Where Do Rewards Come From

By Andrew Barto, Richard L. Lewis, Satinder Singh. 2009

275 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation (2016, methodology): “This setup is similar to optimizing over the space of optimal reward functions to maximize fitness [25].”
Towards Continual Reinforcement Learning: A Review and Perspectives (2020, background): “This concept is foundational to work on intrinsic curiosity (Schmidhuber, 1991; Chentanez, Barto, and Singh, 2005; Singh, Lewis, and Barto, 2009; Barto, 2013; Achiam and Sastry, 2017) and learning an agent’s own curriculum (Schmidhuber, 2013; Justesen and Risi, 2018; Srinivasan, Bahdanau,…”
Playing hard exploration games by watching YouTube (2018, background): “Despite the challenges of designing reward functions [18, 36] or learning them using inverse reinforcement learning [1, 49], we also achieve human-level performance even in the absence of an environment reward signal.”
Inverse Reward Design (2017, background): “However, it provides no information about where this reward function comes from (Singh et al., 2010).”
Learning by Playing - Solving Sparse Reward Tasks from Scratch (2018, methodology): “(Hierarchical) Reinforcement Learning with the help of so called “intrinsic motivation” rewards (Chentanez et al., 2005; Singh et al., 2009) has, furthermore, been studied for controlling real robots by Ngo et al.”
Evolved Policy Gradients (2018, background): “Effectively, humans have access to very well shaped internal reward functions, derived from prior experience on other motor tasks, or perhaps from listening to and playing other musical instruments (36; 49).”
What Can Learned Intrinsic Rewards Capture? (2019, methodology): “To this end, we adopt the multi-lifetime setting of the Optimal Rewards Framework (Singh et al., 2009) in which an agent is initialised randomly at the start of each lifetime and then faces a stationary or non-stationary task drawn from some distribution.”
Emerging social awareness: Exploring intrinsic motivation in multiagent learning (2011, methodology): “Background • Intrinsically Motivated Reinforcement Learning (IMRL) • Extends classic Reinforcement Learning (RL) • Reward behaviors rather than consequences • Agent must maximize total reward = extrinsic + intrinsic • Extrinsic reward: • “normal” reward in RL, related with task (e.g. fulfillment of ”
Eureka: Human-Level Reward Design via Coding Large Language Models (2023, unknown): “Reward engineering is a long-standing challenge in reinforcement learning (Singh et al., 2010; Sutton & Barto, 2018).”
Neuronal Reward and Decision Signals: From Theories to Data. (2015, background): “Movements and effort may not only constitute a cost but also have intrinsic reward value (31, 358, 546, 547).”
Learning to Interactively Learn and Assist (2019, background): “hod can scale to raw pixel observations. 2 Related Work The traditional means of passing task information to an agent include specifying a reward function [4, 6] that can be hand-crafted for the task [46, 30, 9] and providing demonstrations [44, 1] before the agent starts training. More recent works”
Intrinsic Motivation and Reinforcement Learning (2013, background): “(p. 120, Samuelson and Swinkels 2006), Singh et al. (2009, 2010) placed reward processes in an evolutionary context, formulating a notion of an optimal reward function given an evolutionary fitness function and a distribution of environments.”
Safe Reinforcement Learning (2015, background): “Researchers have shown that often the most obvious reward function for a problem is not the most appropriate (Singh et al., 2009, Sorg et al., 2010).”
PERCEPTUAL GOAL SPECIFICATIONS FOR REINFORCEMENT LEARNING (2017, background): “Early work aimed to find rewards that generalized across multiple environments [50].”