Richard Sutton

University of Alberta | UAlberta · Department of Computing Science

Many reinforcement learning algorithms are built on an assumption that an agent interacts with an environment over fixed-duration, discrete time steps. However, physical systems are continuous in time, requiring a choice of time-discretization granularity when digitally controlling them. Furthermore, such systems do not wait for decisions to be mad...

Off-policy prediction—learning the value function for one policy from data generated while following another policy—is one of the most challenging problems in reinforcement learning. This article makes two main contributions: 1) it empirically studies 11 off-policy prediction learning algorithms with emphasis on their sensitivity to parameters, lea...

We show that discounted methods for solving continuing reinforcement learning problems can perform significantly better if they center their rewards by subtracting out the rewards' empirical average. The improvement is substantial at commonly used discount factors and increases further as the discount factor approaches one. In addition, we show tha...

To achieve the ambitious goals of artificial intelligence, reinforcement learning must include planning with a model of the world that is abstract in state and time. Deep learning has made progress with state abstraction, but temporal abstraction has rarely been used, despite extensively developed theory based on the options framework. One reason f...

Importance sampling is a central idea underlying off-policy prediction in reinforcement learning. It provides a strategy for re-weighting samples from a distribution to obtain unbiased estimates under another distribution. However, importance sampling weights tend to exhibit extreme variance, often leading to stability issues in practice. In this w...

In this work, we present a perspective on the role machine intelligence can play in supporting human abilities. In particular, we consider research in rehabilitation technologies such as prosthetic devices, as this domain requires tight coupling between human and machine. Taking an agent-based view of such devices, we propose that human–machine col...

In this paper, we explore an approach to auxiliary task discovery in reinforcement learning based on ideas from representation learning. Auxiliary tasks tend to improve data efficiency by forcing the agent to learn auxiliary prediction and control objectives in addition to the main task of maximizing reward, and thus producing better representation...

We show two average-reward off-policy control algorithms, Differential Q Learning (Wan, Naik, \& Sutton 2021a) and RVI Q Learning (Abounadi Bertsekas \& Borkar 2001), converge in weakly-communicating MDPs. Weakly-communicating MDPs are the most general class of MDPs that a learning algorithm with a single stream of experience can guarantee obtainin...

Herein we describe our approach to artificial intelligence research, which we call the Alberta Plan. The Alberta Plan is pursued within our research groups in Alberta and by others who are like minded throughout the world. We welcome all who would join us in this pursuit.

Value iteration (VI) is a foundational dynamic programming method, important for learning and planning in optimal control and reinforcement learning. VI proceeds in batches, where the update to the value of each state must be completed before the next batch of updates can begin. Completing a single batch is prohibitively expensive if the state spac...

We propose a new objective for option discovery that emphasizes the computational advantage of using options in planning. For a given set of episodic tasks and a given number of options, the objective prefers options that can be used to achieve a high return by composing few options. By composing few options, fast planning can be achieved. When fac...

We present three new diagnostic prediction problems inspired by classical-conditioning experiments to facilitate research in online prediction learning. Experiments in classical conditioning show that animals such as rabbits, pigeons, and dogs can make long temporal associations that enable multi-step prediction. To replicate this remarkable abilit...

To achieve the ambitious goals of artificial intelligence, reinforcement learning must include planning with a model of the world that is abstract in state and time. Deep learning has made progress in state abstraction, but, although the theory of time abstraction has been extensively developed based on the options framework, in practice options ha...

We extend the options framework for temporal abstraction in reinforcement learning from discounted Markov decision processes (MDPs) to average-reward MDPs. Our contributions include general convergent off-policy inter-option learning algorithms, intra-option algorithms for learning values and models, as well as sample-based planning variants of our...

Many off-policy prediction learning algorithms have been proposed in the past decade, but it remains unclear which algorithms learn faster than others. We empirically compare 11 off-policy prediction learning algorithms with linear function approximation on two small tasks: the Rooms task, and the High Variance Rooms task. The tasks are designed su...

