EFTA02508316.pdf

From: jeffrey E. <[email protected]> Sent: Friday, March 9, 2018 10:40 AM To: Joscha Bach Subject: Re: I would think of it more of a space / field effects = Not recursive algorithm s > wrote: Last week I got to know Steve Hyman, Daniel Kahneman and Bo= Horvitz. Telefonica invited all of us to a two day workshop with Pablo Ro=riguez, Ken Morse and a few others, where we were meant to advise them on =ow to use Al for health applications. I told them that I think the goal of=therapeutic invention is not to increase happiness, but integrity. Happine=s is merely an indicator, not the benchmark. Current apps tend to subvert =he motivation of people, but I don't think that this is necessary or t=e best strategy. Humans are meant to be programmable, not subverted. They =erceive their programming as "higher purpose". If we can come fr=m the top, supporting purpose, instead of from the bottom, subverting atte=tion, we might be more successful. (Downside might be that we create cults=) Of the bunch, Hyman managed to be the most interesting (Kahneman was very c=arismatic but mostly tried to see if he could identify an application for =is system one/system two theory). Gary Marcus was there, too, but annoyed =veryone by being too insecure to deal with his incompetence. Did I tell you that I discovered that Deep Learning might be best understoo= as Second order Al? First order Al was the classical Al that was started by Marvin Minsky in th= 1950ies, and it worked by figuring out how we (or an abstract system) can=perform a task that requires intelligence, and then implementing that algo=ithm directly. It yielded most of the progress we saw until recently: ches= programs, data bases, language parsers etc. Second order Al does not implement the functionality directly, but we write=the algorithms that figure out the functionality by themselves. Second ord=r Al is automated function approximation. Learning has existed for a long =ime in Al of course, but Deep Learning means compositional function approx=mation. Our current approximator paradigm is mostly the neural network, i.e. chaine= normalized weighted sums of real values that we adapt by changing the wei=hts with stochastic gradient descent, using the chain rule. This works wel= for linear algebra and the fat end of compact polynomials, but it does no= work well for conditional loops, recursion and many other constructs that=we might want to learn. Ultimately, we want to learn any kind of algorithm=that runs efficiently on the available hardware. Neural network learning is very slow. The different learning algorithms are=quite similar in the amount of structure they can squeeze out of the same =raining data, but they need far more passes over the data than our nervous=system. The solution might be meta learning: we write algorithms that learn how to =reate learning algorithms. Evolution is meta learning. Meta learning is go=ng to be third order Al and perhaps trigger a similar wave as deep learnin=. I intend to visit NYC for a workshop at NYU on the weekend of the 16th. We just moved into a new apartment; the previous one had only two bedrooms =nd this one has three, so I can have a study. It seems that we are as luck= with the new landlords as with the previous ones. Bests, and thank you for everything! I EFTA_R1_01639307 EFTA02508316  Joscha > On Mar 8, 2018, at 16:37, jeffrey E. <[email protected] <mailto:jeevacation=gmail.com» wrote: > progress? > -- > please note > The information contained in this communication is > confidential, may be attorney-client privileged, may > constitute inside information, and is intended only for > the use of the addressee. It is the property of > JEE > Unauthorized use, disclosure or copying of this > communication or any part thereof is strictly prohibited > and may be unlawful. If you have received this > communication in error, please notify us immediately by > return e-mail or by e-mail to [email protected] <mailto:[email protected]> , and > destroy this communication and all copies thereof, > including all attachments. copyright -all rights reserved =AO please note The information contained in this=communication is confidential, may be attorney-client privileged, mayconstitute inside information, and is intended only for the use of the=addressee. It is the property of JEE Unauthorized use, disclosure or=copying of this communication or any part thereof is strictly prohibite= and may be unlawful. If you have received this communication in err=r, please notify us immediately by return e-mail or by e-mail to [email protected]<=a>, and destroy this communication and all copies thereof, including=all attachments. copyright -all rights reserved --001a114bbafaba37e30566f869bd-- conversation-id 12974 date-last-viewed 0 date-received 1520591995 flags 8590195713 gmail-label-ids 7 6 remote-id 803133 2 EFTA_R1_01639308 EFTA02508317