🤖 Automating AI Model Design for Edge AI
DeepGate has introduced an automated model design system for microcontrollers (MCU). The technology combines Neural Architecture Search (NAS) using LLM agents and Supernet NAS, working in conjunction with a proprietary compiler and real-world hardware measurements.
🌍 The transition to an automated "design — training — hardware measurement" cycle allows complex ML models to be ported to extremely low-cost devices, expanding the Edge AI market to billions of microcontrollers.
👤 This means that AI functions (such as voice control or anomaly detection) will soon become standard even in the cheapest sensors and gadgets, operating locally and without the cloud.
Source 1: https://deepgate.ai/blog/neural-architecture-search