Continuous speech emotion recognition from audio segments with supervised learning and reinforcement learning approaches