machine learning and artificial intelligence 2 major frameworks uapache mxnet uproduct...

4
11/21/19 1 Machine Learning and Artificial Intelligence Powered by AWS 1 Machine Learning used by Amazon u Product Recommendations u Robotics Path Optimization u Alexa voice u Computer vision technology by Prime Air 2 Machine Learning Offered by AWS u Frameworks and Infrastructure u Machine Learning Platforms u API-Driven services u Models u Optimized instances with GPUs and Machine Images 3

Upload: others

Post on 20-May-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Machine Learning and Artificial Intelligence 2 Major Frameworks uApache MXNet uProduct recommendations engines uCaffe and Caffe 2 uComputer vision projects uTensorflow 4 Machine Learning

11/21/19

1

Machine Learning and Artificial Intelligence

Powered by AWS

1

Machine Learning used by Amazon

u Product Recommendations

u Robotics Path Optimization

u Alexa voice

u Computer vision technology by Prime Air

2

Machine Learning Offered by AWS

u Frameworks and Infrastructure

u Machine Learning Platforms

u API-Driven services

u Models

u Optimized instances with GPUs and Machine Images

3

Page 2: Machine Learning and Artificial Intelligence 2 Major Frameworks uApache MXNet uProduct recommendations engines uCaffe and Caffe 2 uComputer vision projects uTensorflow 4 Machine Learning

11/21/19

2

Major Frameworks

u Apache MXNetu Product recommendations engines

u Caffe and Caffe 2u Computer vision projects

u Tensorflow

4

Machine Learning Platforms

u Apache Spark on Amazon Elastic Map Reduceu Process large amounts of data

u Machine Learningu Real-Time Streamingu Interactive Analytics

u SparkMLu Machine learning library using Spark

u In-memory data

5

API based tools

u Amazon Rekognitionu Facial and object recognition

u Amazon Pollyu Turn text into speech over voices / languages

u Amazon Lexu Intelligent conversational interfacesu Same technology used in Alexa

6

Page 3: Machine Learning and Artificial Intelligence 2 Major Frameworks uApache MXNet uProduct recommendations engines uCaffe and Caffe 2 uComputer vision projects uTensorflow 4 Machine Learning

11/21/19

3

AWS Pricing

u Standard instances: 0.05 – 6.45 per houru Computer optimized – 0.27 – 4.83 per hour

u Accelerated Computing – 1.26 – 34.272 per hour

u HardwareInstance Type vCPU GPU Mem GPU Mem Network Performance

Ml.t2.medium 2 - 4 - Low to moderateMl.r5.large 2 - 16 - Up to 10 Gbps

Ml.g4dn.16xlarge 64 1xT4 256 16 50 Gbps

7

Google Cloud Offerings

u Cloud translation API: $76 per hour

u Cloud speech to text: $0.04 per minute

u Cloud Natural Language API: $5.00 per 1k classification

u Cloud Vision API

u Computer Engine API

u Dataflow API

8

Demonstrations

u GCP Image Detectionuhttps://www.qwiklabs.com/focuses/1823?parent=catalog

9

Page 4: Machine Learning and Artificial Intelligence 2 Major Frameworks uApache MXNet uProduct recommendations engines uCaffe and Caffe 2 uComputer vision projects uTensorflow 4 Machine Learning

11/21/19

4

Creating an Object Detection Application Using TensorFlow

u Create a virtual machine (VM) using Google Compute Engine.

u Install the Object Detection API library.

u Install and launch an object detection web application.

u Test the web application with uploaded images.

10

Demonstrations

u GCP Online Predictionu https://www.qwiklabs.com/focuses/1789?parent=catalog

11

Scikit-learn Model Serving with Online Prediction Using AI Platform

u Create and save a m odel to a file

u Upload the saved m odel to Google C loud Storage

u Create a m odel resource in AI Platform

u Create a m odel version (linking your scikit-learn m odel)

u M ake an online prediction

12