs, though the learned policies are often based on black-box neural networks. This makes them difficult to interpret and to impose desired specification constraints during learning. We present an iterative framework, MORL, for improving the learned policies using program synthesis. Concretely, we propose to use synthesis techniques to obtain a symbolic representation of the learned policy, which can then be debugged manually or automatically using program repair. After the repair step, we use behavior cloning to obtain the policy corresponding to the repaired program, which is then further improved using gradient descent. This process continues until the learned policy satisfies desired constraints. We instantiate MORL for the simple CartPole problem and show that the programmatic representation allows for high-level modifications that in turn lead to improved learning of the policies.

Suppose you could merely imagine a computation, and a digital prostheses, an extension of your biological brain, would turn it into code that instantly realizes what you had in mind. Imagine looking at an image, dataset or set of equations and wanting to analyze and explore its meaning as an artistic whim or part of a scientific investigation. I don't mean you would use an existing software suite to produce a standard visualization, but rather you would make use of an extensive repository of existing code to assemble a new program analogous to how a composer draws upon a repertoire of musical motifs, themes and styles to construct new works, and tantamount to having a talented musical amanuensis who, in addition to copying your scores, takes liberties with your prior work, making small alterations here and there and occasionally adding new works of its own invention, novel but consistent with your taste and sensibilities. Perhaps the interaction would be wordless and you would express your objective by simply focusing your attention and guiding your imagination, the prostheses operating directly on patterns of activation arising in your primary sensory, proprioceptive and associative cortex that have become part of an extensive vocabulary that you now share with your personal digital amanuensis. Or perhaps it would involve a conversation conducted in subvocal, unarticulated speech in which you specify what it is you want to compute and your assistant asks questions to clarify your intention and the two of you share examples of input and output to ground your internal conversation in concrete terms. More than thirty years ago, Charles Rich and Richard Waters published an MIT AI Lab technical report [68] entitled The Programmer's Apprentice: A Research Overview. Whether they intended it or not, it would have been easy in those days for someone to misremember the title and inadvertently refer to it as "The Sorcerer's Apprentice" since computer programmers at the time were often characterized as wizards and most children were familiar with the Walt Disney movie Fantasia, featuring music written by Paul Dukas inspired by Goethe's poem of the same name

We introduce TextWorld, a sandbox learning environment for the training and evaluation of RL agents on text-based games. TextWorld is a Python library that handles interactive play-through of text games, as well as backend functions like state tracking and reward assignment. It comes with a curated list of games whose features and challenges we have analyzed. More significantly, it enables users to handcraft or automatically generate new games. Its generative mechanisms give precise control over the difficulty, scope, and language of constructed games, and can be used to relax challenges inherent to commercial text games like partial observability and sparse rewards. By generating sets of varied but similar games, TextWorld can also be used to study generalization and transfer learning. We cast text-based games in the Reinforcement Learning formalism, use our framework to develop a set of benchmark games, and evaluate several baseline agents on this set and the curated list.

The AI for Prosthetics challenge is one of NIPS 2018 Competition tracks. In this challenge, the participants seek to build an agent that can make a 3D model of human with prosthetics run. This challenge is a continuation of the Learning to Run challenge (shown below) that was part of NIPS 2017 Competition Track. To start the challenge, you first need to install few packages with Anaconda. Here is a detailed description of the installation process.

Due to burdensome data requirements, learning from demonstration often falls short of its promise to allow users to quickly and naturally program robots. Demonstrations are inherently ambiguous and incomplete, making a correct generalization to unseen situations difficult without a large number of demonstrations in varying conditions. By contrast, humans are often able to learn complex tasks from a single demonstration (typically observations without action labels) by leveraging context learned over a lifetime. Inspired by this capability, we aim to enable robots to perform one-shot learning of multi-step tasks from observation by leveraging auxiliary video data as context. Our primary contribution is a novel action localization algorithm that identifies clips of activities in auxiliary videos that match the activities in a user-segmented demonstration, providing additional examples of each. While this auxiliary video data could be used in multiple ways for learning, we focus on an inverse reinforcement learning setting. We empirically show that across several tasks, robots can learn multi-step tasks more effectively from videos with localized actions, compared to unsegmented videos.

In this work we aim to solve a large collection of tasks using a single reinforcement learning agent with a single set of parameters. A key challenge is to handle the increased amount of data and extended training time. We have developed a new distributed agent IMPALA (Importance Weighted Actor-Learner Architecture) that not only uses resources more efficiently in singlemachine training but also scales to thousands of machines without sacrificing data efficiency or resource utilisation. We achieve stable learning at high throughput by combining decoupled acting and learning with a novel off-policy correction method called V-trace. We demonstrate the effectiveness of IMPALA for multi-task reinforcement learning on DMLab-30 (a set of 30 tasks from the DeepMind Lab environment (Beattie et al., 2016)) and Atari-57 (all available Atari games in Arcade Learning Environment (Bellemare et al., 2013a)). Our results show that IMPALA is able to achieve better performance than previous agents with less data, and crucially exhibits positive transfer between tasks as a result of its multi-task approach. The source code is publicly available at github.com/deepmind/scalable

Transfer learning can greatly speed up reinforcement learning for a new task by leveraging policies of relevant tasks. Existing works of policy reuse either focus on only selecting a single best source policy for transfer without considering contexts, or cannot guarantee to learn an optimal policy for a target task. To improve transfer efficiency and guarantee optimality, we develop a novel policy reuse method, called Context-Aware Policy reuSe (CAPS), that enables multi-policy transfer. Our method learns when and which source policy is best for reuse, as well as when to terminate its reuse. CAPS provides theoretical guarantees in convergence and optimality for both source policy selection and target task learning. Empirical results on a grid-based navigation domain and the Pygame Learning Environment demonstrate that CAPS significantly outperforms other state-of-the-art policy reuse methods.

One of the key challenges in applying reinforcement learning to real-life problems is that the amount of train-and-error required to learn a good policy increases drastically as the task becomes complex. One potential solution to this problem is to combine reinforcement learning with automated symbol planning and utilize prior knowledge on the domain. However, existing methods have limitations in their applicability and expressiveness. In this paper we propose a hierarchical reinforcement learning method based on abductive symbolic planning. The planner can deal with user-defined evaluation functions and is not based on the Herbrand theorem. Therefore it can utilize prior knowledge of the rewards and can work in a domain where the state space is unknown. We demonstrate empirically that our architecture significantly improves learning efficiency with respect to the amount of training examples on the evaluation domain, in which the state space is unknown and there exist multiple goals.

This post assumes that you are familiar with the basics of Reinforcement Learning(RL) and Markov Decision Processes, if not please refer to this previous post first. Let's consider a problem where the agent can be in various states and can choose an action from a set of actions. Such type of problems are called Sequential Decision Problems. The solution to an MDP is an optimal policy which refers to the choice of action for every state that maximizes overall cumulative reward. Thus, the transition model that represents an agent's environment(when the environment is known) and the optimal policy which decides what action the agent needs to perform in each state are required elements for training the agent learn a specific behavior.