Description
I am using Tensorflow 2.7 and trying to reproduce the results with the code in https://modal-python.readthedocs.io/en/latest/content/examples/Keras_integration.html. Instead of importing KerasClassifier from Keras Wrappers that through a deprecation warning, I have installed SciKeras and imported the KerasClassifier. The following is the code:
from tensorflow import keras
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPooling2D
from scikeras.wrappers import KerasClassifier
# build function for the Keras' scikit-learn API
def create_keras_model():
"""
This function compiles and returns a Keras model.
Should be passed to KerasClassifier in the Keras scikit-learn API.
"""
model = Sequential()
model.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=(28, 28, 1)))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(10, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adadelta', metrics=['accuracy'])
return model
classifier = KerasClassifier(create_keras_model)
import numpy as np
from tensorflow.keras.datasets import mnist
# read training data
(X_train, y_train), (X_test, y_test) = mnist.load_data()
X_train = X_train.reshape(60000, 28, 28, 1).astype('float32') / 255
X_test = X_test.reshape(10000, 28, 28, 1).astype('float32') / 255
y_train = keras.utils.to_categorical(y_train, 10)
y_test = keras.utils.to_categorical(y_test, 10)
# assemble initial data
n_initial = 100
initial_idx = np.random.choice(range(len(X_train)), size=n_initial, replace=False)
X_initial = X_train[initial_idx]
y_initial = y_train[initial_idx]
# generate the pool
# remove the initial data from the training dataset
X_pool = np.delete(X_train, initial_idx, axis=0)
y_pool = np.delete(y_train, initial_idx, axis=0)
from modAL.models import ActiveLearner
# initialize ActiveLearner
learner = ActiveLearner(
estimator=classifier,
X_training=X_initial, y_training=y_initial,
verbose=1
)
I have taken 100 initial data samples as aforementioned. But running the code showed me this: That is, the data has been divided into 4 batches whereas in the documentation example, we see the data is divided into "n_initial" batches.
4/4 [==============================] - 11s 32ms/step - loss: 2.2845 - accuracy: 0.1200
The I ran this part of the code where i wanted to generate 10 queries and each time I take 100 instances.
n_queries = 10
for idx in range(n_queries):
print('Query no. %d' % (idx + 1))
query_idx, query_instance = learner.query(X_pool, n_instances=100, verbose=0)
learner.teach(
X=X_pool[query_idx], y=y_pool[query_idx], only_new=True,
verbose=1
)
# remove queried instance from pool
X_pool = np.delete(X_pool, query_idx, axis=0)
y_pool = np.delete(y_pool, query_idx, axis=0)
And I got this: The data has been divided into 4 batches for each query but in the documentation, it was divided into 100 batches (100/100).
Query no. 1
4/4 [==============================] - 0s 6ms/step - loss: 2.3089 - accuracy: 0.0700
Query no. 2
4/4 [==============================] - 1s 7ms/step - loss: 2.2615 - accuracy: 0.2400
Query no. 3
4/4 [==============================] - 0s 6ms/step - loss: 2.3040 - accuracy: 0.0800
Query no. 4
4/4 [==============================] - 0s 7ms/step - loss: 2.2629 - accuracy: 0.1300
Query no. 5
4/4 [==============================] - 0s 6ms/step - loss: 2.3116 - accuracy: 0.1300
Query no. 6
4/4 [==============================] - 0s 6ms/step - loss: 2.3290 - accuracy: 0.0900
Query no. 7
4/4 [==============================] - 0s 6ms/step - loss: 2.3691 - accuracy: 0.0300
Query no. 8
4/4 [==============================] - 0s 6ms/step - loss: 2.2598 - accuracy: 0.2500
Query no. 9
4/4 [==============================] - 0s 7ms/step - loss: 2.2914 - accuracy: 0.1000
Query no. 10
4/4 [==============================] - 0s 6ms/step - loss: 2.2839 - accuracy: 0.1500
Requesting assistance in this regard.