I like the enforced indentation of Python that many don't like because I hate parenthetic typing and redundant semicolons. I like the shell interface, but why do some think Python is de facto for machine learning?
Even with straight rectified linear activation, because of sheer dimensionality, ...
From https://stackoverflow.com/questions/36370129/does-tensorflow-use-automatic-or-symbolic-gradients, I understood TensorFlow requires all the operations in the Graph to be explicit formulas (instead of black-boxes, such as raw python functions) to do Automatic Differentiation. Then it will do s...
@JohnDoucette I was reading old questions so that to clean them up and I found this question. You had told me before that you had done one of your thesis on novelty search. Is that right?
Btw, the paper by Lehman and Stanley that I cite is actually very accessible. It is actually a good introduction to the topic. It feels like a tutorial.
re novelty search have been investigating this area heavily last few yrs, think its extremely promising & maybe a rosetta stone, & collected very many refs but havent run into stanleys research, coincidentally did recently, heres a nice writeup.
Some discussions were started in the meta-section which questions are offtopic questions and should be closed next. I have to admit that I'm supporting such attempts because the SE.AI forum has in comparison to other Stackexchange websites a small close-percentage.
It make sense to go a step bac...
@JohnDoucette You gave the following answer https://ai.meta.stackexchange.com/a/1596/2444 several weeks ago. I can't really understand how you could think that this user isn't actually just TROLLING or just a BOT.
> Playing chess with a blind search algorithm like Monte Carlo tree search is a very good idea, because it helps to raise the energy consumption of your 64 core workstation. Using the maximum capacity of a cpu and ignoring any possible kind of heuristics is a best practice method in proofing real scientific progress.
You got to be kidding, if you think that this user is a normal user, who only has a language problem.
I can't understand how this user got so much reputation and therefore power on this site. Clearly, the site has been malfunctioning.
To me, I say it publicly, the presence of this user (and related ones, such as Douglas Daseeco) on this site is the biggest problem we have had. He, she or it really deteriorated the overall quality of the site (by providing very poor questions and answers). Usually, trolls do not live long, but this is not the case, unfortunately.
@DukeZhou and @JadenTravnik Is it possible to know WHY you didn't ban or delete this user? Moderators can suspend or delete users. See Who are the site moderators, and what is their role here?. There's absolutely no excuse to let this type of users incredibly blossom in terms of reputation while deteriorating the site. Isn't the following quoted paragraph sufficient to understand that this user isn't anything but a troll?
> Playing chess with a blind search algorithm like Monte Carlo tree search is a very good idea, because it helps to raise the energy consumption of your 64 core workstation. Using the maximum capacity of a cpu and ignoring any possible kind of heuristics is a best practice method in proofing real scientific progress.
@ManuelRodriguez If I become a moderator and you post more spam answers or questions, you will be immediately suspended or deleted (if that's possible). I'm warning you. I won't tolerate spam or trolling.
@nbro yeah, I did a thesis on novelty search back in 2009. Your answer was a nice summary of that paper, and answered the user's question well.
I have meet a lot of 'larger than life' people over time. Douglas does not appear to me to be a troll, but he has a very unconventional background, and tended to fixate on issues that stemmed from misuse of terminology (a bit like in that Tree vs. Graph search question a few weeks back).
Manuel seems to me to be someone of unclear motives. Some actions he takes seem to be explainable only as trolling or sarcasm. But others seem genuine, and suggest to me a language or knowledge barrier. Again, I've met a lot of people over time who act in ways that don't make a lot of sense to me, but do so from a well meaning place, by and large. Because of this, I wouldn't feel confident labeling Manuel a troll.
@JohnDoucette Have you seen the comment response he's given to the comment I've written under the mentioned answer?
I wrote
> "Playing chess with a blind search algorithm like Monte Carlo tree search is a very good idea, because it helps to raise the energy consumption of your 64 core workstation", if you are not trolling, then I don't know what you are doing.
He writes
> The OP is trying to play a board game with minimax, alpha-beta pruning and MCTS. A look into performance statistics will show that none of these algorithm is prepared for doing so. All of them will generate a large gametree. It's important to make clear what the limits of the mentioned algorithms are.
which makes me think that he did not understand the comment or is just trolling. Any other explanation?
After reading over that question and answer, I think the most probable explanation is a language issue coupled with a mistaken belief about the feasibility of using search to play boardgames.
I think he's trying to say 'MCTS is probably not going to be able to solve this problem without expending a huge amount of computational effort.'
There is absolutely no language issue here. His answers are very often grammatically correct and mention specific and even advanced topics, which unfortunately are often out of context.
I guess I find his use of language to be consistent with people I have meet for whom English is a second language. His grammar is decent, but his vocab and idioms are off. He speaks like some Germans I've met.
