TI authors evaluate machine-learning techniques for analog waveform prediction

A Texas Instruments ISQED 2023 paper evaluates a machine-learning behavioral-model method for time-domain waveform prediction using feature extraction, waveform segmentation, and circuit partitioning. In an operational-amplifier proof of concept, the authors report an average output-prediction signal-to-noise ratio of 32 dB. The result is an evaluated proof of concept, not evidence of production deployment.