Tweet
Tweet

Slide 1

Slide 1 text

Machine Learning For Developers Danilo Poccia @danilop danilop AWS Technical Evangelist

Slide 2

Slide 2 text

Credit: Gerry Cranham/Fox Photos/Getty Images http://www.telegraph.co.uk/travel/destinations/europe/united-kingdom/england/london/galleries/The-history-of-the-Tube-in-pictures-150-years-of-London-Underground/1939-ticket-examin/

Slide 3

Slide 3 text

1939 London Underground Credit: Gerry Cranham/Fox Photos/Getty Images http://www.telegraph.co.uk/travel/destinations/europe/united-kingdom/england/london/galleries/The-history-of-the-Tube-in-pictures-150-years-of-London-Underground/1939-ticket-examin/

Slide 4

Slide 4 text

Data Predictions

Slide 5

Slide 5 text

Data Model Predictions

Slide 6

Slide 6 text

Model

Slide 7

Slide 7 text

http://www.thehudsonvalley.com/articles/60-years-ago-today-local-technology-demonstrated-artificial-intelligence-for-the-first-time 1959 Arthur Samuel

Slide 8

Slide 8 text

Machine Learning

Slide 9

Slide 9 text

Machine Learning Supervised Learning Inferring a model from labeled training data

Slide 10

Slide 10 text

Machine Learning Supervised Learning Unsupervised Learning Inferring a model from labeled training data Inferring a model to describe hidden structure from unlabeled data

Slide 11

Slide 11 text

Reinforcement Learning Perform a certain goal in a dynamic environment Machine Learning Supervised Learning Unsupervised Learning

Slide 12

Slide 12 text

Driving a vehicle Playing a game against an opponent

Slide 13

Slide 13 text

Clustering

Slide 14

Slide 14 text

Clustering

Slide 15

Slide 15 text

Tip: Try topic modeling with your own emails ;-) Topic Modeling Discovering abstract “topics” that occur in a collection of documents For example, looking for “infrequent” words that are used more often in a document

Slide 16

Slide 16 text

Regression “How many bikes will be rented tomorrow?” Happy, Sad, Angry, Confused, Disgusted, Surprised, Calm, Unknown Binary Classification Multi-Class Classification “Is this email spam?” “What is the sentiment of this tweet, or of this social media comment?” 1, 0, 100K Yes / No True / False %

Slide 17

Slide 17 text

Training the Model Minimizing the Error of using the Model on the Labeled Data

Slide 18

Slide 18 text

Validation How well is this Model working on New Data?

Slide 19

Slide 19 text

Be Careful of Overfitting

Slide 20

Slide 20 text

Be Careful of Overfitting

Slide 21

Slide 21 text

Be Careful of Overfitting

Slide 22

Slide 22 text

Better Fitting

Slide 23

Slide 23 text

Better Fitting

Slide 24

Slide 24 text

Different Models ⇒ Different Predictions

Slide 25

Slide 25 text

Labeled Data

Slide 26

Slide 26 text

Labeled Data 70% 30% Training Validation

Slide 27

Slide 27 text

Neural Networks

Slide 28

Slide 28 text

1943 Warren McCulloch, Walter Pitts Threshold Logic Units

Slide 29

Slide 29 text

1962 Frank Rosenblatt Perceptron

Slide 30

Slide 30 text

∑ w1 w2 w3 wn w0 = output weights (parameters) activation function input

Slide 31

Slide 31 text

f(∑) w1 w2 w3 wn w0 = weights (parameters) activation function output input

Slide 32

Slide 32 text

f(∑) input output

Slide 33

Slide 33 text

1969 Marvin Minsky, Seymour Papert Perceptrons: An Introduction to Computational Geometry A perceptron can only solve linearly separable functions (e.g. no XOR)

Slide 34

Slide 34 text

f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) input layer hidden layer output layer input output Multiple Layers Lots of Parameters Backpropagation

Slide 35

Slide 35 text

Microprocessor Transistor Counts 1971-2011 Intel E7 CPU 4-24 cores NVIDIA K80 GPU 2,496 cores https://en.wikipedia.org/wiki/Moore's_law

Slide 36

Slide 36 text

LeCun, Gradient-Based Learning Applied to Document Recognition,1998 Hinton, A Fast Learning Algorithm for Deep Belief Nets, 2006 Bengio, Learning Deep Architectures for AI, 2009 Advances in Research 1998-2009

Slide 37

Slide 37 text

Image Processing

Slide 38

Slide 38 text

f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) f(∑) output How to give images in input to a Neural Network? Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 39

Slide 39 text

Convolution Matrix 0 0 0 0 1 0 0 0 0 Identity Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 40

