Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userCross-Domain Perceptual Reward Functions
84
0
0.0
(
0
)
Added by
Ashley Edwards
Publication date
2017
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
In reinforcement learning, we often define goals by specifying rewards within desirable states. One problem with this approach is that we typically need to redefine the rewards each time the goal changes, which often requires some understanding of the solution in the agents environment. When humans are learning to complete tasks, we regularly utilize alternative sources that guide our understanding of the problem. Such task representations allow one to specify goals on their own terms, thus providing specifications that can be appropriately interpreted across various environments. This motivates our own work, in which we represent goals in environments that are different from the agents. We introduce Cross-Domain Perceptual Reward (CDPR) functions, learned rewards that represent the visual similarity between an agents state and a cross-domain goal image. We report results for learning the CDPRs with a deep neural network and using them to solve two tasks with deep reinforcement learning.
rate research
Read More
Reinforcement learning problems are often described through rewards that indicate if an agent has completed some task. This specification can yield desirable behavior, however many problems are difficult to specify in this manner, as one often needs to know the proper configuration for the agent. When humans are learning to solve tasks, we often learn from visual instructions composed of images or videos. Such representations motivate our development of Perceptual Reward Functions, which provide a mechanism for creating visual task descriptions. We show that this approach allows an agent to learn from rewards that are based on raw pixels rather than internal parameters.
Traditional Reinforcement Learning (RL) problems depend on an exhaustive simulation environment that models real-world physics of the problem and trains the RL agent by observing this environment. In this paper, we present a novel approach to creating an environment by modeling the reward function based on empirical rules extracted from human domain knowledge of the system under study. Using this empirical rewards function, we will build an environment and train the agent. We will first create an environment that emulates the effect of setting cabin temperature through thermostat. This is typically done in RL problems by creating an exhaustive model of the system with detailed thermodynamic study. Instead, we propose an empirical approach to model the reward function based on human domain knowledge. We will document some rules of thumb that we usually exercise as humans while setting thermostat temperature and try and model these into our reward function. This modeling of empirical human domain rules into a reward function for RL is the unique aspect of this paper. This is a continuous action space problem and using deep deterministic policy gradient (DDPG) method, we will solve for maximizing the reward function. We will create a policy network that predicts optimal temperature setpoint given external temperature and humidity.
Autonomous agents optimize the reward function we give them. What they dont know is how hard it is for us to design a reward function that actually captures what we want. When designing the reward, we might think of some specific training scenarios, and make sure that the reward will lead to the right behavior in those scenarios. Inevitably, agents encounter new scenarios (e.g., new types of terrain) where optimizing that same reward may lead to undesired behavior. Our insight is that reward functions are merely observations about what the designer actually wants, and that they should be interpreted in the context in which they were designed. We introduce inverse reward design (IRD) as the problem of inferring the true objective based on the designed reward and the training MDP. We introduce approximate methods for solving IRD problems, and use their solution to plan risk-averse behavior in test MDPs. Empirical results suggest that this approach can help alleviate negative side effects of misspecified reward functions and mitigate reward hacking.
The design of reward functions in reinforcement learning is a human skill that comes with experience. Unfortunately, there is not any methodology in the literature that could guide a human to design the reward function or to allow a human to transfer the skills developed in designing reward functions to another human and in a systematic manner. In this paper, we use Systematic Instructional Design, an approach in human education, to engineer a machine education methodology to design reward functions for reinforcement learning. We demonstrate the methodology in designing a hierarchical genetic reinforcement learner that adopts a neural network representation to evolve a swarm controller for an agent shepherding a boids-based swarm. The results reveal that the methodology is able to guide the design of hierarchical reinforcement learners, with each model in the hierarchy learning incrementally through a multi-part reward function. The hierarchy acts as a decision fusion function that combines the individual behaviours and skills learnt by each instruction to create a smart shepherd to control the swarm.
In many real-world tasks, it is not possible to procedurally specify an RL agents reward function. In such cases, a reward function must instead be learned from interacting with and observing humans. However, current techniques for reward learning may fail to produce reward functions which accurately reflect user preferences. Absent significant advances in reward learning, it is thus important to be able to audit learned reward functions to verify whether they truly capture user preferences. In this paper, we investigate techniques for interpreting learned reward functions. In particular, we apply saliency methods to identify failure modes and predict the robustness of reward functions. We find that learned reward functions often implement surprising algorithms that rely on contingent aspects of the environment. We also discover that existing interpretability techniques often attend to irrelevant changes in reward output, suggesting that reward interpretability may need significantly different methods from policy interpretability.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
