Meta-World: A Benchmark and Evaluation for Multi-Task and Meta #ReinforcementLearning
Abstract: #Meta-reinforcement learning algorithms can enable robots to acquire new skills much more quickly, by leveraging prior experience to learn how to learn. However, much of the current research on meta-reinforcement learning focuses on task distributions that are very narrow. For example, a commonly used meta-reinforcement learning benchmark uses different running velocities for a simulated robot as different tasks. When policies are meta-trained on such narrow task distributions, they cannot possibly generalize to more quickly acquire entirely new tasks. Therefore, if the aim of these methods is to enable faster acquisition of entirely new behaviors, we must evaluate them on task distributions that are sufficiently broad to enable generalization to new behaviors. In this paper, we propose an open-source simulated benchmark for meta-reinforcement learning and multi-task learning consisting of 50 distinct robotic manipulation tasks. Our aim is to make it possible to develop algorithms that generalize to accelerate the acquisition of entirely new, held-out tasks. We evaluate 6 state-of-the-art metareinforcement learning and multi-task learning algorithms on these tasks. Surprisingly, while each task and its variations (e.g., with different object positions) can be learned with reasonable success, these algorithms struggle to learn with multiple tasks at the same time, even with as few as ten distinct training tasks. Our analysis and open-source environments pave the way for future research in multi-task learning and meta-learning that can enable meaningful generalization, thereby unlocking the full potential of these methods.
Paper
🔭 @DeepGravity
Abstract: #Meta-reinforcement learning algorithms can enable robots to acquire new skills much more quickly, by leveraging prior experience to learn how to learn. However, much of the current research on meta-reinforcement learning focuses on task distributions that are very narrow. For example, a commonly used meta-reinforcement learning benchmark uses different running velocities for a simulated robot as different tasks. When policies are meta-trained on such narrow task distributions, they cannot possibly generalize to more quickly acquire entirely new tasks. Therefore, if the aim of these methods is to enable faster acquisition of entirely new behaviors, we must evaluate them on task distributions that are sufficiently broad to enable generalization to new behaviors. In this paper, we propose an open-source simulated benchmark for meta-reinforcement learning and multi-task learning consisting of 50 distinct robotic manipulation tasks. Our aim is to make it possible to develop algorithms that generalize to accelerate the acquisition of entirely new, held-out tasks. We evaluate 6 state-of-the-art metareinforcement learning and multi-task learning algorithms on these tasks. Surprisingly, while each task and its variations (e.g., with different object positions) can be learned with reasonable success, these algorithms struggle to learn with multiple tasks at the same time, even with as few as ten distinct training tasks. Our analysis and open-source environments pave the way for future research in multi-task learning and meta-learning that can enable meaningful generalization, thereby unlocking the full potential of these methods.
Paper
🔭 @DeepGravity
Learning human objectives by evaluating hypothetical behaviours
TL;DR: We present a method for training #ReinforcementLearning agents from human feedback in the presence of unknown unsafe states.
#DeepMind
Link
🔭 @DeepGravity
TL;DR: We present a method for training #ReinforcementLearning agents from human feedback in the presence of unknown unsafe states.
#DeepMind
Link
🔭 @DeepGravity
Deepmind
Learning human objectives by evaluating hypothetical behaviours
We present a new method for training reinforcement learning agents from human feedback in the presence of unknown unsafe states.
#ReinforcementLearning for ArtiSynth
This repository holds the plugin for the #biomechanical simulation environment of ArtiSynth. The purpose of this work is to bridge in between the biomechanical and reinforcement learning domains of research.
Link
🔭 @DeepGravity
This repository holds the plugin for the #biomechanical simulation environment of ArtiSynth. The purpose of this work is to bridge in between the biomechanical and reinforcement learning domains of research.
Link
🔭 @DeepGravity
GitHub
GitHub - amir-abdi/artisynth-rl: Reinforcement Learning plugin and models for ArtiSynth
Reinforcement Learning plugin and models for ArtiSynth - GitHub - amir-abdi/artisynth-rl: Reinforcement Learning plugin and models for ArtiSynth
Our reinforcement learning architect designs have been just published on #NeurIPS2019 AI Art Gallery:
https://lnkd.in/dFZ37BN
Seems it draws like a baby now, but is growing and hopefully would be a skillful #RL artist very soon.
