In reinforcement learning, an agent learns to reach a set of goals by means of an external reward signal. In the natural world, intelligent organisms learn from internal drives, bypassing the need for external signals, which is beneficial for a wide range of tasks. Motivated by this observation, we propose to formulate an intrinsic objective as the mutual information between the goal states and the controllable states. This objective encourages the agent to take control of its environment. Subsequently, we derive a surrogate objective of the proposed reward function, which can be optimized efficiently. Lastly, we evaluate the developed framework in different robotic manipulation and navigation tasks and demonstrate the efficacy of our approach. A video showing experimental results is available at https://youtu.be/CT4CKMWBYz0
Your Fantastic Institute（优范研究院）；Their Fantastic Institute（泽范研究院）
The mutual information is a core statistical quantity that has applications in all areas of machine learning, whether this is in training of density models over multiple data modalities, in maximising the efficiency of noisy transmission channels, or when learning behaviour policies for exploration by artificial agents. Most learning algorithms that involve optimisation of the mutual information rely on the Blahut-Arimoto algorithm --- an enumerative algorithm with exponential complexity that is not suitable for modern machine learning applications. This paper provides a new approach for scala…
Reinforcement learning (RL) is one of the three basic paradigms of machine learning. It has demonstrated impressive performance in many complex tasks like Go and StarCraft, which is increasingly involved in smart manufacturing and autonomous driving. However, RL consistently suffers from the exploration-exploitation dilemma. In this paper, we investigated the problem of improving exploration in RL and introduced the intrinsically-motivated RL. In sharp contrast to the classic exploration strategies, intrinsically-motivated RL utilizes the intrinsic learning motivation to provide sustainable ex…
Politecnico di Milano（米兰理工大学）；Università di Bologna（博洛尼亚大学）
What is a good exploration strategy for an agent that interacts with an environment in the absence of external rewards? Ideally, we would like to get a policy driving towards a uniform state-action visitation (highly exploring) in a minimum number of steps (fast mixing), in order to ease efficient learning of any goal-conditioned policy later on. Unfortunately, it is remarkably arduous to directly learn an optimal policy of this nature. In this paper, we propose a novel surrogate objective for learning highly exploring and fast mixing policies, which focuses on maximizing a lower bound to the …
University of Edinburgh（爱丁堡大学）；University of Wisconsin–Madison（威斯康星大学麦迪逊分校）
Intrinsic rewards can improve exploration in reinforcement learning, but the exploration process may suffer from instability caused by non-stationary reward shaping and strong dependency on hyperparameters. In this work, we introduce Decoupled RL (DeRL) as a general framework which trains separate policies for intrinsically-motivated exploration and exploitation. Such decoupling allows DeRL to leverage the benefits of intrinsic rewards for exploration while demonstrating improved robustness and sample efficiency. We evaluate DeRL algorithms in two sparse-reward environments with multiple types…
ENSTA Paris（巴黎高等先进技术学院）；Institut Polytechnique de Paris（巴黎理工学院）
Despite the recent progress in deep learning and reinforcement learning, transfer and generalization of skills learned on specific tasks is very limited compared to human (or animal) intelligence. The lifelong, incremental building of common sense knowledge might be a necessary component on the way to achieve more general intelligence. A promising direction is to build world models capturing the true physical mechanisms hidden behind the sensorimotor interaction with the environment. Here we explore the idea that inferring the causal structure of the environment could benefit from well-chosen …
In continuous control, exploration is often performed through undirected strategies in which parameters of the networks or selected actions are perturbed by random noise. Although the deep setting of undirected exploration has been shown to improve the performance of on-policy methods, they introduce an excessive computational complexity and are known to fail in the off-policy setting. The intrinsically motivated exploration is an effective alternative to the undirected strategies, but they are usually studied for discrete action domains. In this paper, we investigate how intrinsic motivation …
RLeXplore: Accelerating Research in Intrinsically-Motivated Reinforcement Learning
Department of Computing（计算系）；The Hong Kong Polytechnic University（香港理工大学）；Mila Québec AI Institute & Université de Montréal（魁北克AI研究所及蒙特利尔大学）；Ningbo Institute of Digital Twin（宁波数字孪生研究所）；Eastern Institute of Technology, Ningbo（宁波东部技术研究所）
Massachusetts Institute of Technology (MIT)（麻省理工学院）；Higher Institute for Applied Sciences and Technology (HIAST)（高等应用科学与技术学院）
Discovering successful coordinated behaviors is a central challenge in Multi-Agent Reinforcement Learning (MARL) since it requires exploring a joint action space that grows exponentially with the number of agents. In this paper, we propose a mechanism for achieving sufficient exploration and coordination in a team of agents. Specifically, agents are rewarded for contributing to a more diversified team behavior by employing proper intrinsic motivation functions. To learn meaningful coordination protocols, we structure agents' interactions by introducing a novel framework, where at each timestep…
