Journal of Machine Learning Research· 2015Q1· Review
A comprehensive survey on safe reinforcement learning
- 1,196citations
- Q1SCImago
- 2015year
Short summary
This survey categorizes and analyzes two primary approaches to Safe Reinforcement Learning (SRL): modifying the optimality criterion with a safety factor, and altering the exploration process using external knowledge or risk metrics.
AI-generated from the title and abstract; the full text is not read.
Key points
- Safe Reinforcement Learning (SRL) aims to balance maximizing returns with ensuring system safety.
- Two core SRL approaches are identified: modifying optimality criteria and modifying exploration processes.
- Modifying optimality criteria involves adding a safety factor to standard reward calculations.
- Modifying exploration involves using external knowledge or risk metrics to guide learning.
- The survey analyzes existing literature and proposes future research avenues in SRL.
AI-generated from the title and abstract; the full text is not read.
Abstract
Safe Reinforcement Learning can be defined as the process of learning policies that maximize the expectation of the return in problems in which it is important to ensure reasonable system performance and/or respect safety constraints during the learning and/or deployment processes. We categorize and analyze two approaches of Safe Reinforcement Learning. The first is based on the modification of the optimality criterion, the classic discounted finite/infinite horizon, with a safety factor. The second is based on the modification of the exploration process through the incorporation of external knowledge or the guidance of a risk metric. We use the proposed classification to survey the existing literature, as well as suggesting future directions for Safe Reinforcement Learning.
The authors' abstract, as published at the source. Journal of Machine Learning Research, 2015 · DOI ↗
Continue with a free account
Ask the paper: 3 free questions a day about this paper; save it, get its citation, new summaries every day for your field. Takeaways are Premium.
Continue free on the webSign in with Google or Apple; no card needed. You come back to this paper.
On your phone:
Field: Software
SoftwareComputer Science