This chapter proposes to analyze two configurations of neural networks to compose the expert set in the development of a multilevel speech signal pattern recognition system of 30 commands in the Brazilian Portuguese language. Then, multilayer perceptron (MLP) and learning vector quantization (LVQ) networks have their performances verified during the training, validation and test stages in the speech signal recognition, whose patterns are given by two-dimensional time matrices, result from mel-cepstral coefficients coding by the discrete cosine transform (DCT). In order to avoid the pattern separability problem, the patterns are modified by a nonlinear transformation to a high-dimensional space through a suitable set of Gaussian radial base functions (GRBF). The performance of MLP and LVQ experts is improved and configurations are trained with few examples of each modified pattern. Several combinations were performed for the neural network topologies and algorithms previously established to determine the network structures with the best hit and generalization results.

The diagnosis of epilepsy is conducted through visual inspection of electroencephalogram (EEG) signal recordings. However, due to the variations in convulsive disorders, it can be challenging for clinicians to constantly monitor the patient for seizure type, especially because EEG records contain hours of signal. Nevertheless, these patterns present in EEG signals can also be identified through signal classification methods based on signal processing and machine learning approaches. In light of this, this study proposes the development of a methodology for epileptic seizure type classification based on analysis of time-frequency characteristics of EEG signals, using Continuous Wavelet Transform (CWT) and joint moments of time-frequency distribution. Epileptic seizure classification was performed using a convolutional neural network (CNN), employing k-fold cross-validation methods. Accuracy, sensitivity, specificity, and area under the curve (AUC) metrics were obtained to validate this algorithm. The achieved results for the CNN classifier were 96.54% accuracy, 96.54% sensitivity, 96.54% specificity, and AUC = 0.90%.