Figure 12.

(a) Illustration of the dynamic neural network (DNN) for the performance of the low-resonant-frequency piezoelectric MEMS energy harvester. Adapted with permission from Nevlydov et al. [180]. (b) Visual representations of the gaussian mixture model (GMM)-based machine learning algorithm for the speaker recognition, and the dataset of 90% used for training data, and 10% for testing data by 2800 training data of 40 people. Adapted with permission from Han et al. [184]. (c) Raw data transmission of multiple sensors through Wi-Fi for data transmission. Adapted with permission from Suh et al. [185]. (d) Schematic illustration of the promising applications using flexible piezoelectric acoustic sensors in response to the speaker’s voice and the data were trained using a machine learning-based model. Adapted with permission from Jung et al. [186]. (e) Effects of motion on fabric, the motion signals from sensors were located on both a rigid base and loosely attached to the base via the fabric. Adapted with permission from Michael et al. [187].