Resources aka msAIML 40 repo aka ms9 Tips

  • Slides: 44
Download presentation

Resources aka. ms/AIML 40 repo aka. ms/9 Tips. Prod. ML aka. ms/mymsignitethetour

Resources aka. ms/AIML 40 repo aka. ms/9 Tips. Prod. ML aka. ms/mymsignitethetour

Recap Machine Learning Applications Autonomous Agents Predictive Maintenance Forecasting Language Understanding E-commerce Recommendations Computer

Recap Machine Learning Applications Autonomous Agents Predictive Maintenance Forecasting Language Understanding E-commerce Recommendations Computer Vision

Let’s Dive Deeper: Trends In Language Understanding Ambient Computing Unlocking Unstructured Data Language Understanding

Let’s Dive Deeper: Trends In Language Understanding Ambient Computing Unlocking Unstructured Data Language Understanding Automation and Solution Scale

The Story Existing pre-built models for text analytics Custom models trained on specific data

The Story Existing pre-built models for text analytics Custom models trained on specific data

Demo Text Analytics

Demo Text Analytics

/Docs alert Get started with Azure Cognitive Services Text Analytics API aka. ms/Text. Cog.

/Docs alert Get started with Azure Cognitive Services Text Analytics API aka. ms/Text. Cog. Svc

From Cognitive Services to Azure Machine Learning

From Cognitive Services to Azure Machine Learning

What is Azure Machine Learning Service? Set of Azure Cloud Services Python SDK That

What is Azure Machine Learning Service? Set of Azure Cloud Services Python SDK That enables you to: ü Prepare Data ü Build Models ü Train Models ü Manage Models ü Track Experiments ü Deploy Models

Azure ML Service Key Artifacts Workspace

Azure ML Service Key Artifacts Workspace

Building a Baseline with Auto. ML Input Azure Auto ML Intelligently test multiple models

Building a Baseline with Auto. ML Input Azure Auto ML Intelligently test multiple models in parallel Optimized model Enter data Define goals Apply constraints

Demo From Text Analytics to Auto ML

Demo From Text Analytics to Auto ML

/Docs alert Get started with Automated Machine Learning https: //aka. ms/Automated. MLDoc

/Docs alert Get started with Automated Machine Learning https: //aka. ms/Automated. MLDoc

Advanced Azure ML Services

Advanced Azure ML Services

NLP-Architect Aspect Based Sentiment Analysis

NLP-Architect Aspect Based Sentiment Analysis

Azure Machine Learning Typical E 2 E Process Experiment Prepare Deploy … Orchestrate

Azure Machine Learning Typical E 2 E Process Experiment Prepare Deploy … Orchestrate

Azure Machine Learning Typical E 2 E Process Experiment Prepare Deploy … Orchestrate

Azure Machine Learning Typical E 2 E Process Experiment Prepare Deploy … Orchestrate

Managing Tailwind Traders ML Challenges

Managing Tailwind Traders ML Challenges

Challenges of distributed training Dependencies and Containers Handle failures Schedule jobs Secure Access Distribute

Challenges of distributed training Dependencies and Containers Handle failures Schedule jobs Secure Access Distribute data Gather results Scale resources Provision clusters of VMs

Features of Distributed Training Horovod distributed training with

Features of Distributed Training Horovod distributed training with

/Docs alert Get Started by setting up your Azure ML Environment https: //aka. ms/amlsdk

/Docs alert Get Started by setting up your Azure ML Environment https: //aka. ms/amlsdk

Distributed Hyperparameter Tuning

Distributed Hyperparameter Tuning

What are Hyperparameters? Config 1= {“learning_rate”: 0. 2, “num_layers”: 2, …} { “learning_rate”: uniform(0,

What are Hyperparameters? Config 1= {“learning_rate”: 0. 2, “num_layers”: 2, …} { “learning_rate”: uniform(0, 1), “num_layers”: choice(2, 4, 8) … } Config 2= {“learning_rate”: 0. 5, “num_layers”: 4, …} Config 3= {“learning_rate”: 0. 9, “num_layers”: 8, …} …

Typical ‘manual’ approach to hyperparameter tuning Dataset Training Algorithm 1 Algorithm 2 Hyperparameter Values

Typical ‘manual’ approach to hyperparameter tuning Dataset Training Algorithm 1 Algorithm 2 Hyperparameter Values – config 1 Values – config 2 Hyperparameter Values – config 3 Values – config 4 Model Training Infrastructure Complex Tedious Repetitive Time consuming Expensive Model 1 Model 2 Model 3 Model 4

Automated Hyperparameter Tuning Manage Active Jobs Evaluate training runs for specified primary metric Early

Automated Hyperparameter Tuning Manage Active Jobs Evaluate training runs for specified primary metric Early terminate poor performing training runs. Early termination policies include: Early terminate poorly performing runs

/Docs alert Get started with Azure Machine Learning Service Hyperparameter Tuning https: //aka. ms/Azure.

/Docs alert Get started with Azure Machine Learning Service Hyperparameter Tuning https: //aka. ms/Azure. MLHyper. Drive

How Azure ML Workspace Docker Image Compute Target My Computer Data Store

How Azure ML Workspace Docker Image Compute Target My Computer Data Store

The 8 Azure ML Train Steps

The 8 Azure ML Train Steps

Demo: Operationalizing Open Source with Azure ML and NLP Architect

Demo: Operationalizing Open Source with Azure ML and NLP Architect

Wrapping Up

Wrapping Up

Review

Review

Takeaways

Takeaways

/Docs alert Train an image classification model with MNIST data and Scikit-learn https: //aka.

/Docs alert Train an image classification model with MNIST data and Scikit-learn https: //aka. ms/Azure. MLTrain

/Docs alert Getting Started Series… https: //aka. ms/Text. Cog. Svc https: //aka. ms/Automated. MLDoc

/Docs alert Getting Started Series… https: //aka. ms/Text. Cog. Svc https: //aka. ms/Automated. MLDoc https: //aka. ms/Azure. MLHyper. Drive https: //aka. ms/AA 3 dzht Other Materials Azure ML Best Practices Repos github. com/microsoft/nlp-recipes/ • Computer Vision github. com/microsoft/Computer. Vision • Recommenders github. com/microsoft/Recommenders https: //aka. ms/9 MLTips

/MS Learn alert aka. ms/AIML 40 MSLearn. Collection

/MS Learn alert aka. ms/AIML 40 MSLearn. Collection

/Microsoft Certification alert aka. ms/Data. Scientist. Cert aka. ms/Data. Engineer. Cert Get hired, stay

/Microsoft Certification alert aka. ms/Data. Scientist. Cert aka. ms/Data. Engineer. Cert Get hired, stay ahead, and receive the recognition you deserve #MSIgnite. The. Tour

Resources Get Certified aka. ms/AIML 40 repo Azure Data Scientist: aka. ms/Data. Scientist. Cert

Resources Get Certified aka. ms/AIML 40 repo Azure Data Scientist: aka. ms/Data. Scientist. Cert Azure Data Engineer Associate: aka. ms/Data. Engineer. Cert aka. ms/mymsignitethetour #MSIgnite. The. Tour

Microsoft. com/Certifications Microsoft. com/Learn aka. ms/Learning. Partner

Microsoft. com/Certifications Microsoft. com/Learn aka. ms/Learning. Partner

Session alert

Session alert