automation, intelligence and knowledge modelling

Post on 16-Apr-2017

293 Views

Category:

Science

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Automation, intelligence and knowledge modelling

Veselin Pizurica

web11.org September 30th, 2016

Veselin Pizurica CTO, veselin@waylay.io

So you graduated! How will you spend your time?

http://ml-games.tomasz-rewak.com/

Reinforced Learning

People are looking into Biology for inspiration, but things are sometimes a little bit other way around...

Machine vision, NLP, ...

Google: DeepMind to cut data center energy bills

http://www.theverge.com/2016/7/21/12246258/google-deepmind-ai-data-center-cooling

Deep Learning● Choosing the correct feature representation of input data, is a way that people can

bring prior knowledge of a domain to increase an algorithm's computational performance and accuracy.

● To move towards general artificial intelligence, algorithms need to be less dependent on this feature engineering and better learn to identify the explanatory factors of input data on their own.

● Deep learning tries to move in this direction by capturing a 'good' representation of input data by using compositions of non-linear transformations.

But there is a problem...

http://nautil.us/issue/40/learning/is-artificial-intelligence-permanently-inscrutable#disqus_thread

Pneumonia/Asthma story

DARPA: Modern learning algorithms show a tradeoff between human interpretability, or explainability, and their accuracy. Deep learning is both the most accurate and the least interpretable.

What vs. Why?

WHAT IS INTELLIGENCE?

Is he the Smartest Man in the World?

WHAT IS INTELLIGENCE?

WHAT IS INTELLIGENCE?

WHAT IS INTELLIGENCE?

Pick your book

Swarm Intelligence

Why are kids so annoying?

Why is ice slippery?

Why is ice slippery?

Y = f (X) Y = f (X)

Rule engine is a

knowledge modeling problem

Y = f (X)

Modelling problem

http://www.theatlantic.com/science/archive/2016/06/how-consciousness-evolved/485558/Attention Schema Theory (AST)

Decision Tree

Learn rules from dataApply each rule at each nodeClassification is at the leafs of the tree

Let’s do it!

How many distinct decision trees we have with n Boolean attributes?= number of distinct truth tables with 2^n rows = 2^n^nWith 6 Boolean attributes 18,446,744,073,709,551,616

Let’s talk about pancakes!

Bayes Nets

Let’s be Engineer!

How do you express that car needs both battery and fuel to function? Easy.

How do you say that if your lights are not working, most likely it is a battery fault, but it could be as well that just lights are broken? Still the fact that lights are not working point to most likely cause of the battery fault.

If you only model via composition and add behavior separately – what most of the tools do these days – you are heading for complexity!

But there is a problem...

http://web.mit.edu/cocosci/Papers/Science-2015-Lake-1332-8.pdf

Glenn Roberts and Motorcycle Mojo Magazine

Pick them all!

top related