News for hackers. Stuff that matters.

A comment on For AI to thrive, it must explain itself. Parent comment

Re:For AI to thrive, it must explain itself by rdlecler1 on Sunday February 18, 2018 @02:13PM (#16407637)
We only know which ones are spurious if we run the trimming algorithms. I wrote a paper about 10 years ago showing how to use generic algorithms on the weight matrix to remove spurious interactions.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2538912/

[ Parent | on HN ]
Re:For AI to thrive, it must explain itself by kazinator on Sunday February 18, 2018 @10:30PM (#16410561)
Do we know which ones are absolutely spurious? Or do we just know which ones are spurious with respect to a particular result that we got?

Like for this image of Grandma, these cells are recognizing it, so the others are spurious. Oops, for this other image of Grandma, a different subset are recognizing it.

If, say, the same 90% of the structure is not involved in recognizing any instances, then why wouldn't we have made the whole thing 10X smaller in the first place and trained that much smaller network? Or why don't we make it 10X smaller now? Trim away the dead weight and make it lighter and faster.

[ Parent | on HN ]
Re:For AI to thrive, it must explain itself by rdlecler1 on Tuesday February 20, 2018 @10:32AM (#16421060)
That’s what the genetic algorithm is for. You essentially add a entropy function that trims connections. If the performance is neutral or improves then the leaner network topology is subject to positive selection in a population.
[ Parent | on HN ]
Re:For AI to thrive, it must explain itself by DonaldFisk on Sunday February 18, 2018 @08:07PM (#16410028)
Genetic algorithms?
[ Parent | on HN ]
Re:For AI to thrive, it must explain itself by rdlecler1 on Sunday February 18, 2018 @08:30PM (#16410117)
Yes, sorry. Typo.
[ Parent | on HN ]