Artificial Intelligence Chris Piech CSBridge 2019 CSBridge 17 A - PowerPoint PPT Presentation

Artificial Intelligence Chris Piech CSBridge 2019 CSBridge ‘17

A Little AI CSBridge ‘17

Something big is happening in the world of AI… CSBridge ‘17

Where is my robot? CSBridge ‘17

Sci-Fi Has Promised Me Robots CSBridge ‘17

House Cleaning Robot CSBridge ‘17

House Cleaning Robot CSBridge ‘17

Robots? Mind Body CSBridge ‘17

Robots? Mind Body CSBridge ‘17

What is AI? CSBridge ‘17

[suspense] CSBridge ‘17

AI: The study and design of intelligent agents CSBridge ‘17

Computer programs AI: The study and design of intelligent agents Better than As well as chance humans CSBridge ‘17

Narrow Intelligence Translate Turkish Play Chess Play Breakout Drive a Car CSBridge ‘17

General Intelligence Translate Turkish Play Chess Play Breakout Drive a Car CSBridge ‘17

Brief History CSBridge ‘17

Early Optimism 1950s 1952 CSBridge ‘17

Early Optimism 1950s “Machines will be capable, within twenty years, of doing any work a man can do.” –Herbert Simon, 1952 CSBridge ‘17

Underwhelming Results 1950s to 1980s The world is too complex CSBridge ‘17

CSBridge ‘17

Big Milestones CSBridge ‘17

Told Speech Was 30 Years Out Almost perfect… CSBridge ‘17

The Last Remaining Board Game CSBridge ‘17

Computers Making Art CSBridge ‘17

Self Driving Cars CSBridge ‘17

What is going on? CSBridge ‘17

[more suspense] CSBridge ‘17

Story of Modern AI CSBridge ‘17

Focus on one problem CSBridge ‘17

Make a Harry Potter Sorting Hat CSBridge ‘17

Classification Logistic Regression is like the Harry Pottery Sorting Hat That is a picture of a one CSBridge ‘17

Classification Logistic Regression is like the Harry Pottery Sorting Hat That is a picture of a zero CSBridge ‘17

Classification That is a picture of an zero * It doesn’t have to be correct all of the time CSBridge ‘17

Can you do it? CSBridge ‘17

CSBridge ‘17

CSBridge ‘17

How about now? What a computer sees 0 0 1 0 1 0 1 0 0 0 1 1 1 0 1 1 0 0 1 0 1 1 1 0 1 0 0 0 0 0 1 1 1 0 1 0 0 1 1 0 0 1 0 1 0 1 1 1 1 1 0 0 0 0 0 1 1 0 1 1 0 0 0 1 1 0 0 1 0 0 0 1 1 1 0 1 0 0 1 1 0 0 0 1 0 1 1 1 1 0 1 1 0 1 1 0 0 1 1 0 1 1 1 0 0 1 0 1 0 0 1 0 0 1 0 0 1 1 1 1 0 0 0 0 1 0 1 0 1 1 0 0 1 1 1 0 1 1 0 0 0 0 0 1 1 1 1 1 1 0 0 0 1 0 1 1 1 0 0 0 1 0 0 0 0 0 1 1 1 0 1 0 0 1 0 0 0 0 0 1 1 1 0 0 0 0 0 0 0 0 1 0 0 1 1 0 0 0 0 0 0 0 0 1 1 1 1 0 0 1 0 0 1 1 1 0 1 0 1 1 0 0 0 1 0 What a human sees CSBridge ‘17

Why is it easy for Humans? About 30% of your cortex is used from vision 3% is used to process hearing CSBridge ‘17

Very hard to Program ?? public class HarryHat extends ConsoleProgram { public void run() { println( “Todo: Write program” ); } } CSBridge ‘17

Perhaps there is an insight? CSBridge ‘17

Two Great Ideas 1. Artificial Neurons 2. Learn by Example CSBridge ‘17

Two Great Ideas 1. Artificial Neurons 2. Learn by Example CSBridge ‘17

1. Artificial Neurons CSBridge ‘17

Neuron CSBridge ‘17

Neuron CSBridge ‘17

Neuron CSBridge ‘17

Neuron CSBridge ‘17

Neuron CSBridge ‘17

Some Inputs are More Important CSBridge ‘17

Artificial Neuron + CSBridge ‘17

Inputs + Numbers between 0 and 1 CSBridge ‘17

Inputs + Numbers between 0 and 1 CSBridge ‘17

Inputs + Numbers between 0 and 1 CSBridge ‘17

Weights + Negative or positive numbers CSBridge ‘17

Weights + CSBridge ‘17

Weights + CSBridge ‘17

Weighted Sum + double weightedSum = 0; weightedSum = weightedSum + input0 * weight0; weightedSum = weightedSum + input1 * weight1; weightedSum = weightedSum + input2 * weight2; weightedSum = weightedSum + input3 * weight3; CSBridge ‘17

