sunkevin1214 / codes Goto Github PK

During the creation of the initial population, the pseudocode in the article (Algorithm 2) and the code you put Github look different.

In the article, while creating an individual, a convolutional or pooling layer is added depending on a possibility, after this stage is finished, the addition of the fully connected layer is started. However, according to the code on Github (individual.py -> init_one_individual()), regardless of probability , layers are added sequentially as
convolutional, pooling and fully connected.

Is there something wrong here? I would be glad if you explain.

You can find the related image attached.

For some reason the EDCNN code is still having Errors:

1.Error:
File ".../PycharmProjects/main.py", line 47, in
restart_evolve(0.9, 0.05, 0.2, 0.05, pop_size, None, None, None, None, number_of_channel, epochs, batch_size, train_data_length, validate_data_length, total_generation_number, eta)
File ".../PycharmProjects/main.py", line 15, in restart_evolve
gen_no, pops, _= load_population()
File "...users\PycharmProjects\utils.py", line 19, in load_population
with open(path, 'rb') as file_handler:

2. Also the error of missing pops.dat file:

FileNotFoundError: [Errno 2] No such file or directory: '...\PycharmProjects/pops.dat'