Tijekom treninga polaznici će usvojiti vještine i znanja potrebne za upravljanje rješenjima strojnog učenja na cloud razini koristeći Azure Machine Learning.
Osim toga, polaznici će naučiti kako primijeniti svoje postojeće znanje o Pythonu i strojnom učenju za upravljanje unosom i pripremom podataka, obuku modela i implementaciju te praćenje rješenja za strojno učenje uz Azure Machine Learning i MLflow.
Što ćete naučiti
Kome je namijenjeno
- Podatkovnim znanstvenicima (engl. Data scientists) koji posjeduju znanje o Pythonu i okvirima za strojno učenje kao što su Scikit-Learn, PyTorch i Tensorflow, a koji žele izgraditi i upravljati rješenjima za strojno učenje u oblaku.
Preduvjeti
- Osnovna znanja o konceptima računalstva u oblaku (eng. cloud computing) i znanosti o podacima te iskustvo u radu s alatima i tehnikama strojnog učenja (eng. machine learining).
- Stvaranje resursa u oblaku u Microsoft Azureu.
- Korištenje Pythona za istraživanje i vizualizaciju podataka.
- Obuka i provjera valjanosti modela strojnog učenja pomoću uobičajenih okvira kao što su Scikit-Learn, PyTorch i TensorFlow.
- Explore Microsoft cloud concepts
- Create machine learning models
- Administer containers in Azure
Nastavni plan
Pregledaj
- Module 1: Design a data ingestion strategy for machine learning projects Learn how to design a data ingestion solution for training data used in machine learning projects. After completing this module, students will be able to:
- Identify your data source and format
- Choose how to serve data to machine learning workflows
- Design a data ingestion solution
- Identify machine learning tasks
- Choose a service to train a model
- Choose between compute options
- Understand how a model will be consumed
- Decide whether to deploy your model to a real-time or batch endpoint
- Explore an MLOps architecture
- Design for monitoring
- Design for retraining
- Create an Azure Machine Learning workspace.
- Identify resources and assets
- Train models in the workspace
- The Azure Machine Learning studio
- The Python Software Development Kit (SDK)
- The Azure Command Line Interface (CLI)
- Work with Uniform Resource Identifiers (URIs)
- Create and use datastores
- Create and use data assets
- Choose the appropriate compute target
- Create and use a compute instance
- Create and use a compute cluster
- Understand environments in Azure Machine Learning
- Explore and use curated environments
- Create and use custom environments
- Prepare your data to use AutoML for classification
- Configure and run an AutoML experiment
- Evaluate and compare models
- Configure to use MLflow in notebooks
- Use MLflow for model tracking in notebooks
- Convert a notebook to a script
- Test scripts in a terminal
- Run a script as a command job
- Use parameters in a command job
- Use MLflow when you run a script as a job
- Review metrics, parameters, artifacts, and models from a run
- Define a hyperparameter search space
- Configure hyperparameter sampling
- Select an early-termination policy
- Run a sweep job
- Create components
- Build an Azure Machine Learning pipeline
- Run an Azure Machine Learning pipeline
- Log models with MLflow
- Understand the MLmodel format
- Register an MLflow model in Azure Machine Learning
- Understand Azure Machine Learning's built-in components for responsible AI
- Create a Responsible AI dashboard
- Explore a Responsible AI dashboard
- Use managed online endpoints
- Deploy your MLflow model to a managed online endpoint
- Deploy a custom model to a managed online endpoint
- Test online endpoints
- Create a batch endpoint
- Deploy your MLflow model to a batch endpoint
- Deploy a custom model to a batch endpoint
- Invoke batch endpoints
Za što vas priprema?
Certifikacijski ispit: Exam DP-100: Designing and Implementing a Data Science Solution on Azure Certifikat: Microsoft Certified: Azure Data Scientist Associate