
BI ViDA Panel Discussion: Deep RL and Dopamine
Brain Inspired
00:00
Defend Your Colleagues, Sir?
i suppose the fact that the training process has is not completely logically realistic might lead us astray. But on the other hand, i think having a model that worksand and simulates many of the encompasses, many of the things we are looking for and expect, is much beter then alternative,. which is not having a model. That's sounds like an important research direction. Ok. I think was also, i haven't read the reward as enough paper, a myself, so i just was curious what that paper was really saying. Were they really arguing against the importance of self supervised learning and sensory, sensory prediction, or what were they actually saying? Ecause that seems like a very
Transcript
Play full episode