#reinforcementlearning #deeplearning #ai #artificialintelligence #art #deepreinforcementlearning #creativeart #neurips
🔭 @DeepGravity
https://lnkd.in/dFZ37BN
Seems it draws like a baby now, but is growing and hopefully would be a skillful #RL artist very soon.
#reinforcementlearning #deeplearning #ai #artificialintelligence #art #deepreinforcementlearning #creativeart #neurips
🔭 @DeepGravity
AI Art Gallery
Yuta Akizuki, Mathias Bernhard, Reza Kakooee, Marirena Kladeftira, Benjamin Dillenburger - AI Art Gallery
Generative Modelling with Design Constraints – Reinforcement Learning for Furniture Generation (2019) Generative design has been…
#ReinforcementLearning and #Control as #ProbabilisticInference: Tutorial and Review
The framework of reinforcement learning or #OptimalControl provides a mathematical formalization of intelligent decision making that is powerful and broadly applicable. While the general form of the reinforcement learning problem enables effective reasoning about uncertainty, the connection between reinforcement learning and inference in probabilistic models is not immediately obvious. However, such a connection has considerable value when it comes to algorithm design: formalizing a problem as probabilistic inference in principle allows us to bring to bear a wide array of approximate inference tools, extend the model in flexible and powerful ways, and reason about compositionality and partial observability. In this article, we will discuss how a generalization of the reinforcement learning or optimal control problem, which is sometimes termed maximum entropy reinforcement learning, is equivalent to exact probabilistic inference in the case of deterministic dynamics, and variational inference in the case of stochastic dynamics. We will present a detailed derivation of this framework, overview prior work that has drawn on this and related ideas to propose new reinforcement learning and control algorithms, and describe perspectives on future research.
Link
🔭 @DeepGravity
The framework of reinforcement learning or #OptimalControl provides a mathematical formalization of intelligent decision making that is powerful and broadly applicable. While the general form of the reinforcement learning problem enables effective reasoning about uncertainty, the connection between reinforcement learning and inference in probabilistic models is not immediately obvious. However, such a connection has considerable value when it comes to algorithm design: formalizing a problem as probabilistic inference in principle allows us to bring to bear a wide array of approximate inference tools, extend the model in flexible and powerful ways, and reason about compositionality and partial observability. In this article, we will discuss how a generalization of the reinforcement learning or optimal control problem, which is sometimes termed maximum entropy reinforcement learning, is equivalent to exact probabilistic inference in the case of deterministic dynamics, and variational inference in the case of stochastic dynamics. We will present a detailed derivation of this framework, overview prior work that has drawn on this and related ideas to propose new reinforcement learning and control algorithms, and describe perspectives on future research.
Link
🔭 @DeepGravity
Interestingness Elements for Explainable #ReinforcementLearning: Understanding Agents' Capabilities and Limitations
We propose an explainable reinforcement learning (XRL) framework that analyzes an agent's history of interaction with the environment to extract interestingness elements that help explain its behavior. The framework relies on data readily available from standard RL algorithms, augmented with data that can easily be collected by the agent while learning. We describe how to create visual explanations of an agent's behavior in the form of short video-clips highlighting key interaction moments, based on the proposed elements. We also report on a user study where we evaluated the ability of humans in correctly perceiving the aptitude of agents with different characteristics, including their capabilities and limitations, given explanations automatically generated by our framework. The results show that the diversity of aspects captured by the different interestingness elements is crucial to help humans correctly identify the agents' aptitude in the task, and determine when they might need adjustments to improve their performance.
Link
🔭 @DeepGravity
We propose an explainable reinforcement learning (XRL) framework that analyzes an agent's history of interaction with the environment to extract interestingness elements that help explain its behavior. The framework relies on data readily available from standard RL algorithms, augmented with data that can easily be collected by the agent while learning. We describe how to create visual explanations of an agent's behavior in the form of short video-clips highlighting key interaction moments, based on the proposed elements. We also report on a user study where we evaluated the ability of humans in correctly perceiving the aptitude of agents with different characteristics, including their capabilities and limitations, given explanations automatically generated by our framework. The results show that the diversity of aspects captured by the different interestingness elements is crucial to help humans correctly identify the agents' aptitude in the task, and determine when they might need adjustments to improve their performance.
