Machine Learning at Google Scale ML APIs and TensorFlow Michel - PowerPoint PPT Presentation

Machine Learning at Google Scale ML APIs and TensorFlow

Michel Pereira Google Cloud Customer Engineer @michelpereira@

What is Neural Network and Deep Learning

Neural Network is a function that can learn

How about this?

More hidden layers = More hierarchies of features

How about this?

We need to go deeper neural network From: Convolutional Deep Belief Networks for Scalable Unsupervised Learning of Hierarchical Representations, Honglak Lee et al.

Machine Learning use cases at Google services

Search machine learning for search engines RankBrain: a deep neural network for search ranking #1 #3 signal improvement for Search ranking, to ranking quality out of hundreds in 2+ years 12

Google Photos [glacier] 13 13

Smart reply in Inbox by Gmail 10% of all responses sent on mobile 14

Google Translate with Neural Machine Translation

Deep Learning usage at Google Used across products: Android Apps Gmail Maps Photos Speech Search Translation YouTube and many others ... 2012 2013 2014 2015 16

Externalizing the power with ML APIs

Machine Learning products from Google Easy-to-Use, for non-ML engineers TensorFlow Cloud Machine Learning ML API Customizable, for Data Scientists

Cloud Vision API Image analysis with pre-trained models No Machine Learning skill required REST API: receives an image and returns a JSON $1.50 per 1,000 units GA - cloud.google.com/ vision

Label OCR Faces Detect entities from furniture to Read and extract text, with Faces, facial landmarks, emotions transportation support for > 10 languages Safe Search Landmarks & Image Properties Logos Detect explicit content - adult, Detect landmarks & dominant Identify product logos violent, medical and spoof color of image Google Cloud Platform Confidential & Proprietary 20

Demo 21 21

Cloud Speech API Pre-trained models. No ML skill required REST API: receives audio and returns texts Supports 80+ languages Streaming or non-streaming Public Beta - cloud.google.com/ speech

Features Inappropriate Content Automatic Speech Recognition Global Vocabulary Streaming Recognition Filtering Automatic Speech Recognition (ASR) Recognizes over 80 Returns partial Filter inappropriate powered by deep learning neural languages and variants recognition results content in text results. networking to power your with an extensive immediately, as they applications like voice search or vocabulary. become available. speech transcription. Noisy Audio Handling Integrated API Real-time or Buffered Audio Support Handles noisy audio from many Audio files can be uploaded in the Audio input can be captured by an application’s environments without requiring request and, in future releases, microphone or sent from a pre-recorded audio additional noise cancellation. integrated with Google Cloud file. Multiple audio file formats are supported, Storage. including FLAC, AMR, PCMU and linear-16. Google Cloud Platform Confidential & Proprietary 23

Demo 24 24

Cloud Natural Language API Pre-trained models. No ML skill required REST API: receives text and returns analysis results Supports English, Spanish and Japanese GA - cloud.google.com/ natural-language

Features Syntax Analysis Entity Recognition Extract sentence, identify parts of Identify entities and label by types such speech and create dependency parse as person, organization, location, events, trees for each sentence. products and media. Sentiment Analysis Understand the overall sentiment of a block of text. Google Cloud Platform Confidential & Proprietary 26

Demo 27 27

Cloud Translation API Premium Pre-trained models. No ML skill required REST API: receives text and returns translated text 8 languages : English to Chinese, French, German, Japanese, Korean, Portuguese, Spanish, Turkish Public Beta - cloud.google.com/ translate

Demo 29 29

Cloud Video Intelligence API Video analysis with pre-trained models No Machine Learning skill required REST API: receives a video and returns a JSON Private Beta - cloud.google.com/ video-intelligence

Demo 31 31

TensorFlow: An open source library for Machine Intelligence

What is TensorFlow? Google's open source library for machine intelligence tensorflow.org launched in Nov 2015 Used by many production ML projects

# define the network import tensorflow as tf x = tf.placeholder(tf.float32, [None, 784]) W = tf.Variable(tf.zeros([784, 10])) b = tf.Variable(tf.zeros([10])) y = tf.nn.softmax(tf.matmul(x, W) + b) # define a training step y_ = tf.placeholder(tf.float32, [None, 10]) xent = -tf.reduce_sum(y_*tf.log(y)) step = tf.train.GradientDescentOptimizer(0.01).minimize(xent)

TensorBoard: visualization tool

Portable and Scalable Training on: Mac/Windows GPU server GPU cluster / Cloud Prediction on: Android and iOS RasPi and TPU

Sharing our tools with researchers and developers around the world Released in Nov. 2015 #1 repository for “machine learning” category on GitHub From: http://deliprao.com/archives/168

TensorFlow community and ecosystem

From: https://www.qualcomm.com/news/snapdragon/2017/01/09/tensorflow-machine-learning-now-optimized-snapdragon-835-and-hexagon-682

Google Cloud is The Datacenter as a Computer Enterprise

<OnBoard> //Participe do treinamento sobre os fundamentos de Cloud e conheça as novas tecnologias da nuvem do Google. //Google Cloud OnBoard é para desenvolvedores, programadores e especialistas em TI. Aprimore-se com os melhores instrutores do Google. //Entre em goo.gl/cWXaaE e cadastre-se para receber mais informações. Agenda(2_de_junho) Local(Google_Campus_SP)+6_cidades_BR </OnBoard>

goo.gl/cWXaaE

Thank you!