Note:
The Content was modified from the Microsoft Content team.
Deck Owner: Nitah Onsongo
Tech/Msg Review: Cesar De La Torre, Simon Tao, Clarke Rahrig
---
Event: Insider Dev Tour Berlin
Event Description: Microsoft is going on a world tour with the announcements of Build 2019. The Insider Dev Tour focuses on innovations related to Microsoft 365 from a developer's perspective.
Date: June 7th, 2019
Event link: https://www.microsoft.com/de-de/techwiese/news/best-of-build-insider-dev-tour-am-7-juni-in-berlin.aspx
Linkedin: http://linkedin.com/in/mia-chang/
4. Examples
Classify if a customer is a retention risk
Identify objects in images and videos
Detect defects in an industrial process
Some problems are difficult to solve using traditional algorithms and
procedural programming.
These examples are good candidates for machine learning.
5. When can you use Machine Learning?
A repetitive decision
or process
Solution lacks an
explicit definition
A lot of training data is
available
6. #insiderDevTour
90%
Reduced time from
documents to insights
$300K
Cost savings per day
40x
More insurance cases reviewed
200K
Customers interacting with
chatbots
+20%
Increase in
customer satisfaction
+32%
Increase in sales
8. #insiderDevTour
Prepare Your Data
Quickly launch and scale
Spark on demand
Rich interactive workspace
and notebooks
Seamless integration with all
Azure data services
Build and Train
Broad frameworks and tools
support
TensorFlow, Cognitive Toolkit,
Caffe2, Keras, MxNET, PyTorch
Deploy
Docker containers
Windows AI Platform
Azure Machine Learning
Machine Learning Landscape
9. #insiderDevTour
Machine Learning for Developers
Enable Multi-Input
Azure Cognitive Services
Ink Recognizer
Increase Functionality
Windows AI platform
Sentiment Analysis
ML.NET
Price prediction
10. #insiderDevTour
Machine Learning for Developers
Enable Multi-Input
Azure Cognitive Services
Ink Recognizer
Increase Functionality
Windows AI platform
Sentiment Analysis
ML.NET
Price prediction
12. Azure Cognitive Services
The most comprehensive pre-trained AI
Language
Vision
Speech
Decision
Web search
Bing Spell Check
Custom
Vision
Personalizer
Form Recognizer
Neural Text-to-Speech
Anomaly Detector Content
Moderator
Content Moderator
Custom Speech
Speech transcription
Text-to-Speech
Conversation transcription
capability
Face
Video
Indexer
Ink Recognizer
Computer
Vision Language
Understanding
QnA Maker
Text Analytics
Translator Text
Bing Web
Search
Bing Custom
Search
Bing
Video Search
Bing Image Search
Bing
Local Business
Search
Bing Visual Search
Bing Entity Search
Bing News
Search
Bing Autosuggest
19. #insiderDevTour
ML.NET
Machine Learning framework for building custom ML Models
Proven at scale
Azure, Office, Windows
Extensible
TensorFlow, ONNX and Infer.NET
Cross-platform and open-source
Runs everywhere
Easy to use tools
CLI + UI-based tool for building models
21. #insiderDevTour
How much is the taxi fare for 1 passenger going from Cape Town to Johannesburg?
AutoML with ML.NET
ML.NET CLI global tool accelerates productivity
22. #insiderDevTour
Criterion
Loss
Min Samples Split
Min Samples Leaf
XYZ
Parameter 1
Parameter 2
Parameter 3
Parameter 4
…
Distance
Trip time
Car type
Passengers
Time of day
…
Gradient Boosted
Nearest Neighbors
SGD
Bayesian Regression
LGBM
…
Distance Gradient Boosted
30%
Model
Car type
Passengers
Getting started w/machine learning can be hard
ML.NET takes the guess work out of data prep, feature selection & hyperparameter tuning
Which algorithm? Which parameters?Which features?
23. #insiderDevTour
N Neighbors
Weights
Metric
P
ZYX
Which algorithm? Which parameters?Which features?