Off-policy prediction -- learning the value function for one policy from data generated while following another policy -- is one of the most challenging subproblems in reinforcement learning. This paper presents empirical results with eleven prominent off-policy learning algorithms that use linear function approximation: five Gradient-TD methods, t...

In this article we hypothesise that intelligence, and its associated abilities, can be understood as subserving the maximisation of reward. Accordingly, reward is enough to drive behaviour that exhibits abilities studied in natural and artificial intelligence, including knowledge, learning, perception, social intelligence, language, generalisation...

In model-based reinforcement learning (MBRL), Wan et al. (2019) showed conditions under which the environment model could produce the expectation of the next feature vector rather than the full distribution, or a sample thereof, with no loss in planning performance. Such expectation models are of interest when the environment is stochastic and non-...

Catastrophic forgetting remains a severe hindrance to the broad application of artificial neural networks (ANNs), however, it continues to be a poorly understood phenomenon. Despite the extensive amount of work on catastrophic forgetting, we argue that it is still unclear how exactly the phenomenon should be quantified, and, moreover, to what degre...

We present three problems modeled after animal learning experiments designed to test online state construction or representation learning algorithms. Our test problems require the learning system to construct compact summaries of their past interaction with the world in order to predict the future, updating online and incrementally on each time ste...

Despite empirical success, the theory of reinforcement learning (RL) with value function approximation remains fundamentally incomplete. Prior work has identified a variety of pathological behaviours that arise in RL algorithms that combine approximate on-policy evaluation and greedification. One prominent example is policy oscillation, wherein an...

Intelligent assistants that follow commands or answer simple questions, such as Siri and Google search, are among the most economically important applications of AI. Future conversational AI assistants promise even greater capabilities and a better user experience through a deeper understanding of the domain, the user, or the user's purposes. But w...

Value-based methods for reinforcement learning lack generally applicable ways to derive behavior from a value function. Many approaches involve approximate value iteration (e.g., $Q$-learning), and acting greedily with respect to the estimates with an arbitrary degree of entropy to ensure that the state-space is sufficiently explored. Behavior base...

We introduce improved learning and planning algorithms for average-reward MDPs, including 1) the first general proven-convergent off-policy model-free control algorithm without reference states, 2) the first proven-convergent off-policy model-free prediction algorithm, and 3) the first learning algorithms that converge to the actual value function...

We explore fixed-horizon temporal difference (TD) methods, reinforcement learning algorithms for a new kind of value function that predicts the sum of rewards over a fixed number of future time steps. To learn the value function for horizon h, these algorithms bootstrap from the value function for horizon h−1, or some shorter horizon. Because no va...

We explore fixed-horizon temporal difference (TD) methods, reinforcement learning algorithms for a new kind of value function that predicts the sum of rewards over a $\textit{fixed}$ number of future time steps. To learn the value function for horizon $h$, these algorithms bootstrap from the value function for horizon $h-1$, or some shorter horizon...

Distribution and sample models are two popular model choices in model-based reinforcement learning (MBRL). However, learning these models can be intractable, particularly when the state and action spaces are large. Expectation models, on the other hand, are relatively easier to learn due to their compactness and have also been widely used for deter...

Distribution and sample models are two popular model choices in model-based reinforcement learning (MBRL). However, learning these models can be intractable, particularly when the state and action spaces are large. Expectation models, on the other hand, are relatively easier to learn due to their compactness and have also been widely used for deter...

There is a long history of using meta learning as representation learning, specifically for determining the relevance of inputs. In this paper, we examine an instance of meta-learning in which feature relevance is learned by adapting step size parameters of stochastic gradient descent---building on a variety of prior work in stochastic approximatio...

Emphatic Temporal Difference (ETD) learning has recently been proposed as a convergent off-policy learning method. ETD was proposed mainly to address convergence issues of conventional Temporal Difference (TD) learning under off-policy training but it is different from conventional TD learning even under on-policy training. A simple counterexample...

