Taking Models to the Next Level with Azure Machine Learning
summary
Starts with Azure Cognitive Services for quick sentiment and key-phrase extraction then builds an AutoML baseline on a 1.8 MB Tailwind Traders clothing reviews dataset (AutoML run ≈28 minutes) to get a logistic-regression baseline. Moves to Intel NLP Architect for Aspect-Based Sentiment Analysis, trains on Azure Machine Learning with compute clusters and registers model version 12, deploys to ACI/AKS, and runs HyperDrive hyperparameter search (≈16–18 trials) that improves F1 and returns aspect–opinion outputs like “love → sweater” and “hate → pants.”
AI-written summary · about
chapters (23)
- 0:00Intro
- 0:57The Story
- 1:38Why Azure Cognitive Services
- 5:03programming
- 7:03Building a Baseline with AutoML
- 13:36NLP-Architect Aspect Based Sentiment Analysis
- 14:07Managing Tailwind Traders ML Challenges
- 15:38Challenges of distributed training
- 16:14Azure Machine Learning Typical end to end ML process Prepare
- 22:14Create Compute
- 28:19Create an experiment
- 28:47Create a training file
- 29:00Create an estimator
- 29:53Submit the experiment to the cluster
- 37:56Register the model
- 39:42AMLS to deploy
- 40:56Inference config inference_config - InferenceConfig
- 41:06Deployment using AML
- 41:22Deploy to ACI
- 49:31Typical 'manual' approach to hyperparameter tu
- 50:23Automated Hyperparameter Tuning Manage Active Jobs
- 55:20Wrap up
- 56:16Learning Resources Get started with Azure ML Services and the Python SDK aka.ms/AA3dzht
related talks
51mTaking Models to the Next Level with Azure Machine Learning
01h 14mAI of Things
58mDeep Dive into Azure Automated Machine Learning
01h 01mAI and ML imbedded in Azure Data Platform
01h 00m5 Levels of AI in Microsoft Azure
01h 00mLearn Live: Get started with AI on Azure
01h 03mMachine Learning for Muggles
01h 05mAzure Machine Learning for the Developer