دانا نادری

Technologies have always influenced human life, resulting in changes in the course of human civilization that have led to modern humanity's formation, and in which we have reached a new level of living with a wide range of facilities and comforts. Scientific approaches for treatment, manipulation, and simulation of human behavior according to the complex and multiple functions of the brain have created a new field called brain-machine interface (BMI). BMI is an ambitious and advanced technology that can be a turning point in human civilization. It is evident that artificial intelligence will play a significant role in improving this technology, and machine learning will be essential for translating recorded activities and signals. The aim of this study is to develop a method for translating and interpreting brain signals in an efficient way. In the presented approach, deep neural networks are used to estimate the parameters of the distribution of observations with a generalized dynamic Bayesian networks. In addition to its flexibility in expression and structure, this model is easier to interpret when combined with the opinions of experts. Also, unlike other detection and translation methods that have a preprocessing step for artifact rejection, noise filtering, outliers, and feature extraction, our model is modeled with raw data, making it suitable for real-time and online systems. The proposed model is validated using a data set related to sleep spindle detection, and the results show that it is more efficient than other models.