Research of a deep predictive coding model for unsupervised representation learning from video and audio data.