This report mainly includes two contents:

I. Sharing of literature review on classification of EEG signals using deep learning techniques in motor imagination

Three questions are discussed by analyzing and researching 89 literatures (preprocessing, input form, deep learning strategy, network structure and performance evaluation)：

1) Is preprocessing necessary for deep learning technology?

2）What forms of input are best for deep learning techniques?

3）What are the current research trends in deep learning technologies?

These questions are finally answered in the discussion section：

1）Preprocessing may increase the computational cost, and it is recommended to use the original signal to build the deep learning network.

2）Topological input is more effective for CNN network. Of course, it is recommended to use raw data.

3）CNN is used to construct neural network and increase the interpretability of deep learning classification EEG.

Ⅱ. Classification results of age-related electroencephalogram (EEG) of vibration-tactile stimulation.

A smaller sub-frequency band is set, and the optimal time-frequency-space features under each sub-frequency band are extracted to improve the accuracy. The current results showed a 12 percent increase in accuracy in older adults and an 8 percent increase in younger adults compared to the original paper.