Filter and Output + CSBridge ‘17

Java Demo CSBridge ‘17

Biological Basis for Neural Networks • A neuron + • Your brain ??? Actually, it’s probably someone else’s brain CSBridge ‘17

Put Many Together + + + + CSBridge ‘17

Put Many Together Input Neurons Hidden Neurons Output Neurons CSBridge ‘17

Making a Prediction Input Neurons Hidden Neurons Output Neurons … CSBridge ‘17

Making a Prediction Input Neurons Hidden Neurons Output Neurons … CSBridge ‘17

Making a Prediction Input Neurons Hidden Neurons Output Neurons … CSBridge ‘17

Making a Prediction Input Neurons Hidden Neurons Output Neurons I think that is a picture of a one! … CSBridge ‘17

Demonstration http://scs.ryerson.ca/~aharley/vis/conv/ CSBridge ‘17

Great Idea: Artificial Neurons + CSBridge ‘17

Neural Networks get their intelligence from their sliders (parameters) CSBridge ‘17

Two Great Ideas 1. Artificial Neurons 2. Learn by Example CSBridge ‘17

Two Great Ideas 1. Artificial Neurons 2. Learn by Example CSBridge ‘17

2. Learn From Experience CSBridge ‘17

Learn by Example CSBridge ‘17

+ CSBridge ‘17

+ CSBridge ‘17

I think that is a picture of a One ! + What do you mean it’ s actually a Zero ? I’ll adjust my sliders so that I do a better job in the future CSBridge ‘17

I think that is a picture of a One ! + What do you mean it’ s actually a Zero ? I’ll adjust my sliders so that I do a better job in the future CSBridge ‘17

+ CSBridge ‘17

+ CSBridge ‘17

+ CSBridge ‘17

I think that is a picture of a One ! + Wahoo I got it right! CSBridge ‘17

+ CSBridge ‘17

+ CSBridge ‘17

I think that is a picture of a zero ! + What do you mean it’ s actually a one ? I’ll adjust my sliders so that I do a better job in the future CSBridge ‘17

I think that is a picture of a zero ! + What do you mean it’ s actually a one ? I’ll adjust my sliders so that I do a better job in the future CSBridge ‘17

Study Hard! CSBridge ‘17

Visualize the Sliders object models object parts (combination of edges) Training set: Aligned images of faces. edges pixels CSBridge ‘17 [Honglak Lee]

Woah… that’s like a brain… CSBridge ‘17

True. CSBridge ‘17

Decomposition Climb Up Step Up Mountain Mountain Pick Beeper Karel Climb Down Mountain Step Down CSBridge ‘17

Image Net Classification … smoothhound, smoothhound shark, Mustelus mustelus American smooth dogfish, Mustelus canis Florida smoothhound, Mustelus norrisi whitetip shark, reef whitetip shark, Triaenodon obseus Atlantic spiny dogfish, Squalus acanthias Stingray Pacific spiny dogfish, Squalus suckleyi hammerhead, hammerhead shark smooth hammerhead, Sphyrna zygaena smalleye hammerhead, Sphyrna tudes shovelhead, bonnethead, bonnet shark, Sphyrna tiburo angel shark, angelfish, Squatina squatina, monkfish electric ray, crampfish, numbfish, torpedo Mantaray smalltooth sawfish, Pristis pectinatus guitarfish roughtail stingray, Dasyatis centroura butterfly ray eagle ray spotted eagle ray, spotted ray, Aetobatus narinari cownose ray, cow-nosed ray, Rhinoptera bonasus manta, manta ray, devilfish Atlantic manta, Manta birostris devil ray, Mobula hypostoma grey skate, gray skate, Raja batis little skate, Raja erinacea … CSBridge ‘17

0.005% 1.5% ? Random guess Pre Neural Networks GoogLeNet CSBridge ‘17 Le, et al., Building high-level features using large-scale unsupervised learning . ICML 2012

0.005% Pre Neural Networks 43.9% 1.5% Random guess GoogLeNet CSBridge ‘17 Szegedy et al, Going Deeper With Convolutions, CVPR 2015

0.005% Pre Neural Networks 66.3% 1.5% Random guess 2016 CSBridge ‘17 http://image-net.org/challenges/LSVRC/2016/results

CSBridge ‘17

Google Brain CSBridge ‘17

Google Brain 1 Trillion Artificial Neurons CSBridge ‘17

A Neuron That Fires When It Sees Cats Optimal stimulus Top stimuli from the test set by numerical optimization CSBridge ‘17 Le, et al., Building high-level features using large-scale unsupervised learning . ICML 2012

CSBridge ‘17

Other Neurons Neuron 1 Neuron 2 Neuron 3 Neuron 4 Neuron 5 CSBridge ‘17 Le, et al., Building high-level features using large-scale unsupervised learning . ICML 2012