[zonbron]

Oorspronkelijk geplaatst door Nr.10

Het dominante optimalisatie algoritme in 'deep learning' heet 'gradient descent'.
Gradient descent: in een iteratief proces wordt de kost-functie J geminimaliseerd.
Gradient descent = een fundamenteel algoritme.
In onderstaande grafiek een voorbeeld voor 1 parameter.

Considering a cost function J(.)
with only a single weight coefficient w,
we can illustrate this concept as follows:

BRON
_{14 dec 2015}
Weerom een chinese bron.

https://www.wired.com/beyond-the-bey...-intelligence/

China issues guideline on artificial intelligence development

Updated: Jul 20,2017 7:52 PM english.gov.cn

The State Council has issued a guideline on developing artificial intelligence (AI), setting a goal of becoming a global innovation center in this field by 2030.

The total output value of artificial intelligence industries should surpass 1 trillion yuan ($147.80 billion). A mature theory and technology system should be formed.

Developing AI is a “complicated and systematic project” according to the guideline. An open and coordinated AI innovation system should be constructed to develop not only the technology, but also products and market.

AI in China should be used to promote the country’s technology, economy, social welfare, maintain national security, and contribute to the world.

Terzijde

Camera met AI voor de consument
https://www.engadget.com/2017/05/11/...curity-camera/


AI op maat bestellen
https://www.neurala.com/examples/