Machine Learning Models
Transform raw machine data into Digital Twins with machine learning models
How the Arch Factory Exchange Works
Transform your raw machine data into insights with machine learning models and digital twins. Deploy these models on the cloud or edge server to maximize performance.
Machine Learning Models Products and Services
Machine Learning Models As-a-Service
Building machine learning models from raw data is one of the most challenging and least appreciated parts of the product stack. When done correctly, the machine model produces machine states and changes in real-time, surfacing downtime reasons and making it possible to produce real-time dashboards that drive action as well as making it possible to build a large variety of historical and predictive applications on top.
Arch’s analytics team builds machine models for our clients to transform raw data into insights. Our team combines cutting edge ML/AI knowledge with industrial domain expertise and practical experience. AI is too urgent not to be investing in and also too important just to outsource. We are your partner to deliver results, capture ROI today, set up training programs, and help empower your own data science teams to use the same tools.
Our products are built on open-source IOTile technology, which provides a growing array of developer tools to customize the complete stack including building your own machine models. The tools used to develop machine models are specific to the architecture upon which the model is running. Specifically: will this model be run in the cloud, on an edge server, on an embedded-Linux edge device, or on low-power, MCU IOTile sensor hardware?
On IOTile sensor hardware, IOTile firmware includes functionality called SensorGraph Embedded™, a patented scheduler that is unique to hardware that follows the IOTile standard. This allows developers to quickly and effectively change the way data is pulled, processed, and communicated. Using SensorGraph Pipelines™, developers can set up stages of filtering and analysis steps that make it easy to successively transform a raw data signal such as current data or accelerometer data into events.
On embedded Linux hardware, IOTile includes functionality called SensorGraph Realtime™ that is a machine-model processing library for deploying states of filtering and analysis steps on data parsed from logs or passed from an IOTap.
Buy a Starter Kit to get the hardware you need and read the open source documentation or reach out to the team to learn more about how to develop your own machine models with IOTile technology.