Rohin Shah
TalkRL: The Reinforcement Learning Podcast
00:00
The Three Types of Robustness in Reactor Learning Agents
A paper mentions three types of robustness. Can you briefly touchin the three types? Yes. So this was basically a categoryzation that we found helpful in generating the tests. And its a somewhat specific to reinforcement learning agents. It can further be a decomposed into two types. Is agent robustness without memory, where even, like, where the the test doesn't require the a i system to have any memory,. There's correct action that seems determinable even if the a system doesn't have memory.
Transcript
Play full episode
