TensorFlow 不兼容形状二进制分类

Question

I have a pandas dataframe of features and samples, and a single series with binary category (0 or 1) values. With that I'm trying to train a neural network, but I am getting the error:

TensorFlow incompatible shapes binary classification

Here is a summary of the code:

X_train, X_test, y_train, y_test = train_test_split(df_x, series_y, random_state=1, test_size=0.25)

best_weight_path = 'best_weights.hdf5'

x = df_x.to_numpy()
y = series_y.to_numpy()

numpy_x_train = X_train.to_numpy()
numpy_y_train = y_train.to_numpy()
numpy_x_test = X_test.to_numpy()
numpy_y_test = y_test.to_numpy()
model = Sequential()
model.add(Dense(20, input_dim=x.shape[1], activation='relu'))
model.add(Dense(10, activation='relu'))
model.add(Dense(2, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam')
monitor = EarlyStopping(monitor='val_loss', min_delta=1e-3, patience=5, verbose=1, mode='auto')
checkpointer = ModelCheckpoint(filepath=best_weight_path, verbose=0, save_best_only=True)
model.fit(x, y, validation_data=(numpy_x_test, numpy_y_test), callbacks=[monitor, checkpointer], verbose=0, epochs=1000)

ValueError: Shapes (None, 1) and (None, 2) are incompatible

Shouldn't the last dense layer have 2 units as there are two possible outcomes, so where is the shape (None, 1) coming from?

Answer 1

The problem is related to the correct choice of an appropriate loss function according to the format of your labels. you have 2 possibilities when using softmax in classification task:

1 possibility: if you have 1D integer encoded target, you can use sparse_categorical_crossentropy as loss function ( this seems to be your case )

n_class = 2
n_features = 100
n_sample = 1000

X = np.random.randint(0,10, (n_sample,n_features))
y = np.random.randint(0,n_class, n_sample)

inp = Input((n_features,))
x = Dense(128, activation='relu')(inp)
out = Dense(n_class, activation='softmax')(x)

model = Model(inp, out)
model.compile(loss='sparse_categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
history = model.fit(X, y, epochs=3)

2 possibility: if you have one-hot encoded your target in order to have 2D shape (n_samples, n_class), you can use categorical_crossentropy

n_class = 2
n_features = 100
n_sample = 1000

X = np.random.randint(0,10, (n_sample,n_features))
y = pd.get_dummies(np.random.randint(0,n_class, n_sample)).values

inp = Input((n_features,))
x = Dense(128, activation='relu')(inp)
out = Dense(n_class, activation='softmax')(x)

model = Model(inp, out)
model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])
history = model.fit(X, y, epochs=3)