Analysis of ECG Data by Energy Efficient Decision Trees on a Reconfigurable ASIC

Explore a tinyML Talk from Germany on analyzing ECG data using energy-efficient decision trees implemented on a reconfigurable ASIC. Dive into the Pilot Innovation Initiative "KI-Sprung" project, which aimed to develop an algorithm for detecting atrial fibrillation episodes in ECG data. Learn about the innovative architecture for decision tree ensembles, feature selection optimization, and hardware implementation strategies. Discover how the team achieved a 92.7% recall and 14.7% fall-out rate while consuming only 42nJ of energy for a 2-minute ECG dataset. Gain insights into feature analysis, peak detection, cost function optimization, classification in hardware, and energy breakdown in this comprehensive 55-minute presentation.