University of Hamburg（汉堡大学）；Knowledge Technology, Department of Informatics（知识技术与信息学系）
Combining model-based and model-free deep reinforcement learning has shown great promise for improving sample efficiency on complex control tasks while still retaining high performance. Incorporating imagination is a recent effort in this direction inspired by human mental simulation of motor behavior. We propose a learning-adaptive imagination approach which, unlike previous approaches, takes into account the reliability of the learned dynamics model used for imagining the future. Our approach learns an ensemble of disjoint local dynamics models in latent space and derives an intrinsic reward…
In this paper, we present a new intrinsically motivated actor-critic algorithm for learning continuous motor skills directly from raw visual input. Our neural architecture is composed of a critic and an actor network. Both networks receive the hidden representation of a deep convolutional autoencoder which is trained to reconstruct the visual input, while the centre-most hidden representation is also optimized to estimate the state value. Separately, an ensemble of predictive world models generates, based on its learning progress, an intrinsic reward signal which is combined with the extrinsic…
School of Electrical Engineering University of Belgrade, Serbia；Faculty of Mathematics University of Belgrade, Serbia
Options represent a framework for reasoning across multiple time scales in reinforcement learning (RL). With the recent active interest in the unsupervised learning paradigm in the RL research community, the option framework was adapted to utilize the concept of empowerment, which corresponds to the amount of influence the agent has on the environment and its ability to perceive this influence, and which can be optimized without any supervision provided by the environment's reward structure. Many recent papers modify this concept in various ways achieving commendable results. Through these var…
Exploration in sparse reward reinforcement learning remains an open challenge. Many state-of-the-art methods use intrinsic motivation to complement the sparse extrinsic reward signal, giving the agent more opportunities to receive feedback during exploration. Commonly these signals are added as bonus rewards, which results in a mixture policy that neither conducts exploration nor task fulfillment resolutely. In this paper, we instead learn separate intrinsic and extrinsic task policies and schedule between these different drives to accelerate exploration and stabilize learning. Moreover, we in…
Institute of Cognitive Science and Technologies, National Research Council（国家研究委员会认知科学与技术研究所）
In symbolic planning systems, the knowledge on the domain is commonly provided by an expert. Recently, an automatic abstraction procedure has been proposed in the literature to create a Planning Domain Definition Language (PDDL) representation, which is the most widely used input format for most off-the-shelf automated planners, starting from `options', a data structure used to represent actions within the hierarchical reinforcement learning framework. We propose an architecture that potentially removes the need for human intervention. In particular, the architecture first acquires options in …
Ludwig Maximilian University of Munich（慕尼黑大学）；Siemens AG（西门子公司）；Tsinghua University（清华大学）；University of California, Berkeley（加州大学伯克利分校）；Horizon Robotics（地平线机器人）
Reinforcement learning has been shown to be highly successful at many challenging tasks. However, success heavily relies on well-shaped rewards. Intrinsically motivated RL attempts to remove this constraint by defining an intrinsic reward function. Motivated by the self-consciousness concept in psychology, we make a natural assumption that the agent knows what constitutes itself, and propose a new intrinsic objective that encourages the agent to have maximum control on the environment. We mathematically formalize this reward as the mutual information between the agent state and the surrounding…
Learning goal-directed behavior in environments with sparse feedback is a major challenge for reinforcement learning algorithms. The primary difficulty arises due to insufficient exploration, resulting in an agent being unable to learn robust value functions. Intrinsically motivated agents can explore new behavior for its own sake rather than to directly solve problems. Such intrinsic behaviors could eventually help the agent solve tasks posed by the environment. We present hierarchical-DQN (h-DQN), a framework to integrate hierarchical value functions, operating at different temporal scales, …
Max Planck Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）
We present a novel intrinsically motivated agent that learns how to control the environment in the fastest possible manner by optimizing learning progress. It learns what can be controlled, how to allocate time and attention, and the relations between objects using surprise based motivation. The effectiveness of our method is demonstrated in a synthetic as well as a robotic manipulation environment yielding considerably improved performance and smaller sample complexity. In a nutshell, our work combines several task-level planning agent structures (backtracking search on task graph, probabilis…
University of Trento（特伦托大学）；Department of Information Engineering and Computer Science (DISI)（信息工程与计算机科学系（DISI））