Link
🔭 @DeepGravity
Explaining #ReinforcementLearning: Active vs Passive
We examine the required elements to solve an RL problem, compare passive and active reinforcement learning, and review common active and passive RL techniques.
Article
🔭 @DeepGravity
We examine the required elements to solve an RL problem, compare passive and active reinforcement learning, and review common active and passive RL techniques.
Article
🔭 @DeepGravity
KDnuggets
Explaining Reinforcement Learning: Active vs Passive - KDnuggets
We examine the required elements to solve an RL problem, compare passive and active reinforcement learning, and review common active and passive RL techniques.
Positive-Unlabeled #RewardLearning
Learning #Reward functions from data is a promising path towards achieving scalable #ReinforcementLearning ( #RL ) for #robotics. However, a major challenge in training agents from learned reward models is that the agent can learn to exploit errors in the reward model to achieve high reward behaviors that do not correspond to the intended task. These reward delusions can lead to unintended and even dangerous behaviors. On the other hand, adversarial imitation learning frameworks tend to suffer the opposite problem, where the discriminator learns to trivially distinguish agent and expert behavior, resulting in reward models that produce low reward signal regardless of the input state. In this paper, we connect these two classes of reward learning methods to positive-unlabeled (PU) learning, and we show that by applying a large-scale PU learning algorithm to the reward learning problem, we can address both the reward under- and over-estimation problems simultaneously. Our approach drastically improves both GAIL and supervised reward learning, without any additional assumptions.
Paper
🔭 @DeepGravity
Learning #Reward functions from data is a promising path towards achieving scalable #ReinforcementLearning ( #RL ) for #robotics. However, a major challenge in training agents from learned reward models is that the agent can learn to exploit errors in the reward model to achieve high reward behaviors that do not correspond to the intended task. These reward delusions can lead to unintended and even dangerous behaviors. On the other hand, adversarial imitation learning frameworks tend to suffer the opposite problem, where the discriminator learns to trivially distinguish agent and expert behavior, resulting in reward models that produce low reward signal regardless of the input state. In this paper, we connect these two classes of reward learning methods to positive-unlabeled (PU) learning, and we show that by applying a large-scale PU learning algorithm to the reward learning problem, we can address both the reward under- and over-estimation problems simultaneously. Our approach drastically improves both GAIL and supervised reward learning, without any additional assumptions.
Paper
🔭 @DeepGravity
Training Agents using Upside-Down #ReinforcementLearning
Traditional Reinforcement Learning (RL) algorithms either predict rewards with value functions or maximize them using policy search. We study an alternative: Upside-Down Reinforcement Learning (Upside-Down RL or #UDRL), that solves RL problems primarily using supervised learning techniques. Many of its main principles are outlined in a companion report [34]. Here we present the first concrete implementation of UDRL and demonstrate its feasibility on certain episodic learning problems. Experimental results show that its performance can be surprisingly competitive with, and even exceed that of traditional baseline algorithms developed over decades of research.
#JürgenSchmidhuber
Paper
🔭 @DeepGravity
Traditional Reinforcement Learning (RL) algorithms either predict rewards with value functions or maximize them using policy search. We study an alternative: Upside-Down Reinforcement Learning (Upside-Down RL or #UDRL), that solves RL problems primarily using supervised learning techniques. Many of its main principles are outlined in a companion report [34]. Here we present the first concrete implementation of UDRL and demonstrate its feasibility on certain episodic learning problems. Experimental results show that its performance can be surprisingly competitive with, and even exceed that of traditional baseline algorithms developed over decades of research.
