You are here

Machine Learning Solutions | Cypress Semiconductor

Machine Learning Solutions

Machine Learning is progressing at a rapid rate with new Deep Learning algorithms playing a critical role in advancing the next phase of the IoT revolution. Artificial Intelligence and the Internet of Things have come together to create AIoT, which enables connected products to perform intelligent tasks.

The rapid increase in connected devices has led to an explosion in the amount of data being generated at the edge. Using this data to effectively gain insights is shifting towards using ML algorithms at the edge which are typically run on the cloud. Privacy, reliability and latency are barriers to using traditional approaches like uploading all your data to the cloud for analysis and machine learning. Running these algorithms efficiently on edge eliminates these barriers and allow AIoT products to scale rapidly.

Edge ML Solutions for the AIoT



Voice is quickly becoming a natural interface to technology like activity trackers and smart home assistants. Deep Learning-based Machine Learning algorithms are gaining popularity in almost all of the various stages of the voice pipeline, from wake word detection to intent recognition and Natural Language Processing (NLP). Moving ML to the edge for these voice applications significantly reduce latency and alleviate privacy concerns.


Predictive maintenance

Predictive maintenance

Predictive maintenance techniques detect anomalies in equipment before those turn into system-critical failures, allowing maintenance to be scheduled before the equipment actually breaks down. Monitoring the condition of equipment make use of various sensors for vibration analysis, sound anomaly detection, current sensing etc. to gain meaningful insights into equipment’s health.

Deploying predictive maintenance algorithms directly on Edge devices can significantly reduce the data transfer and connectivity requirement of the final system by being able to classify commonly known anomalies and send lower probability cases to the cloud for further analysis.



Intelligent event detection using sensors

Machine Learning for the Edge has particularly broad applications when processing and combining data from various sensors can detect specific events. A few key example of these are:

Glassbreak detection

Glassbreak detection

Glassbreak– when a window or door glass is broken, ML algorithms can the signature pattern of using audio sensors in combination with change in pressure inside the room to then trigger a glass break alarm. This fusion leads to a more robust detection algorithm with less false-positives.


Medical wearables

Medical wearables

Wearable medical technology is an application where privacy and latency are key design constraints. Anomalous events such as human fall detection need to run on edge devices for sending a high priority alert with low latency.



Radar-based presence detection and classification

The Infineon XENSIV™ radar sensor provides an anonymous alternative to cameras for complex presence detection applications such as people counting in smart entrances and buildings. Using ML algorithms running locally on the edge, counting people can become fully automated and anonymous because it only recognizes objects. These insights can unlock a variety of different use-cases, such as help us avoid crowded areas.

Machine Learning Workflow and Software

Developing ML algorithms for AIoT products traverse various technical teams, ranging from data scientists to embedded firmware developers. Infineon provides state-of-the-art tooling for Machine Learning model deployment on edge devices using ModusToolbox™ and works closely with major partners who can provide solutions for training and model selection for your specific application.


Hardware Products

The PSoC™ 6 family is built on an ultra-low-power architecture, and the MCUs feature low-power design techniques to extend battery life up to a full week for battery powered applications.

PSoC 6

PSoC™ 6

The XMC™ microcontroller family is based on ARM® Cortex®-M cores. It is dedicated to applications in the segments of power conversion, factory and building automation, transportation and home appliances.



The XENSIV™ family was developed to meet today’s sensing challenges in automotive, industrial and consumer applications. Building on the company’s in-depth system understanding, it is the broadest portfolio of sensor types on the market, giving customers the widest selection of ready-to-use solutions offering fast time-to-market.