Distance
Trip time
Car type
Passengers
Time of day
…
Gradient Boosted
Nearest Neighbors
SGD
Bayesian Regression
LGBM
…
Nearest Neighbors
Criterion
Loss
Min Samples Split
Min Samples Leaf
XYZ
50%
Model
Iterate
30%
Gradient BoostedDistance
Car brand
Year of make
Car type
Passengers
Trip time
Getting started w/machine learning can be hard
ML.NET takes the guess work out of data prep, feature selection & hyperparameter tuning
25. #insiderDevTour
70%95% Feature importance
Distance
Trip time
Car type
Passengers
Time of day
0 1
Model B (70%)
Distance
0 1
Trip time
Car type
Passengers
Time of day
Feature importance Model A (95%)
ML.NET accelerates model development
with model explainability
27. #insiderDevTour
# STEP 1: Load data
IDataView trainingDataView = mlContext.Data.LoadFromTextFile<TaxiTrip>( ... )
IDataView testDataView = mlContext.Data.LoadFromTextFile<TaxiTrip>( ... )
# Display first few rows of the training data
ConsoleHelper.ShowDataViewInConsole(mlContext, trainingDataView)
# STEP 2: Initialize user-defined progress handler that AutoML will invoke after each model
var progressHandler = new RegressionExperimentProgressHandler()
# STEP 3: Run AutoML regression experiment
ExperimentResult<RegressionMetrics> experimentResult = mlContext.Auto()
.CreateRegressionExperiment(ExperimentTime)
.Execute(trainingDataView, LabelColumnName, progressHandler: progressHandler)
# Print top models found by AutoML
PrintTopModels(experimentResult)
# STEP 4: Evaluate the model on test data
RunDetail<RegressionMetrics> best = experimentResult.BestRun
ITransformer trainedModel = best.Model
# Run best model on test data
IDataView predictions = trainedModel.Transform(testDataView)
# STEP 5: Save trained model to a .ZIP file
mlContext.Model.Save(trainedModel, trainingDataView.Schema, ModelPath)
28. #insiderDevTour
Machine Learning for Developers
Enable Multi-Input
Azure Cognitive Services
Ink Recognizer
Increase Functionality
Windows AI platform
Sentiment Analysis
ML.NET
Price prediction
aka.ms/idt2019resources
29. #insiderDevTour
WinML
Practical, simple model-based API for ML
inferencing on Windows
DirectML
Realtime, high control ML operator API; part
of DirectX family
Compute Driver Model
Robust hardware reach/abstraction layer for
compute and graphics silicon
DirectML API
DirectX12
WinML API
ONNX Runtime
Compute Driver Model
GPU VPU xPU CPU
Windows AI platform
31. #insiderDevTour
# Load a model
var modelFile = await StorageFile.GetFileFromApplicationUriAsync(new Uri(modelPath))
LearningModel model = wait LearningModel.LoadFromStorageFileAsync(modelFile)
# Bind a model
LearningModelBinding binding = new LearningModelBinding(session)
ImageFeatureValue image = ImageFeatureValue.CreateFromVideoFrame(inputFrame)
binding.Bind(inputName, image);
# Evaluate
var result = await session.EvaluateAsync(binding, "0")
32. #insiderDevTour
✔ Azure Custom Vision makes it easy to create an image model
✔ WinMLTools converts existing models from TensorFlow, Keras, CoreML, scikit-learn, LIBSVM and
XGBoost
✔ Azure Machine Learning Service provides an end-to-end solution for preparing your data and
training your model
✔ ML.NET
Getting an ONNX Model
33. #insiderDevTour
WinML Benefits
• Low latency
• Cost Effectiveness
• Flexibility
DirectML API
DirectX12
WinML API
ONNX Runtime
Compute Driver Model
GPU VPU xPU CPU
Windows AI platform
34. Azure AI
Ink Recognizer
Windows AI platform
WinML
ML.NET
AutoML
What we’ve seen today…
…and two exciting tools announced at Build
DirectML
Vision Skills
35. ✔ Powers Windows ML hardware
acceleration
✔ Delivers broad hardware
support through DirectX12
✔ Built for real-time performance
✔ Raytracing
DirectML API
DirectX12
WinML API
ONNX Runtime
Compute driver model
GPU VPU xPU CPU
What is DirectML?
37. #insiderDevTour
Why AI and ML?
✔ The future of Apps
Azure Cognitive Services
✔ Comprehensive pre-trained AI
✔ Many new updates!
ML.NET
✔ AutoML
✔ Model Builder
The Windows AI platform
✔ WinML
✔ DirectML
✔ Vision Skills
Recap