Training.py

import numpy as np
import pandas as pd
from tensorflow import keras
from tensorflow.keras import layers
import HelperFunctions

emnist_train = pd.read_csv('data/emnist-balanced-train.csv')
emnist_test = pd.read_csv('data/emnist-balanced-test.csv')

print(f'Shape of emnist_train: {emnist_train.shape}\nShape of emnist_test: {emnist_test.shape}')

# %%


emnist_train_x, emnist_train_y = HelperFunctions.emnist_preprocess(emnist_train)
emnist_test_x, emnist_test_y = HelperFunctions.emnist_preprocess(emnist_test)
del emnist_train
del emnist_test

print(f'{emnist_train_x.shape}\n{emnist_train_y.shape}\n{emnist_test_x.shape}\n{emnist_test_y.shape}')


# %%
BATCH_SIZE = 128
EPOCHS = 20

def cnn2d():
    x_train = HelperFunctions.convert_1d_to_2d(emnist_train_x)
    input_shape = (28, 28, 1)
    model = keras.Sequential(
        [
            keras.Input(shape=input_shape),
            layers.Conv2D(64, kernel_size=(3, 3), padding="same", activation="relu"),
            layers.MaxPooling2D(pool_size=(2, 2)),
            layers.BatchNormalization(),
            layers.Dropout(0.3),
            layers.Conv2D(128, kernel_size=(3, 3), padding="same", activation="relu"),
            layers.MaxPooling2D(pool_size=(2, 2)),
            layers.BatchNormalization(),
            layers.Dropout(0.3),
            layers.Conv2D(256, kernel_size=(3, 3), padding="same", activation="relu"),
            layers.MaxPooling2D(pool_size=(2, 2)),
            layers.BatchNormalization(),
            layers.Flatten(),
            layers.Dense(128, activation="relu"),
            layers.Dense(64, activation="relu"),
            layers.Dropout(0.3),
            layers.Dense(47, activation="softmax"),
        ]
    )

    model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=["accuracy"])
    model.fit(x_train, emnist_train_y, batch_size=BATCH_SIZE, epochs=EPOCHS, validation_split=0.1)
    return model
    

def cnn1d_3layer():
    input_shape = 784
    train = np.expand_dims(emnist_train_x, axis=2)
    model = keras.Sequential([
        layers.Conv1D(64, 8, padding='same', input_shape=(input_shape, 1)),
        layers.MaxPool1D(2),
        layers.BatchNormalization(),
        layers.Conv1D(128, 7, padding='same'),
        layers.MaxPool1D(2),
        layers.Dropout(0.2),
        layers.BatchNormalization(),
        layers.Conv1D(128, 6, padding='same'),
        layers.MaxPool1D(2),
        layers.BatchNormalization(),
        layers.Flatten(),
        layers.Dense(256, activation="relu"),
        layers.Dense(128, activation="relu"),
        layers.Dropout(0.2),
        layers.Dense(47, activation="softmax"),
    ])
    
    model.compile(loss="mse", optimizer="adam", metrics=["accuracy"])
    model.fit(train, emnist_train_y, batch_size=BATCH_SIZE, epochs=EPOCHS, validation_split=0.2)
    return model


# %%
if __name__ == "__main__":
    path = "models"
    name = "2dmodel"
    model = cnn2d()
    HelperFunctions.save_model(model, path, name)