AIfES Update:

Since the beginning of the year an AIfES® update (version 2.1.1) is available. It can be downloaded directly from the Arduino IDE using the Arduino Library Manager or alternatively from GitHub.

Here is an overview of the new features:

- The new AIfES-Express API:

AIfES-Express is an alternative and greatly simplified API that is integrated directly into the library. The new features allow you to run and train a feed-forward neural network (FNN) with just a few lines of code.

(A tutorial is available here)
 

- Q7 Weight Quantization:

The update enables simple Q7 (8 bit) quantization of the weights of a trained FNN. This significantly reduces the amount of memory required by the FNN and, depending on the controller type, also provides a significant speed increase.
 

- Arm® CMSIS integration enhanced by NN functions:

AIfES® offers the possibility to use the Arm® CMSIS (DSP and NN) library for faster runtime (e.g. faster inference).
 

A lot of new examples:

- Simple gesture recognition that can be trained directly on the device.

- Play tic-tac-toe against your microcontroller. The pre-trained net is practically impossible to defeat.

- Dual Core Training for the Arduino Portenta H7. The example shows how to train in the background on one core while the other does a completely different task.

 

Follow AIfES on our LinkedIn channel and AIfES® on GitHub to stay up to date.

 

Certified MLOps training with AIfES® introduction and practical example:

Together with our colleagues from Fraunhofer IAIS, we have prepared the certified online training "Certified Data Scientist Specialized in MLOps". In the training, you will learn methods and tools for operationalizing machine learning applications ("Machine Learning Operations (MLOps)"). There will also be an introduction to the requirements of Edge/Embedded AI and how to easily implement it on a microcontroller using AIfES®. In addition, a practical example will be implemented with the participants.

Description and registration for the training.