# 提取特定参数的数据
x_downsampled = x[downsampled_indices][:, :, selected_params]
y_downsampled = y[downsampled_indices]

# 数据标准化
scaler = MinMaxScaler()
n_samples, time_steps, n_features = x_downsampled.shape
x_downsampled_flat = x_downsampled.reshape(n_samples, -1)
x_downsampled_flat = scaler.fit_transform(x_downsampled_flat)
x_downsampled = x_downsampled_flat.reshape(n_samples, time_steps, n_features)

# 划分训练集和测试集
x_train, x_test, y_train, y_test = train_test_split(x_downsampled, y_downsampled, test_size=0.2, random_state=42)

# 定义CNN-LSTM模型
model = Sequential()
model.add(Conv1D(filters=64, kernel_size=3, activation='relu', input_shape=(time_steps, n_features)))
model.add(MaxPooling1D(pool_size=2))
model.add(LSTM(64, return_sequences=True))
model.add(LSTM(64, return_sequences=False))
model.add(Dense(4, activation='softmax'))  # 假设有4个类别的输出

model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

# 打印模型摘要以检查层次结构
model.summary()

# 训练模型
model.fit(x_train, y_train, epochs=10, batch_size=64, validation_split=0.2)

feature_extractor = Model(inputs=model.input, outputs=model.layers[3].output)
# 提取特征
x_train_features = feature_extractor.predict(x_train)
x_test_features = feature_extractor.predict(x_test)

# 使用t-SNE对特征进行降维
tsne = TSNE(n_components=2, random_state=42)
features_2d = tsne.fit_transform(np.vstack((x_train_features, x_test_features)))

# 可视化
plt.figure(figsize=(10, 6))
plt.scatter(features_2d[:x_train_features.shape[0], 0], features_2d[:x_train_features.shape[0], 1], c=y_train, cmap='viridis', label='Training Data')
plt.scatter(features_2d[x_train_features.shape[0]:, 0], features_2d[x_train_features.shape[0]:, 1], c=y_test, cmap='viridis', label='Test Data')
plt.colorbar(label='Classes')
plt.title('t-SNE Visualization of CNN-LSTM Features')
plt.xlabel('Dimension 1')
plt.ylabel('Dimension 2')
plt.legend()
plt.show()

Then it keeps throwing errors like this one：ValueError: The layer sequential_4 has never been called and thus has no defined input.
I really don’t know how to solve it.

# 创建一个新模型，以获取中间层的输出
intermediate_layer_model = tf.keras.Model(inputs=model.input,
                                          outputs=[model.get_layer('dense_1').output])

# 获取中间层输出
intermediate_output = intermediate_layer_model.predict(x_downsampled)


# 对原始输入和中间层输出进行 t-SNE 降维
x_downsampled_flat = x_downsampled.reshape(n_samples, -1)
tsne = TSNE(n_components=2, random_state=42)

x_tsne = tsne.fit_transform(x_downsampled_flat)
intermediate_output_tsne = tsne.fit_transform(intermediate_output)


def plot_tsne(data, labels, title):
    plt.figure(figsize=(8, 8))
    for label in np.unique(labels):
        indices = np.where(labels == label)
        plt.scatter(data[indices, 0], data[indices, 1], label=f'Class {label}', alpha=0.6)
    plt.legend()
    plt.title(title)
    plt.show()

# 可视化原始输入数据的 t-SNE 降维结果
plot_tsne(x_tsne, y_downsampled, 't-SNE of Original Input Data')

# 可视化中间层输出数据的 t-SNE 降维结果
plot_tsne(intermediate_output_tsne, y_downsampled, 't-SNE of Dense Layer Output Data')

I just want to extract the data from the output layer, but it is unsuccessful.I want to know why the error is happening.maybe i know little about what i’m doing.