Azure Machine Learning
In a nutshell, Azure Machine Learning (ML) is Microsoft’s fully managed cloud-based ML environment for making predictions from data. It is comprised of a collection of services and tools that allow you to train, deploy, and manage ML models at an enterprise scale.
The value of ML to an organization is the ability to predict future trends or behavior; Azure ML’s capabilities allow these predictions to be made better, faster, simpler, and with a lower cost barrier.
The Azure ML service provides capabilities for building and deploying models while having complete control of the design of the algorithm, training, and your data.
The models can be trained and deployed at scale using web interfaces and Software Development Kits (SDKs). In addition, open source technologies can be used, including Python frameworks such as PyTorch, TensorFlow, and scikit-learn; Azure ML also supports Python and R. The following diagram outlines an Azure ML solution approach:
Figure 5.7 – Azure Machine Learning solution approach
As we learned in the Artificial intelligence solutions section, ML is a data science technique and a subset of AI.
The most common use cases for ML within AI are image analysis, natural interaction, speech comprehension, and making predictions from data.
The core concept of ML is based on algorithms; this can be thought of as the magic sauce, as it were. An algorithm is a way to solve a problem or carry out dataset analysis using a sequence of calculations and rules. For example, an algorithm, in its simplest form, can determine what similar category objects should be classified. The following figure shows an example of where an algorithm would be used to figure out what’s a dog or a muffin when classifying these images:
Figure 5.8 – Sadie, Mrs Smile’s Muffins, or Daisy
This section looked at Azure Machine Learning. In the next section, we will look at Azure Cognitive Services.