#JürgenSchmidhuber
Paper
🔭 @DeepGravity
Crowdfunding Dynamics Tracking: A #ReinforcementLearning Approach
Recent years have witnessed the increasing interests in research of crowdfunding mechanism. In this area, dynamics tracking is a significant issue but is still under exploration. Existing studies either fit the fluctuations of time-series or employ regularization terms to constrain learned tendencies. However, few of them take into account the inherent decision-making process between investors and crowdfunding dynamics. To address the problem, in this paper, we propose a Trajectory-based Continuous Control for Crowdfunding (TC3) algorithm to predict the funding progress in crowdfunding. Specifically, actor-critic frameworks are employed to model the relationship between investors and campaigns, where all of the investors are viewed as an agent that could interact with the environment derived from the real dynamics of campaigns. Then, to further explore the in-depth implications of patterns (i.e., typical characters) in funding series, we propose to subdivide them into fast-growing and slow-growing ones. Moreover, for the purpose of switching from different kinds of patterns, the actor component of TC3 is extended with a structure of options, which comes to the TC3-Options. Finally, extensive experiments on the Indiegogo dataset not only demonstrate the effectiveness of our methods, but also validate our assumption that the entire pattern learned by TC3-Options is indeed the U-shaped one.
Paper
🔭 @DeepGravity
Recent years have witnessed the increasing interests in research of crowdfunding mechanism. In this area, dynamics tracking is a significant issue but is still under exploration. Existing studies either fit the fluctuations of time-series or employ regularization terms to constrain learned tendencies. However, few of them take into account the inherent decision-making process between investors and crowdfunding dynamics. To address the problem, in this paper, we propose a Trajectory-based Continuous Control for Crowdfunding (TC3) algorithm to predict the funding progress in crowdfunding. Specifically, actor-critic frameworks are employed to model the relationship between investors and campaigns, where all of the investors are viewed as an agent that could interact with the environment derived from the real dynamics of campaigns. Then, to further explore the in-depth implications of patterns (i.e., typical characters) in funding series, we propose to subdivide them into fast-growing and slow-growing ones. Moreover, for the purpose of switching from different kinds of patterns, the actor component of TC3 is extended with a structure of options, which comes to the TC3-Options. Finally, extensive experiments on the Indiegogo dataset not only demonstrate the effectiveness of our methods, but also validate our assumption that the entire pattern learned by TC3-Options is indeed the U-shaped one.
Paper
🔭 @DeepGravity
A New Framework for Query Efficient Active #ImitationLearning
We seek to align agent policy with human expert behavior in a #reinforcementlearning (#RL) setting, without any prior knowledge about dynamics, reward function, and unsafe states. There is a human expert knowing the rewards and unsafe states based on his preference and objective, but querying that human expert is expensive. To address this challenge, we propose a new framework for imitation learning (IL) algorithm that actively and interactively learns a model of the user's reward function with efficient queries. We build an adversarial generative model of states and a successor feature (SR) model trained over transition experience collected by learning policy. Our method uses these models to select state-action pairs, asking the user to comment on the optimality or safety, and trains a adversarial neural network to predict the rewards. Different from previous papers, which are almost all based on uncertainty sampling, the key idea is to actively and efficiently select state-action pairs from both on-policy and off-policy experience, by discriminating the queried (expert) and unqueried (generated) data and maximizing the efficiency of value function learning. We call this method adversarial reward query with successor representation. We evaluate the proposed method with simulated human on a state-based 2D navigation task, robotic control tasks and the image-based video games, which have high-dimensional observation and complex state dynamics. The results show that the proposed method significantly outperforms uncertainty-based methods on learning reward models, achieving better query efficiency, where the adversarial discriminator can make the agent learn human behavior more efficiently and the SR can select states which have stronger impact on value function. Moreover, the proposed method can also learn to avoid unsafe states when training the reward model.