I think that sarcasm and trolling are distinct actions, but agree that, in this case, his answer is not helpful.
I know a little bit of German and I've met several German people in the past, but I've never noticed this
It seems to me that, when he gets something right in his answers, it's because he's put a lot of effort into connecting his sentences in a coherent way, which makes me think that he's a bot.
@nbro @nbro It sounds a bit uncommon, but I'm happy with receiving downvotes for my answers. It shows, that the community is trying to improve the overall quality and creates a pressure to the users who are attempt to answer a question. In the given case with the checkers/MCTS problem, the question has received a lot of upvotes (+12) while all answers together have accumulated -8 downvotes. This kind of relationship is great.
@nbro Perhaps it make sense to explain how group interaction is working. Not a single user is allowed to moderate, but a Q&A website produces a chaotic behavior which is the result of conflicting social roles. There is a person who creates a question, another person provides an answer, the next person is judging about the quality and so on. Every user can be replaced by someone who is able to play the same social role.
@nbro That was a case where there were serious concerns about vote rigging, for which our prior mod took the lead on suspensions. Total bans should be used with sparingly, and DD often had a number of upvotes on his questions, problematic as they were.
End of the day, downvoting is the best remedy. (Typically bans would be for more conventional transgressivity, such as injecting charged political or social topics.)
@nbro be diplomatic, even when strongly arguing a point
I am recalled to another user (not mentioned in this convo) who I suspected was suffering from mental illness (either mania or mild schizophrenia.) Their posts were problematic, and diminished the stack, but it seemed best to deal with it gently, with compassion.
Ultimately I think they recognized their posts weren't getting traction and chose to forgo continued posting.
@DukeZhou The terminology of remedy and transgressivity was introduced by sociology to describe group behavior with a certain bias. The problem with the models used in sociology is, that they are based on information centralization. The assumption is, that somewhere in the core, the management is located which is observing and directing the group. This kind of vocabulary was useful in the 19th century for handling easy moderation problems but it's less suited for a modern scrum based management.
You allow spam to proper, at the expense of the quality of the site! This problem is as severe as voting irregularities or even worse! Well done!
> The terminology of remedy and transgressivity was introduced by sociology to describe group behavior with a certain bias. The problem with the models used in sociology is, that they are based on information centralization. The assumption is, that somewhere in the core, the management is located which is observing and directing the group. This kind of vocabulary was useful in the 19th century for handling easy moderation problems but it's less suited for a modern scrum based management.
But the problem is, voting irregularities, or transgressive content, is much more objective. Where a greater degree of subjectivity (or fuzziness) is involved, the measures must be less stringent.
I will probably not become a moderator, but if I become a moderator, I will not tolerate trolls or spams. There's no room for niceness with trolls who have nothing better to do than bothering people (who just need help) or spreading wrong information.
I just like to keep in mind that we're often dealing with confidence levels as opposed to perfect certainty. (There have been times when I've been certain and wrong.)
I think that AI is really a nice and promising field, but it makes me sad when there's a site like this with people that, rather than raising the credibility of the field, only deteriorate it.
I'm not always right, that's for sure, I can even provide wrong answers, but I am pretty sure about this case.
And I think people need to be cognizant that you're one of the favorites for the upcoming election, and you're not the only user calling for stricter management.
I don't deny we've been pursuing a "greedy policy" in the prior era, but, with the elections, we're going to be entering a new era for the stack.
Our metrics on Area 51 suggest we can afford to start being stricter.
@nbro There is no need that SE.AI has a high credibility, it's enough, if users get the answer they need. Trust is something which is given to the website, if an OP accepts an answer and it has to be earned in each single case.
@ManuelRodriguez No, in this case, credibility is very important and credibility will only come with quality. The answers will be available to future readers and not just to the initial asker, so our site can potentially have more utility than simply helping the specific user that asks the question. So, we should have the responsibility of providing high-quality questions and answers, which can serve as a future reference!
@ManuelRodriguez If you keep on spreading wrong information, providing wrong answers, etc., our site loses quality and thus credibility. So, please, stop.
Correct. The combination of a dialogue based answering plus the ability to search in the archive is a unique feature, not available in normal publications. The papers which are provided at Arxiv aren't interactive and chatbased interaction with a supervisor during phd writing isn't archived. So it make sense to argue, that SE.AI is able to earn a dedicated credibility.
I like the enforced indentation of Python that many don't like because I hate parenthetic typing and redundant semicolons. I like the shell interface, but why do some think Python is de facto for machine learning?
Even with straight rectified linear activation, because of sheer dimensionality, ...
I am currently learning C#. Is it a good language to get started with AI? Or should use another programming language?
