You will be surprised by how much accuracy you can achieve in just a few kylobytes of resources: Decision Tree, Random Forest and XGBoost (Extreme Gradient Boosting) are now available on your microcontrollers: highly RAM-optmized implementations for super-fast classification on embedded devices.
Fourier Transform is probably the most well known algorithm for feature extraction from time-dependent data (in particular speech data), where frequency holds a great deal of information. Sadly, computing the transform over the whole spectrum of the signal still requires O(NlogN) with the best implementation (FFT - Fast Fourier Transform); we would like to achieve faster computation on our microcontrollers.
In this post I propose a partial, naive linear-time implementation of the Fourier Transform you can use to extract features from your data for Machine Learning models.
Let's revamp the post I wrote about word classification using Machine Learning on Arduino, this time using a proper microphone (the MP34DT05 mounted on the Arduino Nano 33 BLE Sense) instead of a chinese, analog one: will the results improve?
Are you looking for a top-performer classifiers with a minimal amount of parameters to tune? Look no further: Gaussian Naive Bayes is what you're looking for. And thanks to EloquentML you can now port it to your microcontroller.
A brand new binary classifier that's tiny and accurate, perfect for embedded scenarios: easily achieve 90+ % accuracy with a minimal memory footprint!
When working with Machine Learning projects on microcontrollers and embedded devices the dimension of features can become a limiting factor due to the lack of RAM: dimensionality reduction (eg. PCA) will help you shrink your models and even achieve higher prediction accuracy.
In earlier posts I showed you can run incremental binary classification on your microcontroller with Stochastic Gradient Descent or Passive-Aggressive classifier. Now it is time to upgrade your toolbelt with a new item: One-vs-One multiclass classifier.
Stochastic gradient descent is a well know algorithm to train classifiers in an incremental fashion: that is, as training samples become available. This saves you critical memory on tiny devices while still achieving top performance! Now you can use it on your microcontroller with ease.
When working with memory constrained devices you may not able to keep all the training data in memory: passive-aggressive classifiers may help solve your memory problems.