This paper investigates the problem of online prediction learning, where learning proceeds continuously as the agent interacts with an environment. The predictions made by the agent are contingent on a particular way of behaving, represented as a value function. However, the behavior used to select actions and generate the behavior data might be di...

Temporal difference (TD) learning is an important approach in reinforcement learning, as it combines ideas from dynamic programming and Monte Carlo methods in a way that allows for online and incremental model-free learning. A key idea of TD learning is that it is learning predictive knowledge about the environment in the form of value functions, f...

We apply neural nets with ReLU gates in online reinforcement learning. Our goal is to train these networks in an incremental manner, without the computationally expensive experience replay. By studying how individual neural nodes behave in online training, we recognize that the global nature of ReLU gates can cause undesirable learning interference...

In this paper, we introduce a method for adapting the step-sizes of temporal difference (TD) learning. The performance of TD methods often depends on well chosen step-sizes, yet few algorithms have been developed for setting the step-size automatically for TD learning. An important limitation of current methods is that they adapt a single step-size...

The relationship between a reinforcement learning (RL) agent and an asynchronous environment is often ignored. Frequently used models of the interaction between an agent and its environment, such as Markov Decision Processes (MDP) or Semi-Markov Decision Processes (SMDP), do not capture the fact that, in an asynchronous environment, the state of th...

The relationship between a reinforcement learning (RL) agent and an asynchronous environment is often ignored. Frequently used models of the interaction between an agent and its environment, such as Markov Decision Processes (MDP) or Semi-Markov Decision Processes (SMDP), do not capture the fact that, in an asynchronous environment, the state of th...

This paper investigates estimating the variance of a temporal-difference learning agent's update target. Most reinforcement learning methods use an estimate of the value function, which captures how good it is for the agent to be in a particular state and is mathematically expressed as the expected sum of discounted future rewards (called the retur...

This paper investigates estimating the variance of a temporal-difference learning agent's update target. Most reinforcement learning methods use an estimate of the value function, which captures how good it is for the agent to be in a particular state and is mathematically expressed as the expected sum of discounted future rewards (called the retur...

This work presents an overarching perspective on the role that machine intelligence can play in enhancing human abilities, especially those that have been diminished due to injury or illness. As a primary contribution, we develop the hypothesis that assistive devices, and specifically artificial arms and hands, can and should be viewed as agents in...

We consider off-policy temporal-difference (TD) learning in discounted Markov decision processes, where the goal is to evaluate a policy in a model-free way by using observations of a state process generated without executing the policy. To curb the high variance issue in off-policy TD learning, we propose a new scheme of setting the $\lambda$-para...

We consider off-policy temporal-difference (TD) learning in discounted Markov decision processes, where the goal is to evaluate a policy in a model-free way by using observations of a state process generated without executing the policy. To curb the high variance issue in off-policy TD learning, we propose a new scheme of setting the $\lambda$-para...

Unifying seemingly disparate algorithmic ideas to produce better performing algorithms has been a longstanding goal in reinforcement learning. As a primary example, TD($\lambda$) elegantly unifies one-step TD prediction with Monte Carlo methods through the use of eligibility traces and the trace-decay parameter $\lambda$. Currently, there are a mul...

We develop an extension of the Rescorla-Wagner model of associative learning. In addition to learning from the current trial, the new model supposes that animals store and replay previous trials, learning from the replayed trials using the same learning rule. This simple idea provides a unified explanation for diverse phenomena that have proved cha...

Representations are fundamental to Artificial Intelligence. Typically, the performance of a learning system depends on its data representation. These data representations are usually hand-engineered based on some prior domain knowledge regarding the task. More recently, the trend is to learn these representations through deep neural networks as the...

The temporal-difference methods TD(lambda) and Sarsa(lambda) form a core part of modern reinforcement learning. Their appeal comes from their good performance, low computational cost, and their simple interpretation, given by their forward view. Recently, new versions of these methods were introduced, called true online TD(lambda) and true online S...

