Anyway, Jaden has recently given an answer, but it would be better if he was a lot more active, not necessarily answering questions, but handling flagged comments, questions, etc. In fact, we have several recently asked questions that should have already been closed, but no moderator has yet taken action...
Can anyone point me in the direction of a nice graph that depicts the "family tree", or hierarchy, of RL algorithms (or models)? For example, it splits the learning into TD and Monte Carlo methods, under which is listed all of the algorithms with their respective umbrella terms. Beneath each algo...
I want to be able to debug my keras LSTM model. For example, I want to be able to check the values of the input/output gates, cell states and hidden states at every time-step during prediction. Are there any tools out there that can achieve this?
Thanks!
@Mithical Yes, I know, but our community is often very slow to do it. You don't have to say this thing to me but to all other community members. But shouldn't moderators intervene anyway? IMHO, they can and they should, in the case a question needs evidently to be closed
I'd argue it's off-topic as a general programming question. It happens to be related to a tool that is used in AI, but hasn't got anything to do with AI itself.
@OliverMason Yes, I also think it's off-topic (although this could fall into the category resource-request), but, ideally, this question should also be on-topic, if we merged Data Science SE with AI SE.
Incredible! 8 people currently here today! That's a record! I wish this chat was always like this, more active and interactive.
@OliverMason Well, I am also hesitant to close these questions because I think the given answers can be helpful. I think we should be more strict when it comes to too broad posts (e.g. posts with multiple questions), primarily opinion-based or clearly off-topic questions, rather than these questions that, in theory, are related to AI, but because DS SE already exists for this purpose, they "should" be considered off-topic.
What would happen if an agent were programmed with two utility functions which are one the opposite of the other? Would one prevail or will they cancel each other out? Are there studies in this direction?
@JohnDoucette You should really nominate yourself. In case I am not chosen, I think you're the person (who is willing to take this responsibility) with the best knowledge of the AI field and mentality to become a moderator. There are some people that think I may not be a good moderator, and maybe they aren't completely wrong (see https://ai.stackexchange.com/election/1#post-18078), so that's another reason for you to nominate yourself.
I want to be able to debug my keras LSTM model. For example, I want to be able to check the values of the input/output gates, cell states and hidden states at every time-step during prediction. Are there any tools out there that can achieve this?
Thanks!
What would happen if an agent were programmed with two utility functions which are one the opposite of the other? Would one prevail or will they cancel each other out? Are there studies in this direction?