Paper
🔭 @DeepGravity
We seek to align agent policy with human expert behavior in a #reinforcementlearning (#RL) setting, without any prior knowledge about dynamics, reward function, and unsafe states. There is a human expert knowing the rewards and unsafe states based on his preference and objective, but querying that human expert is expensive. To address this challenge, we propose a new framework for imitation learning (IL) algorithm that actively and interactively learns a model of the user's reward function with efficient queries. We build an adversarial generative model of states and a successor feature (SR) model trained over transition experience collected by learning policy. Our method uses these models to select state-action pairs, asking the user to comment on the optimality or safety, and trains a adversarial neural network to predict the rewards. Different from previous papers, which are almost all based on uncertainty sampling, the key idea is to actively and efficiently select state-action pairs from both on-policy and off-policy experience, by discriminating the queried (expert) and unqueried (generated) data and maximizing the efficiency of value function learning. We call this method adversarial reward query with successor representation. We evaluate the proposed method with simulated human on a state-based 2D navigation task, robotic control tasks and the image-based video games, which have high-dimensional observation and complex state dynamics. The results show that the proposed method significantly outperforms uncertainty-based methods on learning reward models, achieving better query efficiency, where the adversarial discriminator can make the agent learn human behavior more efficiently and the SR can select states which have stronger impact on value function. Moreover, the proposed method can also learn to avoid unsafe states when training the reward model.
Paper
🔭 @DeepGravity
World Programs for Model-Based Learning and Planning in Compositional State and Action Spaces
Some of the most important tasks take place in environments which lack cheap and perfect simulators, thus hampering the application of model-free #reinforcementlearning (#RL). While model-based RL aims to learn a dynamics model, in a more general case the learner does not know a priori what the action space is. Here we propose a formalism where the learner induces a world program by learning a dynamics model and the actions in graph-based compositional environments by observing state-state transition examples. Then, the learner can perform RL with the world program as the simulator for complex planning tasks. We highlight a recent application, and propose a challenge for the community to assess world program-based planning.
Paper
🔭 @DeepGravity
Some of the most important tasks take place in environments which lack cheap and perfect simulators, thus hampering the application of model-free #reinforcementlearning (#RL). While model-based RL aims to learn a dynamics model, in a more general case the learner does not know a priori what the action space is. Here we propose a formalism where the learner induces a world program by learning a dynamics model and the actions in graph-based compositional environments by observing state-state transition examples. Then, the learner can perform RL with the world program as the simulator for complex planning tasks. We highlight a recent application, and propose a challenge for the community to assess world program-based planning.
Paper
🔭 @DeepGravity
Long-Term Visitation Value for Deep Exploration in Sparse Reward #ReinforcementLearning
Reinforcement learning with sparse rewards is still an open challenge. Classic methods rely on getting feedback via extrinsic rewards to train the agent, and in situations where this occurs very rarely the agent learns slowly or cannot learn at all. Similarly, if the agent receives also rewards that create suboptimal modes of the objective function, it will likely prematurely stop exploring. More recent methods add auxiliary intrinsic rewards to encourage exploration. However, auxiliary rewards lead to a non-stationary target for the Q-function. In this paper, we present a novel approach that (1) plans exploration actions far into the future by using a long-term visitation count, and (2) decouples exploration and exploitation by learning a separate function assessing the exploration value of the actions. Contrary to existing methods which use models of reward and dynamics, our approach is off-policy and model-free. We further propose new tabular environments for benchmarking exploration in reinforcement learning. Empirical results on classic and novel benchmarks show that the proposed approach outperforms existing methods in environments with sparse rewards, especially in the presence of rewards that create suboptimal modes of the objective function. Results also suggest that our approach scales gracefully with the size of the environment. Source code is available at https://github.com/sparisi/visit-value-explore
Paper
🔭 @DeepGravity
Reinforcement learning with sparse rewards is still an open challenge. Classic methods rely on getting feedback via extrinsic rewards to train the agent, and in situations where this occurs very rarely the agent learns slowly or cannot learn at all. Similarly, if the agent receives also rewards that create suboptimal modes of the objective function, it will likely prematurely stop exploring. More recent methods add auxiliary intrinsic rewards to encourage exploration. However, auxiliary rewards lead to a non-stationary target for the Q-function. In this paper, we present a novel approach that (1) plans exploration actions far into the future by using a long-term visitation count, and (2) decouples exploration and exploitation by learning a separate function assessing the exploration value of the actions. Contrary to existing methods which use models of reward and dynamics, our approach is off-policy and model-free. We further propose new tabular environments for benchmarking exploration in reinforcement learning. Empirical results on classic and novel benchmarks show that the proposed approach outperforms existing methods in environments with sparse rewards, especially in the presence of rewards that create suboptimal modes of the objective function. Results also suggest that our approach scales gracefully with the size of the environment. Source code is available at https://github.com/sparisi/visit-value-explore
Paper
🔭 @DeepGravity
GitHub
GitHub - sparisi/visit-value-explore
Contribute to sparisi/visit-value-explore development by creating an account on GitHub.