An important application of interactive machine learning is extending or amplifying the cognitive and physical capabilities of a human. To accomplish this, machines need to learn about their human users' intentions and adapt to their preferences. In most current research, a user has conveyed preferences to a machine using explicit corrective or ins...

Background: Myoelectric prostheses currently used by amputees can be difficult to control. Machine learning, and in particular learned predictions about user intent, could help to reduce the time and cognitive load required by amputees while operating their prosthetic device. Objectives: The goal of this study was to compare two switching-based...

We consider how to learn multi-step predictions efficiently. Conventional algorithms wait until observing actual outcomes before performing the computations to update their predictions. If predictions are made at a high rate or span over a large amount of time, substantial computation can be required to store all relevant observations and to update...

Importance sampling is an essential component of model-free off-policy learning algorithms. Weighted importance sampling (WIS) is generally considered superior to ordinary importance sampling but, when combined with function approximation , it has hitherto required computational complexity that is O(n 2) or more in the number of features. In this p...

The true online TD({\lambda}) algorithm has recently been proposed (van Seijen and Sutton, 2014) as a universal replacement for the popular TD({\lambda}) algorithm, in temporal-difference learning and reinforcement learning. True online TD({\lambda}) has better theoretical properties than conventional TD({\lambda}), and the expectation is that it a...

In this paper we introduce the idea of improving the performance of parametric temporal-difference (TD) learning algorithms by selectively emphasizing or de-emphasizing their updates on different time steps. In particular, we show that varying the emphasis of linear TD($\lambda$)'s updates in a particular way causes its expected update to become st...

In reinforcement learning, the notions of experience replay, and of planning as learning from replayed experience, have long been used to find good policies with minimal training data. Replay can be seen either as model-based reinforcement learning, where the store of past experiences serves as the model, or as a way to avoid a conventional model o...

The present experiment tested whether or not the time course of a conditioned eyeblink response, particularly its duration, would expand and contract, as the magnitude of the conditioned response (CR) changed massively during acquisition, extinction, and reacquisition. The CR duration remained largely constant throughout the experiment, while CR on...

Van Seijen and Sutton (2014) recently proposed a new version of the linear TD(λ) learning algorithm that is exactly equivalent to an online forward view and that empirically performed better than its classical counterpart in both prediction and control problems. However, their algorithm is restricted to on-policy learning. In the more general case...

TD(λ) is a core algorithm of modern reinforcement learning. Its appeal comes from its equivalence to a clear and conceptually simple forward view, and the fact that it can be implemented online in an inexpensive manner. However, the equivalence between TD(λ) and the for-ward view is exact only for the off-line version of the algorithm (in which upd...

Q-learning, the most popular of reinforcement learning algorithms, has always included an extension to eligibility traces to enable more rapid learning and improved asymptotic performance on non-Markov problems. The parameter smoothly shifts on-policy algorithms such as TD(λ) and Sarsa(λ) from a pure bootstrapping form (λ = 0) to a pure Monte Carlo...

In this work we explore the use of reinforcement learning (RL) to help with human decision making, combining state-of-the-art RL algorithms with an application to prosthetics. Managing human-machine interaction is a problem of considerable scope, and the simplification of human-robot interfaces is especially important in the domains of biomedical t...

Efficient planning plays a crucial role in model-based reinforcement learning. Traditionally, the main planning operation is a full backup based on the current estimates of the successor states. Consequently, its computation time is proportional to the number of successor states. In this paper, we introduce a new planning backup that uses only the...

Integrating learned predictions into a prosthetic control system promises to enhance multi-joint prosthesis use by amputees. In this article, we present a preliminary study of different cases where it may be beneficial to use a set of temporally extended predictions - learned and maintained in real time - within an engineered or learned prosthesis...

PREDICTING THE FUTURE has long been regarded as a powerful means to improvement and success. The ability to make accurate and timely predictions enhances our ability to control our situation and our environment. Assistive robotics is one prominent area where foresight of this kind can bring improved quality of life. In this article, we present a ne...

