customer engineer brian vinson machine learning group ... · machine learning engine tensorflow...

Post on 20-May-2020

14 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Machine Learning using Google Technology Machine Learning Group | November 7th, 2017

Brian VinsonCustomer Engineer

Data is EverythingHow well you use your data determines the degree of your success

© 2017 Google Inc. All rights reserved.

How does a machine learn?

How does a machine learn?

PREDICTING HOUSE SALE PRICES

Sq. Footage Price

1,000 $100,000

3,000 $300,000

How much would a 2,000 sq ft. house sell for?

How does a machine learn?

PREDICTING HOUSE SALE PRICES

Sq. Footage Price

1,000 $100,000

3,000 $300,000

2,000 $200,000

1Look at the Data

Sq. FootagePr

ice

1,000 2,000 3,000

100,000

200,000

300,000

2Find a pattern

Sq. FootagePr

ice

1,000 2,000 3,000

100,000

200,000

300,000

3Predict

Sq. FootagePr

ice

1,000 2,000 3,000

100,000

200,000

300,000

A better dataset

PREDICTING HOUSE SALE PRICES

Sq. Footage

Crime Rate (1-100)

School Rating (1-5)

Price

1,000 1 5 $300,000

2,000 50 3 $200,000

4,000 80 1 $100,000

How much would a 3,000 sq ft. house with a crime rate of 30 and a school rating of 2 cost?

It's not who has the best algorithm who wins, it's who has the most data.

— Andrew Ng , Co-Founder of Google Brain

“”

Google’s Mission

Eight cloud products with

ONE BILLIONusers each

© 2017 Google Inc. All rights reserved.

San Francisco

New York

14

Gmail—Smart Reply Inbox

Proprietary + Confidential

10%of all responses sent on mobile

April 1, 2009: April Fool’s Day joke

Nov 5, 2015: Launched Real Product

Proprietary + Confidential

SearchSearch RankingSpeech Recognition

AndroidKeyboard and Speech Input

PlayApp Recommendations Game Developer Experience

GmailSmart ReplySpam Classification

DriveIntelligence in Apps

ChromeSearch by Image

PhotosPhotos Search

YouTubeVideo Recommendations Better Thumbnails

MapsStreet View ImageParsing Local Search

TranslateText, Graphic and Speech Translations

CardboardSmart Stitching

AdsRicher Text AdsAutomated Bidding

Self Driving Car1.5MM miles driven

Data Center Power UsageReduced cooling energy 40%

Alpha GoFirst AI to beat a world Go champion (2016)

Google is the world leader in applying Machine Learning to real-world situations, inside and outside of Google.

Rule #1 of Machine Learning: Don’t build a model if you don’t have to

Custom ML models Pre-trained ML models

Machine Learning Engine

TensorFlow

Vision API

Translation API

Natural Language API

Speech API Jobs API

Two Flavors of Machine Learning

Video Intelligence API

● State of the art machine learning models built by Google experts

● Pre-Trained on massive datasets using Google scale infrastructure

● Work out of the box. No machine learning expertise required

Machine Learning APIs

Vision API

Detect broad sets of categories within an image, ranging from modes of transportation to animals.

Analyze facial features to detect emotions: joy, sorrow, anger.

Detect logos.

Detect and extract text within an image, with support for a broad range of languages, along with support for automatic language identification.

Extract text

Detect different types of inappropriate content from adult to violent content. Powered by Google Safe Search

Detect inappropriate contentObject Recognition Facial sentiment & logos

TRY THE API

Speech API

Powered by deep learning neural networking to power your applications..

No need for signal processing or noise cancellation before calling API. Can handle noisy audio from a variety of environments.

Noise Robustness

Can provide context hints for improved accuracy. Especially useful for device and app use cases.

Word HintsSpeech Recognition

TRY THE API

Recognizes over 80 languages & variants. Can also filter inappropriate content in text results

Over 80 languages

Can stream text results, returning partial recognition results as they become available. Can also be run on buffered or archived audio files.

Real-time results

Translation API

Supports more than 100 languages and thousands of language pairs.

Behind the scenes, Translation API is learning from logs analysis and human translation examples. Existing language pairs improve and new language pairs come online at no additional cost.

Sometimes you don’t know your source text language in advance. Can automatically identify languages with high accuracy.

Automatic language detection

The Premium edition is tailored for users who need precise, long-form translation services (e.g. livestream translations, high volume of emails, detailed articles and documents)

Premium edition BETAText Translation Continuous Updates

TRY THE API

Natural Language API

Identify entities and label by types such as person, organization, location, events, products and media.

Enables you to easily analyze text in multiple languages including English, Spanish and Japanese.

Extract tokens and sentences, identify parts of speech (PoS) and create dependency parse trees for each sentence.

Syntax analysisEntity Recognition Multi-Language Support

TRY THE API

Understand the overall sentiment expressed in a block of text.

Sentiment Analysis

Video Intelligence API

Detect entities within the video, such as "dog", "flower" or "car".

You can now search your video catalog the same way you search text documents..

Extract actionable insights from video files without requiring any machine learning or computer vision knowledge.

Enable Video Search

More features will be added to the Video Intelligence API during the BETA period.

More to come ... BETALabel Detection Insights From Videos

Cloud Detection in Satellite Images

11%of images

Error rate

3%of images

Development time

3months

YearsPreviously

Today

Data Loss Prevention (DLP) API

DLP lets GCP users scan for and redact sensitive data

● Find sensitive data stored in GCP● Redact sensitive data to comply with

regulations or policies● Safely unlock more of the cloud

Automatically discover and redact sensitive data everywhere

RawData

Machine Learning

Secure Sharing

App DevelopmentRedact Data

DLP API

Classify Data

Montoring

Data Governance

TensorFlowWorld’s Most Popular Machine Learning Framework

TensorFlow

● Tensorflow is what we use for our own internal machine learning projects, and now it’s available to you!

● Powers over 100 Google services

● Tutorials to get started at https://www.tensorflow.org

As of 6-24-2017

Most Popular Machine Learning Framework

FrameworkG

itHub

Sta

rsTensorFlow

10,000

20,000

30,000

0

40,000

SKLearn Caffe CNTK MXnet Torch

50,000

60,000

Cloud TPU

● 180 TFlops, 9x faster than NVIDIA P100 GPU

● Faster time to market for new applications

● Train, iterate, and predict faster

● Reduce the cost of Machine Learning work

DEMO!!

Brian VinsonGoogle Cloud Customer Engineer

brianvinson@google.com

Thank You!

top related