i'm working on a university project, my purpose is to create a federated CNN capable of training and testing images of different malware, collected in different folders (for each family).

This is the perfect example of what I would like to do.

I have watched several tutorials on creating federated systems all based on MNIST, but I cannot understand how to apply a federated approach to this code.

I would like to know if it's possible to convert this code using a federated approach.

Here the code i'm using...

path_root = "\Malware_Classification\data\malimg_paper_dataset_imgs\\"

from keras.preprocessing.image import ImageDataGenerator
batches = ImageDataGenerator().flow_from_directory(directory=path_root, target_size=(64,64), batch_size=10000)

imgs, labels = next(batches)

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(imgs/255.,labels, test_size=0.3)

import keras
from keras.models import Sequential, Input, Model
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras.layers.normalization import BatchNormalization
num_classes = 25

def malware_model():
    Malware_model = Sequential()
    Malware_model.add(Conv2D(30, kernel_size=(3, 3),
    Malware_model.add(MaxPooling2D(pool_size=(2, 2)))
    Malware_model.add(Conv2D(15, (3, 3), activation='relu'))
    Malware_model.add(MaxPooling2D(pool_size=(2, 2)))
    Malware_model.add(Dense(128, activation='relu'))
    Malware_model.add(Dense(50, activation='relu'))
    Malware_model.add(Dense(num_classes, activation='softmax'))
    Malware_model.compile(loss='categorical_crossentropy', optimizer = 'adam', metrics=['accuracy'])
    return Malware_model

Malware_model = malware_model()

y_train_new = np.argmax(y_train, axis=1)

from sklearn.utils import class_weight

class_weights = class_weight.compute_class_weight('balanced',

scores = Malware_model.evaluate(X_test, y_test)
print('Final CNN accuracy: ', scores[1])
New contributor
TheNewOne is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.