Machine learning for sensory data analytics

Analyzing an enormous amount of sensory data generated by internet of things (IoT) devices are useful for monitoring and prediction. However, traditional data analysis methods are no longer applicable to sensory data due to its large volume and high complexity. At present, machine learning plays an important role in sensory data analytics. This chapter briefly introduces deep learning models and their applications in IoT, such as convolutional neural networks, recurrent neural networks, and long short-term memory networks. Then, this chapter introduces deep reinforcement learning, analyzes its advantages, and describes different models that are applied in IoT sensory data analytics, which contains value-based methods, policy-based methods, and actor-critic methods. Since a single machine cannot analyze massive and complex sensory data effectively, this chapter gives an overview of distributed machine learning along with its challenges and related knowledge.