Efficient Authentication System Using Wavelet Embeddings of Otoacoustic Emission Signals

Biometrics, which has become integrated with our daily lives, could fall prey to falsification attacks, leading security concerns. In paper, we use Transient Evoked Otoacoustic Emissions (TEOAE) that are generated by the human cochlea in response an external sound stimulus, as a biometric modality. TEOAE robust uniqueness of individual’s inner ear cannot be impersonated. this study, both raw 1D signals, well 2D time-frequency representation signal using Continuous Wavelet Transform (CWT). We and Convolutional Neural Networks (CNN) for former latter, respectively, derive feature maps. The corresponding lower-dimensional maps obtained principal component analysis, is then used features build classifiers machine learning techniques task person identification. T-SNE plots these show they discriminate among subjects. Among various architectures explored, achieve best-performing accuracy 98.95% 100% 1D-CNN 2D-CNN, latter performance being improvement over all earlier works. This makes based identification systems deployable real-world situations, along added advantage robustness attacks.