To receive notifications about scheduled maintenance, please subscribe to the mailing-list gitlab-operations@sympa.ethz.ch. You can subscribe to the mailing-list at https://sympa.ethz.ch

README.md 1.29 KB
Newer Older
Ard Kastrati's avatar
Ard Kastrati committed
1
2
# DL-Project

3
Predicting eye gaze with DL
4
5
6
7
8
9
10
11
    Projects that we try: CNN, InceptionTime, EEGNet, DeepEye



## Leonhard Settings

Please use the command

zpgeng's avatar
zpgeng committed
12
13
14
15
`module load gcc/6.3.0 python_gpu/3.8.5 hdf5/1.10.1`

before training the model. Please make sure that the Tensorflow version should be 2.x.

okiss's avatar
okiss committed
16
## model configuration
okiss's avatar
okiss committed
17
18
19
20
21
Please configure the config.py file correctly before running the main.py file  
config['data_dir']    : indicates the directory where you stored the data  
config['model']       : indicates the model you want to use, choose between 'cnn', 'eegnet', 'inception', 'xception' or'deepeye'  
config['downsampled'] : True if you want to use 125 data points per second instead of 500. Default is False  
config['split']       : True if you want to run a clustered version of the model, please keep it to False as the clustered version is inneficient  
okiss's avatar
okiss committed
22
23


zpgeng's avatar
zpgeng committed
24
25
26
27
28
29
30
31
32
33
34
35
36
## deepEye3 tuning

nb_filter: [32, 64]

depth: [9, 12, 20]

kernel_size:[40, 20]

residual_jump: [3, 4]

Large depth causes overfitting, same for the number of filters. Kernel size seems to have tiny affect on validation. Residual jump for 4 (i.e. `depth % (res_jump) == (res_jump - 1)`) is not so good in our task, but I think it would be useful for future tasks.

The best setting is **nb_filter == 32, depth == 9, kernel_size == 40, res_jump == 3**
37