pierrelux

Uber-Like Models for the Electrical Industry

Posted by pierrelux on Thursday January 19, 2017 @01:19PM
from the orbi.ulg.ac.be dept.

http://orbi.ulg.ac.be/handle/2268/205035

Re:New Draft of “Reinforcement Learning: An Introduction, Second Edition” on Saturday September 24, 2016 @11:45AM (#12571354)

Short and mathematical: "Algorithms for Reinforcement Learning". PDF available: https://sites.ualberta.ca/~szepesva/RLBook.html

Cogitai: new AI startup with investment from Sony

Posted by pierrelux on Tuesday May 17, 2016 @01:26PM
from the engadget.com dept.

http://www.engadget.com/2016/05/17/sony-ai-continual-learning

( Read More... | no comments | 3 points | on HN )

Fluid Control Devices

Posted by pierrelux on Tuesday November 24, 2015 @06:44PM
from the miriam-english.org dept.

http://miriam-english.org/files/fluidics/FluidControlDevices.html

( Read More... | no comments | 2 points | on HN )

Re:Project Panama: Interconnecting JVM and Native Code on Wednesday May 06, 2015 @08:29PM (#9502656)

Could this help Jython to interface with Numpy ? The JyNI project is otherwise constrained to do some pretty serious gymnastics...

Re:To create a super-intelligent machine, start with an equation on Thursday November 28, 2013 @01:26PM (#6815782)

About countably infinite state space: with about 10^56 states, Pacman still remains a very challenging domain ! Just to put things into perspective, my colleague recently ran some experiment where he would exhaust 8Gb of RAM, but could reduce this number down to 4Gb using random projections for dimensionality reduction.

http://www.cs.mcgill.ca/~jpineau/files/whamilton-icml13.pdf

http://www.cs.mcgill.ca/~whamil3/files/cpsr-talk.pdf

Also, don't forget that this is a partially observable domain (with only local sensory information) with no apriori knowledge of the overall goal to achieve. It learns completely from scratch ! If we find this task easy as human beings, it's because we have a lot of prior knowledge that we can transfer into this task