Диагностика и Naïve Bayes класификация на шумове в комуникационни канали за връзка
The report proposes a modular approach for the statistical diagnosis and recognition of unwanted interference in communication link channels. The object of the research is Gaussian and Periodic random noise. Analytical methods are applied to the noises through Normal and Kernel probability density functions. Training, evaluation, and verification of confounding type identification models were conducted using a Naïve Bayes machine learning algorithm. Two concepts are applied with Normal and Kernel distribution of input data when exploring Naïve Bayes classification models. Each concept for generating classification models is evaluated by the Resubstitution and Cross-Validation approaches until the most adequate classifier is established.
More...
