41–50 of 116 posts
If the author's here: some of the math formulas don't render correctly. In particular, 10^170 is parsed as 10^{1}70, and $5478$ shows up without TeX applied to it.
Thanks, fixed those
> A new paper was released a few days detailing a new neural net
I believe you mean "a few days ago"?
Mageek, any reason why they haven't applied this to chess yet?
One of the things that amazed me the most about AlphaGo Zero was that they didn't do any tricks to minimize the latency of the network evaluation!
Still, it's certainly worth a try, I'd be extremely interested to see what style of chess a self-trained MCTS chess version would have :).
The main reason AlphaGo Zero learns so much faster than its predecessors is because it uses temporal-difference learning.[1] This effectively removes a huge amount of the value network's state space for the learning algorithm to search through, since it bakes in the assumption that a move's value ought to equal that of the best available move in the following board position, which is exactly what you'd expect for a g…
The expected value of a contract at time T, estimated at some time t expected estimate, made n months from now, of the value [12-n] months later, will also be $100. This allows you to shrink the state space considerably.
You can usually work out the payoff of a derivatives in different scenarios given rational exercise decisions by all contract participants. The calculation assumes that every market participant makes the best possible decision given the information they had available at the time by either explicitly or implicitly building a tree and working backwards, back-propagating the 'future' value back to the root.
This closely resembles the modeling of a discrete adversarial game, except the payoffs need to make reference to random market variables like the stock price, so the tree nodes are not just indexed by participant action, but also by variables.
There's actually a nice resemblance between the Longstaff-Schwarz method of pricing American options and MCTS + Alphago, except that the former is using kernel regressions instead of deep neural nets and we sample from a continuous space with an assumed probability distribution instead of a discrete space guided by a policy network [1].
[1] https://people.math.ethz.ch/~hjfurrer/teaching/LongstaffSchw...
Earlier quoted context omitted.
What do you mean? AlphaGo Zero has played itself, and they have published 20 of those games.
the OP is referring to the use of a GAN, which is a different type of setup. GAN's were not used for alphaGo, as the article points out Deepmind uses reinforcement learning and MCTS.
If the author's here: some of the math formulas don't render correctly. In particular, 10^170 is parsed as 10^{1}70, and $5478$ shows up without TeX applied to it.
Been following the AlphaGo Zero developments, which leap-frog what was going on in the movie (although still very much worth seeing).
One thing I was curious about is if Go would be considered solved, either hard or weakly solved, since AlphaGo Zero at this point doesn't seem to be able to be beat by any living human. Wikipedia does not list it as solved in either sense, and I was wondering if this was an oversight.
Saw the AlphaGo movie at a festival recently. Been following the AlphaGo Zero developments, which leap-frog what was going on in the movie (although still very much worth seeing). One thing I was curious about is if Go would be considered solved, either hard or weakly solved, since AlphaGo Zero at this point doesn't seem to be able to be beat by any living human. Wikipedia does not list it as solved in either sense,…
I wonder how the STYLE of Alpha Go Zero is regarded by human experts. Is it far different from AlphaGo? Why bother learning from AlphaGo if they can learn from AlphaGo Zero? Did they unleash a second "Master" program? I am wondering if the "better" strategy moves are now super wacky and weird and break all theory.
At least initial reports are that alphaGo Zero is more human-like than Master. Zero packs even more of the inhuman ability to pick the most critical part of the board for each move, but less weird looking stuff. In fact, one of the obvious differences between AlphaGo Zero and top human players, is much more play on safe opening spots, which has been out of fashion among human pros for a hundred years or so.
Empirically, this is not correct. The original AlphaGo achieved a 57% accuracy at predicting expert moves.
I can't find an exact number, but based on the graph in the nature article, AlphaGo Zero has a less than 50% accuracy at predicting human moves. Eyeballing the graph, it looks like the supervized learning variant of AlphaGo Zero scored Of course, it is not clear that ability to predict the move humans will play is the best metric to measure how human like a computer plays. It is just the only objective metric we have. Although, if this were an actual research question, we could probably come up with better metrics.
https://storage.googleapis.com/deepmind-media/alphago/AlphaG...
https://www.nature.com/nature/journal/v550/n7676/pdf/nature2...
The main reason AlphaGo Zero learns so much faster than its predecessors is because it uses temporal-difference learning.[1] This effectively removes a huge amount of the value network's state space for the learning algorithm to search through, since it bakes in the assumption that a move's value ought to equal that of the best available move in the following board position, which is exactly what you'd expect for a g…
Interestingly, the idea behind temporal difference learning is more or less the intuition behind how people price derivatives in finance. The expected value of a contract at time T, estimated at some time t expected estimate, made n months from now, of the value [12-n] months later, will also be $100 . This allows you to shrink the state space considerably. You can usually work out the payoff of a derivatives in diff…
My memory is hazy so there might not be a real connection here.
Mageek, any reason why they haven't applied this to chess yet?