What is machine learning in IoT?

What is machine learning?

IoT and machine learning deliver insights otherwise hidden in data for rapid, automated responses and improved decision making. Machine learning for IoT can be used to project future trends, detect anomalies, and augment intelligence by ingesting images, videos, and audio files.

Why use machine learning for IoT?

Machine learning can help demystify the hidden patterns in IoT data by analyzing massive volumes of data using sophisticated algorithms. Machine learning inference can supplement or replace manual processes with automated systems using statistically derived actions in critical processes.

Sample use cases

Companies are utilizing machine learning for IoT to perform predictive capabilities on a wide variety of use cases that enable the business to gain new insights and advanced automation capabilities.

With machine learning for IoT, you can:

• Ingest and transform data into a consistent format.

• Build a machine learning model.

• Deploy this machine learning model on cloud, edge, and device.

For example, using machine learning, a company can automate quality inspection and defect tracking on its assembly line, track activity of assets in the field, and forecast consumption & demand patterns.

Figure 1: Cumulocity machine learning architecture.

Benefits of machine learning inference for IoT

Machine learning is a key component of Cumulocity’s low-code, self-service IoT platform. The platform comes ready to go with tools you need for fast results: device integration and management, application enablement and integration, as well as streaming analytics, machine learning, and machine learning model deployment. The platform is available on the cloud, on-premises, and/or at the edge. Uniquely with Cumulocity, standalone, edge-only solutions are also supported.

Simplify machine learning model training

Cumulocity is designed to help you quickly build new machine learning models in an easy manner. AutoML support allows the right machine learning model to be chosen for you based on your data, whether that be operational device data captured on the Cumulocity platform or historical data stored in big data archives.

Flexibility to use your data science library of choice

There are a wide variety of data science libraries available (e.g., Tensorflow®, Keras, Scikit-learn) for developing machine learning models. Cumulocity allows models to be developed in data science frameworks of your choice. These models can be transformed into industry-standard formats using open-source tools and made available for scoring within the Cumulocity platform.

Rapid model deployment to operationalize machine learning quickly

Model deployment into production environments is possible wherever needed in one click, either in the cloud or at the edge. Operationalized models can be easily monitored and updated if underlying patterns shift. Additionally, pretrained and verified models are available for immediate model deployment to accelerate adoption.

Pre-built connectors for operational & historical datastores

Cumulocity provides easy access to data residing in operational and historical datastores for model training. It can retrieve this data on a periodic basis and route it through an automated pipeline to transform the data and train a machine learning model. Data can be hosted on Amazon® S3 or Microsoft® Azure® Data Lake Storage, as well as local data storage, and retrieved using pre-built Cumulocity DataHub connectors.

Integration with Cumulocity streaming analytics

Cumulocity enables high-performance scoring of real-time IoT data within Cumulocity streaming analytics. Cumulocity streaming analytics provides a “Machine Learning” building block in its visual analytics builder that allows the user to invoke a specified machine learning model to score real-time data. This provides a no-code environment to integrate machine learning models with streaming analytics workflows.

Notebook integration

Jupyter Notebook, a de facto standard in data science, provides an interactive environment across programming languages. They can be used to prepare and process data, train, deploy, and validate machine learning models. This open-source web application is integrated with Cumulocity.