Stacked AI chip uses fast and slow layers to recognize changes in motion over time

Korean researchers developed a stacked semiconductor AI chip using solid ion-gated transistors with different response speeds to process fast and slow changes in motion over time. This chip architecture can enhance wearable devices' ability to analyse movement and physiological signals by recognising temporal patterns with over 90% accuracy in video classification tasks.

This Wire brief sits within Fusion42's coverage of AI & ML.