Feature Vector Generation for Highly Accurate Traffic Distribution Prediction by Supervised Variational Auto-Encoder

Traffic prediction is an important technique for network link capacity planning. Supervised Variational Auto-Encoder (SVAE), a deep learning technique, suitable approach the The problem with SVAE that mean absolute value error (MAPE) decreases when correlation between feature vector and traffic probability distribution function (PDF) low. In this paper, we propose to increase by analytically obtaining values correlate PDF. Simulations are performed under conditions above demonstrate improvement in MAPE.