#40 · Primary category： Classical Machine Learning Frameworks
EdgeML
This repository provides code for machine learning algorithms for edge devices developed at Microsoft Research India.
Project last updated：05/20/24
GitHub Stars
1.7K
Forks
388
Contributors
40
License
Other
Why we included this project
On-device machine learning means fitting a model into a few kilobytes of memory and still getting predictions in real time, with no cloud connection to lean on. EdgeML, developed at Microsoft Research, collects training routines and model architectures built for exactly that constraint: the Bonsai tree classifier, ProtoNN's prototype-based kNN, and the FastGRNN family of compact recurrent cells. You train on a regular machine using TensorFlow or PyTorch, then load the trained model onto the device for fully offline inference. The repository also includes SeeDot, a floating-point to fixed-point quantization tool, and working examples such as GesturePod, which does gesture recognition on microcontrollers. If you are trying to figure out whether on-device ML is viable for your product, this is a good place to start measuring the tradeoff between accuracy and model size.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
scikit-learn
scikit-learn: machine learning in Python
pyod
A Python library for anomaly detection across tabular, time series, graph, text, image, and audio data. 60+ detectors, benchmark-backed ADEngine orchestration, and an agentic workflow for AI agents.
dowhy
DoWhy is a Python library for causal inference that supports explicit modeling and testing of causal assumptions. DoWhy is based on a unified language for causal inference, combining causal graphical models and potential outcomes frameworks.
umap
Uniform Manifold Approximation and Projection
imbalanced-learn
A Python Package to Tackle the Curse of Imbalanced Datasets in Machine Learning