Rabbits were classically conditioned using compounds of tone and light conditioned stimuli (CSs) presented with either simultaneous onsets (Experiment 1) or serial onsets (Experiment 2) in a delay conditioning paradigm. Training with the simultaneous compound reduced the likelihood of a conditioned response (CR) to the individual CSs ("mutual overs...

Efficient planning plays a crucial role in model-based reinforcement learning. Traditionally, the main planning operation is a full backup based on the current estimates of the successor states. Consequently, its computation time is proportional to the number of successor states. In this paper, we introduce a new planning backup that uses only the...

Intelligent systems promise to amplify, augment, and extend innate human abilities. A principal example is that of assistive rehabilitation robots—artificial intelligence and machine learning enable new electrome-chanical systems that restore biological functions lost through injury or illness. In order for an intelligent machine to assist a human...

Several robot capabilities rely on predictions about the temporally extended consequences of a robot's behaviour. We describe how a robot can both learn and make many such predictions in real time using a standard algorithm. Our experiments show that a mobile robot can learn and make thousands of accurate predictions at 10 Hz. The predictions are a...

Predictions and anticipations form a basis for pattern-recognition-based control systems, including those used in next-generation prostheses and assistive devices. In this work, we outline how new methods in real-time prediction learning provide an approach to one of the principal open problems in multi-function myoelectric control—robust, ongoing,...

The temporal-difference (TD) algorithm from reinforcement learning provides a simple method for incrementally learning predictions of upcoming events. Applied to classical conditioning, TD models suppose that animals learn a real-time prediction of the unconditioned stimulus (US) on the basis of all available conditioned stimuli (CSs). In the TD mo...

Existing robot algorithms demonstrate several capabil-ities that are enabled by a robot's knowledge of the temporally-extended consequences of its behaviour. This knowledge consists of real-time predictions—predictions that are conventionally computed by iterating a small one-timestep model of the robot's dynamics. Given the utility of such predict...

We pursue a life-long learning approach to artificial intelligence that makes extensive use of reinforcement learning algorithms. We build on our prior work with general value functions (GVFs) and the Horde architecture. GVFs have been shown able to represent a wide variety of facts about the world's dynamics that may be useful to a long-lived agen...

Reinforcement learning methods are often con-sidered as a potential solution to enable a robot to adapt to changes in real time to an unpredictable environment. However, with continuous action, only a few existing algorithms are practical for real-time learning. In such a setting, most effective methods have used a parameterized policy structure, o...

A general problem for human-machine interaction occurs when a machine's controllable dimensions outnumber the control channels available to its human user. In this work, we examine one prominent example of this problem: amputee switching between the multiple functions of a powered artificial limb. We propose a dynamic switching approach that learns...

We consider the problem of efficiently learning optimal control policies and value functions over large state spaces in an online setting in which estimates must be available after each interaction with the world. This paper develops an explicitly model-based approach extending the Dyna architecture to linear function approximation. Dynastyle plann...

This paper presents the first actor-critic algorithm for off-policy reinforcement learning. Our algorithm is online and incremental, and its per-time-step complexity scales linearly with the number of learned weights. Previous work on actor-critic algorithms is limited to the on-policy setting and does not take advantage of the recent advances in o...

L’apprentissageparrenforcementestsouventconsidérécommeunesolutionpotentiellepour permettre à un robot de s’adapter en temps réel aux changements imprédictibles d’un environnement ; mais avec des actions continues, peu d’algorithmes existants sont utilisables pour un tel apprentissage temps réel. Les méthodes les plus efficaces utilisent une politiq...

Temporal-difference learning is one of the most successful and broadly applied solutions to the reinforcement learning problem; it has been used to achieve master-level play in chess, checkers and backgammon. The key idea is to update a value function from episodes of real experience, by bootstrapping from future value estimates, and using value fu...