Hot Trends in Smart Software Methods

January 10, 2017

I read “Summary of NIPS 2016.” The write up includes a list of presentations the author found interesting. For me, I noted the three hot trends:

  1. Generative adversarial networks
  2. Reinforcement learning
  3. Bayesian Deep Learning.

You will want to read the documents linked in the original paper.

What struck me about the three hot trends was that none of them were new new. Artificial intelligence has been in the recycling business for decades. The systems and methods are constructed of well known procedures taught in university classes or summarized in books like Peter Norvig’s “Artificial Intelligence: A Modern Approach.”

A good example of this is the tried and true Bayesian method which makes an appearance in the worlds of GANs and RLs. Old chestnut algorithms get new life with plentiful computing and lots of RAM.

If you want to get a sense of what’s in store for smart software, the write up is a useful place to begin.

Stephen E Arnold, January 10, 2016

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta