Chapter 09 · Sustainability & AISustainability & AI

TinyML

Meaning statusEstablishedSource recordDirect document linkedWhy these are different

Definition

Machine learning on severely resource-constrained hardware — microcontrollers and sensors operating on milliwatts of power — enabling intelligence at the far edge.

References

tinyML FoundationtinyML community and foundation

This source provides part of the technical or institutional basis for the definition.

arXivHinton et al. — Distilling the Knowledge in a Neural Network (2015)

This source supports the explanation of how the term is applied, measured or governed in practice.

Overview

What it means

TinyML compresses models through quantisation, pruning and efficient architectures so they run on battery or energy-harvesting devices for months or years. A dedicated research community and industry foundation coordinate methods, benchmarks and tools.

How it is used

Flagship uses are distributed environmental sensing: acoustic monitors recognising chainsaws or species in forests, low-power camera traps, soil and water sensors, and smart-meter analytics — all without mains power or connectivity.

Why it matters

TinyML expands the observable world: it puts pattern recognition where infrastructure is absent and where conservation and agriculture actually happen. It is also the efficiency endgame of AI — capability per milliwatt as the design target.