1chipML is an open source library for basic numerical crunching and machine learning for microcontrollers. As the Internet of Things and Edge Computing are becoming a ubiquitous reality, we need to a reliable and open framework to use on limited and low power demanding hardware. The expected benefits are endless and we expect 1chipML will play an essential role in enabling this new computing revolution!
This library is implemented by team of developers who are refered to as maintainers or contributors, depending on their actual role.
The 1chipML project is hosted as a Sandbox project at the LF AI & Data Foundation.
This package does not necessitate any installation, it is essentially a collection of independent numerical and machine learning methods which target microcontrollers. Therefore, the installation is going to depend on the hardware itself along with the software tools used (e.g., the compiler).
In any case, we do our best to ensure that the library can be compiled by the main compilers provided by AVR and ARM microcontrollers (avr-gcc and Cortex-M compilers) by writing code that is entirely in the ANSI C language.
Therefore, in order to use this library in practical cases, you need to select one or more of the source files which can be found in the src forlder. A very efficient way to start is by studying the source files stored in the tests folder.
Although the target of this library is represented by microcontrollers, it is possible to run it on a common personal computer. This is twofold, on one hand it allows the user to have a feeling about what this library can perform, on the other hand it allows the debug of codes on a computer before deployment on microcontrollers.
To run the tests on computer, simply type the following command in a shell:
make testTo illustrate the usage of each algorithm, some prototypes have been developed. Each prototype located in the prototype folder brings to light how a specific algorithm can be used. We recommend you look at these prototypes to have a better understanding of how each algorithm can be used for development. Most of them have been implemented to work with the platform Arduino IDE and all have a unique README file to better understand their implementation.