Slide 40 text

Convolution Matrix 1 0 -1 2 0 -2 1 0 -1 Left Edges Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 41

Slide 41 text

Convolution Matrix -1 0 1 -2 0 2 -1 0 1 Right Edges Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 42

Slide 42 text

Convolution Matrix 1 2 1 0 0 0 -1 -2 -1 Top Edges Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 43

Slide 43 text

Convolution Matrix -1 -2 -1 0 0 0 1 2 1 Bottom Edges Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 44

Slide 44 text

Convolution Matrix 0.6 -0.6 1.2 -1.4 1.2 -1.6 0.8 -1.4 1.6 Random Values Photo by David Iliff. License: CC-BY-SA 3.0 https://commons.wikimedia.org/wiki/File:Colosseum_in_Rome,_Italy_-_April_2007.jpg

Slide 45

Slide 45 text

Convolutional Neural Networks (CNNs) https://en.wikipedia.org/wiki/Convolutional_neural_network

Slide 46

Slide 46 text

ImageNet Classification Error Over Time 0 5 10 15 20 25 30 2010 2011 2012 2013 2014 2015 2016 Classification Error CNNs

Slide 47

Slide 47 text

2012 ImageNet Classification with Deep Convolutional Neural Networks

Slide 48

Slide 48 text

SuperVision: 8 layers, 60M parameters 0

Slide 49

Slide 49 text

2013 Visualizing and Understanding Convolutional Networks

Slide 50

Slide 50 text

No content

Slide 51

Slide 51 text

No content

Slide 52

Slide 52 text

http://www.asimovinstitute.org/neural-network-zoo/ Lots of Parameters Network Architectures defined by Hyperparameters Dropout Layers for Regularization

Slide 53

Slide 53 text

Generative Adversarial Networks (GANs) Generator Neural Network Discriminator Neural Network Real or Generated? Real Picture Generated Picture

Slide 54

Slide 54 text

2016 Generative Adversarial Networks (GANs)

Slide 55

Slide 55 text

Artificial Intelligence & Deep Learning At Amazon Thousands Of Employees Across The Company Focused on AI Discovery & Search Fulfilment & Logistics Add ML-powered features to existing products Echo & Alexa

Slide 56

Slide 56 text

Artificial Intelligence on AWS P2, F1 & Elastic GPUs Deep Learning AMI and template Investment in Apache MXNet

Slide 57

Slide 57 text

Apache MXNet

Slide 58

Slide 58 text

Deep Learning Frameworks MXNet, Caffe, Tensorflow, Theano, Torch, CNTK and Keras Pre-installed components to speed productivity, such as Nvidia drivers, CUDA, cuDNN, Intel MKL-DNN with MXNet, Anaconda, Python 2 and 3 AWS Integration Deep Learning AMI

Slide 59

Slide 59 text

Amazon AI Bringing Powerful Artificial Intelligence To All Developers

Slide 60

Slide 60 text

Amazon Rekognition Image Recognition And Analysis Powered By Deep Learning 1

Slide 61

Slide 61 text

Amazon Rekognition Deep learning-based image recognition service Search, verify, and organize millions of images Object and Scene Detection Facial Analysis Face Comparison Facial Recognition

Slide 62

Slide 62 text

Amazon Rekognition: Images In, Categories and Facial Analysis Out Amazon Rekognition Car Outside Daytime Driving Objects & Scenes Female Smiling Sunglasses Face ID DetectLabels DetectFaces CompareFaces IndexFaces SearchFacesByImage Faces

Slide 63

Slide 63 text

No content

Slide 64

Slide 64 text

Deep Learning Process Conv 1 Conv 2 Conv n … … Feature Maps Labrador Dog Beach Outdoors Softmax Probability Fully Connected Layer

Slide 65

Slide 65 text

Bynder allows you to easily create, find and use content for branding automation and marketing solutions. With our new AI capabilities, Bynder’s software… now allows users to save hours of admin labor when uploading and organizing their files, adding exponentially more value. Chris Hall CEO, Bynder ” “ With Rekognition, Bynder revolutionizes marketing admin tasks with AI capabilities

Slide 66

Slide 66 text

Amazon Polly Text To Speech Powered By Deep Learning 2

Slide 67

Slide 67 text

Amazon Polly: Text In, Life-like Speech Out Amazon Polly “The temperature in WA is 75°F” “The temperature in Washington is 75 degrees Fahrenheit”

Slide 68

Slide 68 text

TEXT Market grew by > 20%. WORDS PHONEMES { { { { { ˈtwɛn.ti pɚ.ˈsɛnt ˈmɑɹ.kət ˈgɹu baɪ ˈmoʊɹ ˈðæn PROSODY CONTOUR UNIT SELECTION AND ADAPTATION TEXT PROCESSING PROSODY MODIFICATION STREAMING Market grew by more than twenty percent Speech units inventory

