How to acquire CHAMP module.

[sekigh]

I am interested in running Annex.ipynb on jupyter lab but fail in the middle because I can not locate CHAMP module. This CHAMP looks unrelated with champ on the github https://github.com/wweir827/CHAMP. Am I right? How do I get the right Champ ? Thank you.

[laurentperrinet]

hi!

indeed, this repository was in "private" mode. It is now fixed, you can get it @ https://github.com/VictorBoutin/CHAMP

(it has been a while since I have not looked at that code - hope everything still runs fine! :-) )

cheers,
Laurent

[sekigh]

Hi,

Thank you for your quick response. Yes, I can install CHAMP but I find that Raw DataBase is missing. The code says an error as follows:

FileNotFoundError: [Errno 2] No such file or directory: '/tmp/database/Face_DataBase/Raw_DataBase'

Is it possible to share that data for code sanity check purpose ?

Thank you.

[laurentperrinet]

you certainly miss the data which is normally imported through torchvision. when using LoadDatabe sure to set download=True

[VictorBoutin]

The database we are using is the AT&T database that you could find following this link: http://cam-orl.co.uk/facedatabase.html <http://cam-orl.co.uk/facedatabase.html> Victor

…

Le 30 sept. 2020 à 14:41, sekigh ***@***.***> a écrit : Hi, Thank you for your quick response. Yes, I can install CHAMP but I find that Raw DataBase is missing. The code says an error as follows: FileNotFoundError: [Errno 2] No such file or directory: '/tmp/database/Face_DataBase/Raw_DataBase' Is it possible to share that data for code sanity check purpose ? Thank you. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEXHOGKYAQWHR6QM34FIOITSIMRORANCNFSM4R6UILPQ>.

[sekigh]

Hi Victor,

Thank you for your response. Unfortunately, the link to face data, shown on the page of ATT http://cam-orl.co.uk/facedatabase.html is not available any longer. It is disconnected or removed. It is because the ATT link page looks old (created in 2001. The data are now missing.

Hi Laurent,
According to Victor, these data does not look to be included in torchvision. There is CelebA face in torchvision but they are of course different from data you used in the annex. I changed the LoadData argument with download=True, but I failed.

Is there any suggestion to retrieve the data back for code check ?
I really want to see how the SHL is incorporated in neural network like CNN.

Best regards,

Hiroshi Sekiguchi

[VictorBoutin]

Here is the database, https://cloud.int.univ-amu.fr/index.php/s/9EqBTJLdi2WR37S <https://cloud.int.univ-amu.fr/index.php/s/9EqBTJLdi2WR37S> Victor

…

Le 30 sept. 2020 à 15:32, sekigh ***@***.***> a écrit : Hi Victor, Thank you for your response. Unfortunately, the link to face data, shown on the page of ATT http://cam-orl.co.uk/facedatabase.html <http://cam-orl.co.uk/facedatabase.html> is not available any longer. It is disconnected or removed. It is because the ATT link page looks old (created in 2001. The data are now missing. Hi Laurent, According to Victor, these data does not look to be included in torchvision. There is CelebA face in torchvision but they are of course different from data you used in the annex. I changed the LoadData argument with download=True, but I failed. Is there any suggestion to retrieve the data back for code check ? I really want to see how the SHL is incorporated in neural network like CNN. Best regards, Hiroshi Sekiguchi — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEXHOGKJSNX432AJEDEMFCDSIMXPFANCNFSM4R6UILPQ>.

[sekigh]

Hi Victor,

Thank you for url to database. I will import them to the codes.

Best regards,

Hiroshi

[sekigh]

Hi,
I am in the situation to debug codes.
It succeeded in reading Raw_DataBase and executed computation up to the 3rd-2nd line to the last of a part of the codes attached below.;

dico_mask = L1_mask.TrainLayer(
Filtered_L_TrSet, eta=eta, eta_homeo=eta_homeo, nb_epoch=nb_epoch, seed=seed)

Then the last statement of a part of codes;

SaveNetwork(Network=L1_mask, saving_path=ffname)

faces an error as follows: FileNotFoundError: [Errno 2] No such file or directory: 'cache_dir_CNN/CHAMP_low_None.pkl'

I looked into the codes carefully and found no statement to create a folder/file for "cache_dir_CNN/CHAMP_low_None.pkl" before. So it is reasonable to have an error. I am not sure how it ends up with such as this. Any comments are welcome.

---

< A part of the codes having an error: >

sys.path.append('/home/sekigh/Windowsfolder/HULK/CHAMP')

from CHAMP.DataLoader import LoadData
from CHAMP.DataTools import LocalContrastNormalization, FilterInputData, GenerateMask
from CHAMP.Monitor import DisplayDico, DisplayConvergenceCHAMP, DisplayWhere

import os
datapath = os.path.join("/tmp", "database")
path = os.path.join(datapath, "Face_DataBase/Raw_DataBase")
TrSet, TeSet = LoadData('Face', path, decorrelate=False, resize=(65, 65))

MP Parameters

nb_dico = 20
width = 9
dico_size = (width, width)
l0 = 20
seed = 42

Learning Parameters

eta = .05
nb_epoch = 500

TrSet, TeSet = LoadData('Face', path, decorrelate=False, resize=(65, 65))
N_TrSet, _, _, _ = LocalContrastNormalization(TrSet)
Filtered_L_TrSet = FilterInputData(
N_TrSet, sigma=0.25, style='Custom', start_R=15)

mask = GenerateMask(full_size=(nb_dico, 1, width, width), sigma=0.8, style='Gaussian')

from CHAMP.CHAMP_Layer import CHAMP_Layer

from CHAMP.DataTools import SaveNetwork, LoadNetwork
homeo_methods = ['None', 'HAP']

for homeo_method, eta_homeo in zip(homeo_methods, [0., 0.0025]):
ffname = 'cache_dir_CNN/CHAMP_low_' + homeo_method + '.pkl'
try:
L1_mask = LoadNetwork(loading_path=ffname)
except:
L1_mask = CHAMP_Layer(l0_sparseness=l0, nb_dico=nb_dico,
dico_size=dico_size, mask=mask, verbose=1)
dico_mask = L1_mask.TrainLayer(
Filtered_L_TrSet, eta=eta, eta_homeo=eta_homeo, nb_epoch=nb_epoch, seed=seed)
SaveNetwork(Network=L1_mask, saving_path=ffname)

< End of a part of the codes>