Artificial Intelligence is just whatever a computer cant

“Artificial Intelligence is just whatever a computer can’t do yet…”

“Artificial Intelligence is computers doing things that we would normally think of as intelligent in humans. ”

Fact Fact Model Fact Doctor/ Lawyer/ Astronaut/ President

Model Sensory Input Behavior NEURAL NETWORK

Model Data Answers

Algorithms Diseased?

O T E V A H T ’ K N R O O D W U O O Y T H T A ! S M P L W E O H N T I K T U B … E R E H

Predict how much/many Which category? (Regression) Data structure? (Clustering, Recommender) (Classification) Is it weird? (Anomaly) What next? (Reinforcement Learning)

AI Computers doing things that we would normally think of as intelligent in humans MACHINE LEARNING NEURAL NETWORKS

1. NEW COMBO OF MATH the spark 2. BIG DATA the fuel 3. MASSIVE COMPUTATION the horsepower

1. NEW COMBO OF MATH the spark 2. BIG DATA the fuel 3. MASSIVE COMPUTATION the horsepower

1. NEW COMBO OF MATH the spark 2. BIG DATA the fuel 3. MASSIVE COMPUTATION the horsepower

SUMMER 2012

GOOGLE BRAIN 16, 000 CPUS SUMMER 2012

< Geoffrey Hinton, Alex Krizhevsky, & Ilya Sutskever DEEP CNNs on 2 GPUs SUMMER 2012 WINTER 2012

“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run. ” Ray Amara

The way we will make software is not the way we have made software.

Anomaly detection Principal component analysis Python Probability theory Cross-validation Logistic regression Reinforcement learning. Scala Multi-classification K-means clustering R False negatives feature engineering Activation functions Accuracy Training set Normalization Algorithms Multivariate calculus Sentinel values Test set Differential equations Perceptron Stochastic gradient descent Recommender systems Support vector machines Statistics Linear algebra Information theory Bias-variance tradeoff Unit step function Recall False positives Precision Spark clusters Deep belief networks Sigmoid function Complex optimizations Generative adversarial neural networks Sensitivity GPUs Recurrent neural networks Convolutional neural networks Deep belief networks

AI Computers doing things that we would normally think of as intelligent in humans

Microsoft. ML Pick a common ML tool and sample data Quickstart guide See some results – how well did it go Graph data Get curious – ask a question Disease or not disease?

Wednesday 2: 00 PM B 8010 Bot Capabilities, Patterns and Principles Wednesday 3: 30 PM B 8092 Using Microsoft Speech APIs to bring the power of speech recognition to your apps Wednesday 5: 00 PM B 8091 Build intelligence into your Business Apps with ease using Bing APIs in Cognitive Services Computer Vision made easy: from pre-trained models to custom ones, Cognitive Services has you covered. Thursday 10: 30 AM B 8020 Thursday 1: 00 PM B 8903 Open Q&A: Artificial intelligence tools for developers Thursday 4: 00 PM B 8090 How Language APIs can help your apps communicate with people Friday 12: 30 PM B 8089 Let Knowledge APIs utilize the power of the web to improve your apps Friday 2: 00 PM B 8038 Deep learning with Cognitive Toolkit