5 ms·
Ask HN: How to prevent learning in reinforcement learning?
With the dangers of AI lurking around the corner, I am wondering how much research has been done in preventing AI algorithms from learning and how to intentially confuse or disrupt them. For example, what can be done to a data stream to prevent reinforcement learning from learning?
- b20000 4y agocame across this interesting paper - https://onlinelibrary.wiley.com/doi/10.1002/sat.1418?af=R https://onlinelibrary.wiley.com/doi/10.1002/sat.1418?af=R
- mindcrime 4y agoIt's not specific to reinforcement learning, but you might find this of interest: https://en.wikipedia.org/wiki/Adversarial_machine_learning https://en.wikipedia.org/wiki/Adversarial_machine_learning
- b20000 4y agomuch appreciated