Overview

TinyML is a subset of Edge AI that targets the smallest devices, often operating in the milliwatt power range. It enables 'always-on' intelligence in sensors and consumer electronics.

Key Concepts

  • Microcontrollers (MCUs): Low-cost, low-power chips (e.g., ARM Cortex-M) that lack the resources of a full OS.
  • On-device Inference: Running the model entirely on the chip without external help.
  • Ultra-low Power: Devices that can run for years on a single coin-cell battery.

Applications

  • Keyword spotting ('Hey Siri', 'OK Google').
  • Gesture recognition in wearables.
  • Predictive maintenance in industrial sensors.
  • Environmental monitoring (e.g., detecting specific sounds in a forest).

Related Terms