Slide 69

Slide 69 text

aws polly synthesize-speech --text "It was nice to live such a wonderful live show." --output-format mp3 --voice-id Joanna --text-type text output.mp3

Slide 70

Slide 70 text

“Nel mezzo del cammin di nostra vita mi ritrovai per una selva oscura ché la diritta via era smarrita.” https://commons.wikimedia.org/wiki/File:Portrait_de_Dante.jpg

Slide 71

Slide 71 text

Duolingo voices its language learning service Using Polly Duolingo is a free language learning service where users help translate the web and rate translations. With Amazon Polly our users benefit from the most lifelike Text-to-Speech voices available on the market. Severin Hacker CTO, Duolingo ” “ • Spoken language crucial for language learning • Accurate pronunciation matters • Faster iteration thanks to TTS • As good as natural human speech

Slide 72

Slide 72 text

GoAnimate is a cloud-based, animated video creation plarform. Amazon Polly gives GoAnimate users the ability to immediately give voice to the characters they animate using our platform. Alvin Hung CEO, GoAnimate ” “ • Multi-language communication • Training or HR professionals who have to create content in many languages • Video preproduction • Video makers who need to iterate and fine-tune before the text-to- speech is eventually replaced by a professional voiceover • K–12 education • Students who make videos and don’t have access to professional voices or time for or knowledge of voiceover With Polly, GoAnimate gives voice to the characters in their animations

Slide 73

Slide 73 text

” “ Royal National Institute of Blind People creates and distributes accessible information in the form of synthesized content Amazon Polly delivers incredibly lifelike voices which captivate and engage our readers. John Worsfold Solutions Implementation Manager, RNIB • RNIB delivers largest library of audiobooks in the UK for nearly 2 million people with sight loss • Naturalness of generated speech is critical to captivate and engage readers • No restrictions on speech redistributions enables RNIB to create and distribute accessible information in a form of synthesized content RNIB provides the largest library in the UK for people with sight loss

Slide 74

Slide 74 text

Amazon ALEXA (It’s what’s inside Alexa) 3 Natural Language Understanding (NLU) & Automatic Speech Recognition (ASR) Powered By Deep Learning

Slide 75

Slide 75 text

Amazon Lex: Speech Recognition & Natural Language Understanding Amazon Lex Automatic Speech Recognition Natural Language Understanding “What’s the weather forecast?” Weather Forecast

Slide 76

Slide 76 text

Amazon Lex: Speech Recognition & Natural Language Understanding Amazon Lex Automatic Speech Recognition Natural Language Understanding “What’s the weather forecast?” “It will be sunny and 25°C” Weather Forecast

Slide 77

Slide 77 text

Lex Bot Structure Utterances Spoken or typed phrases that invoke your intent BookHotel Intents An Intent performs an action in response to natural language user input Slots Slots are input data required to fulfill the intent Fulfillment Fulfillment mechanism for your intent

Slide 78

Slide 78 text

Hotel Booking City New York City Check In Nov 30th Check Out Dec 2nd Hotel Booking City New York City Check In Check Out “Book a Hotel” Book Hotel NYC “Book a Hotel in NYC” Automatic Speech Recognition Hotel Booking New York City Natural Language Understanding Intent/Slot Model Utterances “Your hotel is booked for Nov 30th” Polly Confirmation: “Your hotel is booked for Nov 30th” a in “Can I go ahead with the booking?”

Slide 79

Slide 79 text

” “ Finding missing persons: ~100,000 active missing persons cases in the U.S. at any given time ~60% are adults, ~40% are children • Motorola Solutions applies Amazon Rekognition, Amazon Polly and Amazon Lex • Image analytics and facial recognition can continually monitor for missing persons • Tools that understand natural language can enable officers to keep eyes up and hands free Motorola Solutions is using Amazon AI to help finding missing persons Motorola Solutions keeps utility workers connected and visible to each other with real-time voice and data communication across the smart grid.

Slide 80

Slide 80 text

I See

Slide 81

Slide 81 text

I see… Amazon Rekognition Amazon Polly Camera Raspberry Pi Voice Synthesize Speech Detect Labels Detect Faces

Slide 82

Slide 82 text

No content

Slide 83

Slide 83 text

Nikola Tesla, 1926 “When wireless is perfectly applied, the whole earth will be converted into a huge brain…”

Slide 84

Slide 84 text

Machine Learning For Developers Danilo Poccia @danilop danilop AWS Technical Evangelist