Neural networks vary based on the number of hidden layers and the activation functions used in each layer. Here are some of the common types of neural networks:

• Deep neural networks: Networks with more than one hidden layer.

• CNN: Commonly used in computer-vision-related learning problems. The CNN hidden layer uses convolution functions as the activation function.
• Recurrent neural networks: Commonly used in problems related to natural language processing.

Current projects/research in the field of improving neural networks in mobile devices include the following:

• MobileNet
• MobileNet V2
• MNasNet—implementing reinforcement learning in mobile devices