Build an AI model

What is machine learning?

New in machine learning (ML)? Get familiar with machine learning.

Use ONNX - Open Neural Network Exchange

ONNX is open format to represent traditional and deep learning models. It’s developed in co-operation between Microsoft, Facebook, AWS and several other partners.

Azure Cognitive Services

Azure Cognitive Services are APIs, SDKs, and services available to help developers build intelligent applications without having direct AI or data science skills or knowledge. Azure Cognitive Services enable developers to easily add cognitive features into their applications.

Ready get started?

• Create your vision AI model with Custom Vision.
• Start development with Azure Machine Learning.
• Find AI models.

Characteristics of main machine learning use cases

	Vision	Audio	Telemetry
Summary	Vision AI is specialized machine learning for images and video that allows you to transform camera input into labels and objects that you can use in your applications and business logic.	Machine learning solutions for audio can be further categorized into two main subcategories - speech to text used for example by smart assistant and anomaly detection based on sound	Telemetry refers to numeric data received from multiple different kinds of sensors for example temperature or pressure. The telemetry data amounts produced are typically significantly smaller than with vision or audio
Typical use cases	Workplace safety, Retail, Quality control (anomaly detection), Smart home/building/city, Access control	Speech to text, anomaly detection (predictive maintenance)	Anomaly detection, Predictive maintenance
Main challenges	Large amounts of data, HW acceleration needed for real time inferencing	Speech to text: Microphone and model dependency, Acquiring audio transcription data for training	Building a model often requires in depth machine learning knowledge. Limited amount of tools available.