Tiny Embedded Intelligence Layer

Tiny Embedded Intelligence Layer (TEIL) is a cutting-edge framework designed to efficiently deploy machine learning (ML) models on resource-scarce embedded systems, such as Arduino, ARM Cortex-M, and similar devices. Developed as part of my PhD thesis, TEIL aims to bridge the gap between advanced ML techniques and the limitations of embedded systems, enabling real-time intelligence in environments with constrained resources.

TEIL offers seamless compatibility with popular machine learning libraries, including Scikit-Learn, and goes a step further by transpiling ML models into optimized C code, ensuring that these models can run effectively on devices with limited memory and processing power. In addition to supporting machine learning models, TEIL also provides capabilities for real-time data analysis, verification of ML models during runtime, and reducing the need for constant cloud communication, making it a versatile solution for autonomous, edge-based intelligence. With TEIL, embedded systems can harness the power of machine learning while maintaining performance and efficiency.

Supported ML Algorithms

Because there are many ML algorithm that exist, this framework is limited to the following:

Algorithm	Task	Documentation
Decision Tree	Classification	readthedocs
Decision Tree	Regression	readthedocs
Multi-Layer Percepton	Classification	readthedocs
Multi-Layer Percepton	Regression	readthedocs
SVM (Linear)	Classification	readthedocs
SVM (Linear)	Regression	readthedocs
SVM (RBF)	Classification	readthedocs
SVM (RBF)	Regression	readthedocs
SVM (Polynomial)	Classification	readthedocs
SVM (Polynomial)	Regression	readthedocs
PCA	Dimensionality Reduction / Decomposition
Gaussian Naive Bayes	Classification
Multinomial Naive Bayes	Classification