In this work we consider partially observable environments with sparse rewards. We present a self-supervised representation learning method for image-based observations, which arranges embeddings respecting temporal distance of observations. This representation is empirically robust to stochasticity and suitable for novelty detection from the error of a predictive forward model. We consider episodic and life-long uncertainties to guide the exploration. We propose to estimate the missing information about the environment with the world model, which operates in the learned latent space. As a mot…
In many complex dynamical systems, artificial or natural, one can observe self-organization of patterns emerging from local rules. Cellular automata, like the Game of Life (GOL), have been widely used as abstract models enabling the study of various aspects of self-organization and morphogenesis, such as the emergence of spatially localized patterns. However, findings of self-organized patterns in such models have so far relied on manual tuning of parameters and initial states, and on the human eye to identify interesting patterns. In this paper, we formulate the problem of automated discovery…
Intrinsically motivated goal exploration algorithms enable machines to discover repertoires of policies that produce a diversity of effects in complex environments. These exploration algorithms have been shown to allow real world robots to acquire skills such as tool use in high-dimensional continuous state and action spaces. However, they have so far assumed that self-generated goals are sampled in a specifically engineered feature space, limiting their autonomy. In this work, we propose to use deep representation learning algorithms to learn an adequate goal space. This is a developmental 2-…
In vision-based reinforcement learning (RL) tasks, it is prevalent to assign auxiliary tasks with a surrogate self-supervised loss so as to obtain more semantic representations and improve sample efficiency. However, abundant information in self-supervised auxiliary tasks has been disregarded, since the representation learning part and the decision-making part are separated. To sufficiently utilize information in auxiliary tasks, we present a simple yet effective idea to employ self-supervised loss as an intrinsic reward, called Intrinsically Motivated Self-Supervised learning in Reinforcement…
Intrinsically motivated spontaneous exploration is a key enabler of autonomous developmental learning in human children. It enables the discovery of skill repertoires through autotelic learning, i.e. the self-generation, self-selection, self-ordering and self-experimentation of learning goals. We present an algorithmic approach called Intrinsically Motivated Goal Exploration Processes (IMGEP) to enable similar properties of autonomous learning in machines. The IMGEP architecture relies on several principles: 1) self-generation of goals, generalized as parameterized fitness functions; 2) select…
This paper studies the coupling of internally guided learning and social interaction, and more specifically the improvement owing to demonstrations of the learning by intrinsic motivation. We present Socially Guided Intrinsic Motivation by Demonstration (SGIM-D), an algorithm for learning in continuous, unbounded and non-preset environments. After introducing social learning and intrinsic motivation, we describe the design of our algorithm, before showing through a fishing experiment that SGIM-D efficiently combines the advantages of social learning and intrinsic motivation to gain a wide repe…
IMT Atlantique（大西洋国立高等工业技术学院）；ENSTA Paris（巴黎国立高等先进技术学院）；Institut Polytechnique de Paris（巴黎理工学院）；Inria（法国国家信息与自动化研究所）
In open-ended continuous environments, robots need to learn multiple parameterised control tasks in hierarchical reinforcement learning. We hypothesise that the most complex tasks can be learned more easily by transferring knowledge from simpler tasks, and faster by adapting the complexity of the actions to the task. We propose a task-oriented representation of complex actions, called procedures, to learn online task relationships and unbounded sequences of action primitives to control the different observables of the environment. Combining both goal-babbling with imitation learning, and activ…
Infants are experts at playing, with an amazing ability to generate novel structured behaviors in unstructured environments that lack clear extrinsic reward signals. We seek to mathematically formalize these abilities using a neural network that implements curiosity-driven intrinsic motivation. Using a simple but ecologically naturalistic simulated environment in which an agent can move and interact with objects it sees, we propose a "world-model" network that learns to predict the dynamic consequences of the agent's actions. Simultaneously, we train a separate explicit "self-model" that allow…
University of Hamburg（汉堡大学）；Knowledge Technology（知识技术学院）
Combining model-based and model-free learning systems has been shown to improve the sample efficiency of learning to perform complex robotic tasks. However, dual-system approaches fail to consider the reliability of the learned model when it is applied to make multiple-step predictions, resulting in a compounding of prediction errors and performance degradation. In this paper, we present a novel dual-system motor learning approach where a meta-controller arbitrates online between model-based and model-free decisions based on an estimate of the local reliability of the learned model. The reliab…