Computational model discovery with #ReinforcementLearning
The motivation of this study is to leverage recent breakthroughs in artificial intelligence research to unlock novel solutions to important scientific problems encountered in computational science. To address the human intelligence limitations in discovering reduced-order models, we propose to supplement human thinking with artificial intelligence. Our three-pronged strategy consists of learning (i) models expressed in analytical form, (ii) which are evaluated a posteriori, and iii) using exclusively integral quantities from the reference solution as prior knowledge. In point (i), we pursue interpretable models expressed symbolically as opposed to black-box neural networks, the latter only being used during learning to efficiently parameterize the large search space of possible models. In point (ii), learned models are dynamically evaluated a posteriori in the computational solver instead of based on a priori information from preprocessed high-fidelity data, thereby accounting for the specificity of the solver at hand such as its numerics. Finally in point (iii), the exploration of new models is solely guided by predefined integral quantities, e.g., averaged quantities of engineering interest in Reynolds-averaged or large-eddy simulations (LES). We use a coupled deep reinforcement learning framework and computational solver to concurrently achieve these objectives. The combination of reinforcement learning with objectives (i), (ii) and (iii) differentiate our work from previous modeling attempts based on machine learning. In this report, we provide a high-level description of the model discovery framework with reinforcement learning. The method is detailed for the application of discovering missing terms in differential equations. An elementary instantiation of the method is described that discovers missing terms in the Burgers' equation.
Paper
🔭 @DeepGravity
The motivation of this study is to leverage recent breakthroughs in artificial intelligence research to unlock novel solutions to important scientific problems encountered in computational science. To address the human intelligence limitations in discovering reduced-order models, we propose to supplement human thinking with artificial intelligence. Our three-pronged strategy consists of learning (i) models expressed in analytical form, (ii) which are evaluated a posteriori, and iii) using exclusively integral quantities from the reference solution as prior knowledge. In point (i), we pursue interpretable models expressed symbolically as opposed to black-box neural networks, the latter only being used during learning to efficiently parameterize the large search space of possible models. In point (ii), learned models are dynamically evaluated a posteriori in the computational solver instead of based on a priori information from preprocessed high-fidelity data, thereby accounting for the specificity of the solver at hand such as its numerics. Finally in point (iii), the exploration of new models is solely guided by predefined integral quantities, e.g., averaged quantities of engineering interest in Reynolds-averaged or large-eddy simulations (LES). We use a coupled deep reinforcement learning framework and computational solver to concurrently achieve these objectives. The combination of reinforcement learning with objectives (i), (ii) and (iii) differentiate our work from previous modeling attempts based on machine learning. In this report, we provide a high-level description of the model discovery framework with reinforcement learning. The method is detailed for the application of discovering missing terms in differential equations. An elementary instantiation of the method is described that discovers missing terms in the Burgers' equation.
Paper
🔭 @DeepGravity
Curriculum for #ReinforcementLearning
A curriculum is an efficient tool for humans to progressively learn from simple concepts to hard problems. It breaks down complex knowledge by providing a sequence of learning steps of increasing difficulty. In this post, we will examine how the idea of curriculum can help reinforcement learning models learn to solve complicated tasks.
Article
🔭 @DeepGravity
A curriculum is an efficient tool for humans to progressively learn from simple concepts to hard problems. It breaks down complex knowledge by providing a sequence of learning steps of increasing difficulty. In this post, we will examine how the idea of curriculum can help reinforcement learning models learn to solve complicated tasks.
Article
🔭 @DeepGravity
Lil'Log
Curriculum for Reinforcement Learning
A curriculum is an efficient tool for humans to progressively learn from simple concepts to hard problems. It breaks down complex knowledge by providing a sequence of learning steps of increasing difficulty. In this post, we will examine how the idea of curriculum…