University of Hertfordshire（赫特福德大学）；University of Waterloo（滑铁卢大学）
A challenge in using robots in human-inhabited environments is to design behavior that is engaging, yet robust to the perturbations induced by human interaction. Our idea is to imbue the robot with intrinsic motivation (IM) so that it can handle new situations and appears as a genuine social other to humans and thus be of more interest to a human interaction partner. Human-robot interaction (HRI) experiments mainly focus on scripted or teleoperated robots, that mimic characteristics such as IM to control isolated behavior factors. This article presents a "robotologist" study design that allows…
Training machine learning models that are robust against adversarial inputs poses seemingly insurmountable challenges. To better understand adversarial robustness, we consider the underlying problem of learning robust representations. We develop a notion of representation vulnerability that captures the maximum change of mutual information between the input and output distributions, under the worst-case input perturbation. Then, we prove a theorem that establishes a lower bound on the minimum adversarial risk that can be achieved for any downstream classifier based on its representation vulner…
We introduce the Self-Adaptive Goal Generation - Robust Intelligent Adaptive Curiosity (SAGG-RIAC) architecture as an intrinsi- cally motivated goal exploration mechanism which allows active learning of inverse models in high-dimensional redundant robots. This allows a robot to efficiently and actively learn distributions of parameterized motor skills/policies that solve a corresponding distribution of parameterized tasks/goals. The architecture makes the robot sample actively novel parameterized tasks in the task space, based on a measure of competence progress, each of which triggers low-lev…
Google Brain（谷歌大脑）；MILA（蒙特利尔学习算法研究所）；Université de Montréal（蒙特利尔大学）；Deepmind（深度思维）
Mutual information maximization has emerged as a powerful learning objective for unsupervised representation learning obtaining state-of-the-art performance in applications such as object recognition, speech recognition, and reinforcement learning. However, such approaches are fundamentally limited since a tight lower bound of mutual information requires sample size exponential in the mutual information. This limits the applicability of these approaches for prediction tasks with high mutual information, such as in video understanding or reinforcement learning. In these settings, such technique…
Berkeley Artificial Intelligence Research Lab（伯克利人工智能研究实验室）；University of California, Berkeley（加州大学伯克利分校）
Intrinsically motivated artificial agents learn advantageous behavior without externally-provided rewards. Previously, it was shown that maximizing mutual information between agent actuators and future states, known as the empowerment principle, enables unsupervised stabilization of dynamical systems at upright positions, which is a prototypical intrinsically motivated behavior for upright standing and walking. This follows from the coincidence between the objective of stabilization and the objective of empowerment. Unfortunately, sample-based estimation of this kind of mutual information is c…
Several recent works in communication systems have proposed to leverage the power of neural networks in the design of encoders and decoders. In this approach, these blocks can be tailored to maximize the transmission rate based on aggregated samples from the channel. Motivated by the fact that, in many communication schemes, the achievable transmission rate is determined by a conditional mutual information term, this paper focuses on neural-based estimators for this information-theoretic quantity. Our results are based on variational bounds for the KL-divergence and, in contrast to some previo…
The Role of Mutual Information in Variational Classifiers
Facultad de Ingeniería, Universidad de Buenos Aires（布宜诺斯艾利斯大学工程学院）；CSC, CONICET（国家科学技术研究委员会计算机科学中心）；International Laboratory on Learning Systems (ILLS)（国际学习系统实验室）；Mila（米拉人工智能研究所）
Xidian University（西安电子科技大学）；Chongqing University of Posts and Telecommunications（重庆邮电大学）；Hong Kong Baptist University（香港浸会大学）；The Chinese University of Hong Kong (Shenzhen)（香港中文大学（深圳））；JD Explore Academy（京东探索研究院）；The University of Sydney（悉尼大学）
Deep neural networks (DNNs) are found to be vulnerable to adversarial noise. They are typically misled by adversarial samples to make wrong predictions. To alleviate this negative effect, in this paper, we investigate the dependence between outputs of the target model and input adversarial samples from the perspective of information theory, and propose an adversarial defense method. Specifically, we first measure the dependence by estimating the mutual information (MI) between outputs and the natural patterns of inputs (called natural MI) and MI between outputs and the adversarial patterns of …
We consider the theoretical problem of designing an optimal adversarial attack on a decision system that maximally degrades the achievable performance of the system as measured by the mutual information between the degraded signal and the label of interest. This problem is motivated by the existence of adversarial examples for machine learning classifiers. By adopting an information theoretic perspective, we seek to identify conditions under which adversarial vulnerability is unavoidable i.e. even optimally designed classifiers will be vulnerable to small adversarial perturbations. We present …
