ValueError：层模型需要 21 个输入，但它收到 1 个输入张量

如何解决ValueError：层模型需要 21 个输入，但它收到 1 个输入张量

我是 Keras 的初学者。我正在关注此 example，我正在使用 Keras 训练二进制分类模型，其中输入数据是从 csv 中获取的结构化数据，但出现以下错误

ValueError: Layer model expects 21 input(s),but it received 1 input tensors. Inputs received: [<tf.Tensor 'ExpandDims:0' shape=(None,1) dtype=float64>]

在线

score = model.evaluate(x=test_labels,y=test_data,verbose=1)

我的代码如下

import os
import numpy as np
import pandas as pd
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.layers.experimental.preprocessing import Normalization

def dataframe_to_dataset(dataframe):
    dataframe = dataframe.copy()
    labels = dataframe.pop("label")
    labels = np.asarray(labels).astype('float32').reshape((-1,1))
    ds = tf.data.Dataset.from_tensor_slices((dict(dataframe),labels))
    ds = ds.shuffle(buffer_size=len(dataframe))
    return ds


os.chdir('datasets/accelerometer/accel-labeled/')
df = pd.read_csv("combined-all-labeled.csv",delimiter=',')

print(f"All size : {df.shape[0]}")

np.random.seed(23)
perm = np.random.permutation(df.index)
m = len(df.index)
train_end = int(.70 * m)
validate_end = int(.25 * m) + train_end

train_ds = dataframe_to_dataset(df.iloc[perm[:train_end]])
validate_ds = dataframe_to_dataset(df.iloc[perm[train_end:validate_end]])
test_df = df.iloc[perm[validate_end:]]

test_labels = test_df['label'].astype('float')
test_data = test_df.iloc[:,2:22].astype('float')
print(test_data.head(2))

print(f"Train Set size : {len(train_ds)}")
print(f"Validation Set size : {len(validate_ds)}")
print(f"Test Set size : {len(test_df)}")


def encode_numerical_feature(feature,name,dataset):
    # Create a Normalization layer for our feature
    normalizer = Normalization()

    # Prepare a Dataset that only yields our feature
    feature_ds = dataset.map(lambda x,y: x[name])
    feature_ds = feature_ds.map(lambda x: tf.expand_dims(x,-1))

    # Learn the statistics of the data
    normalizer.adapt(feature_ds)

    # Normalize the input feature
    encoded_feature = normalizer(feature)
    return encoded_feature


# Numerical features
x_mean = keras.Input(shape=(1,),name="x_mean")
x_median = keras.Input(shape=(1,name="x_median")
x_std_dev = keras.Input(shape=(1,name="x_stdev")
x_raw_min = keras.Input(shape=(1,name="x_raw_min")
x_raw_max = keras.Input(shape=(1,name="x_raw_max")
x_abs_min = keras.Input(shape=(1,name="x_abs_min")
x_abs_max = keras.Input(shape=(1,name="x_abs_max")

y_mean = keras.Input(shape=(1,name="y_mean")
y_median = keras.Input(shape=(1,name="y_median")
y_std_dev = keras.Input(shape=(1,name="y_stdev")
y_raw_min = keras.Input(shape=(1,name="y_raw_min")
y_raw_max = keras.Input(shape=(1,name="y_raw_max")
y_abs_min = keras.Input(shape=(1,name="y_abs_min")
y_abs_max = keras.Input(shape=(1,name="y_abs_max")

z_mean = keras.Input(shape=(1,name="z_mean")
z_median = keras.Input(shape=(1,name="z_median")
z_std_dev = keras.Input(shape=(1,name="z_stdev")
z_raw_min = keras.Input(shape=(1,name="z_raw_min")
z_raw_max = keras.Input(shape=(1,name="z_raw_max")
z_abs_min = keras.Input(shape=(1,name="z_abs_min")
z_abs_max = keras.Input(shape=(1,name="z_abs_max")

all_inputs = [
    x_mean,x_median,x_std_dev,x_raw_min,x_raw_max,x_abs_min,x_abs_max,y_mean,y_median,y_std_dev,y_raw_min,y_raw_max,y_abs_min,y_abs_max,z_mean,z_median,z_std_dev,z_raw_min,z_raw_max,z_abs_min,z_abs_max,]

# Numerical features
x_mean_encoded = encode_numerical_feature(x_mean,"x_mean",train_ds)
x_median_encoded = encode_numerical_feature(x_median,"x_median",train_ds)
x_std_dev_encoded = encode_numerical_feature(x_std_dev,"z_stdev",train_ds)
x_raw_min_encoded = encode_numerical_feature(x_raw_min,"x_raw_min",train_ds)
x_raw_max_encoded = encode_numerical_feature(x_raw_max,"x_raw_max",train_ds)
x_abs_min_encoded = encode_numerical_feature(x_abs_min,"x_abs_min",train_ds)
x_abs_max_encoded = encode_numerical_feature(x_abs_max,"x_abs_max",train_ds)

y_mean_encoded = encode_numerical_feature(y_mean,"y_mean",train_ds)
y_median_encoded = encode_numerical_feature(y_median,"y_median",train_ds)
y_std_dev_encoded = encode_numerical_feature(y_std_dev,train_ds)
y_raw_min_encoded = encode_numerical_feature(y_raw_min,"y_raw_min",train_ds)
y_raw_max_encoded = encode_numerical_feature(y_raw_max,"y_raw_max",train_ds)
y_abs_min_encoded = encode_numerical_feature(y_abs_min,"y_abs_min",train_ds)
y_abs_max_encoded = encode_numerical_feature(y_abs_max,"y_abs_max",train_ds)

z_mean_encoded = encode_numerical_feature(z_mean,"z_mean",train_ds)
z_median_encoded = encode_numerical_feature(z_median,"z_median",train_ds)
z_std_dev_encoded = encode_numerical_feature(z_std_dev,train_ds)
z_raw_min_encoded = encode_numerical_feature(z_raw_min,"z_raw_min",train_ds)
z_raw_max_encoded = encode_numerical_feature(z_raw_max,"z_raw_max",train_ds)
z_abs_min_encoded = encode_numerical_feature(z_abs_min,"z_abs_min",train_ds)
z_abs_max_encoded = encode_numerical_feature(z_abs_max,"z_abs_max",train_ds)

all_features = layers.concatenate(
    [
        x_mean_encoded,x_median_encoded,x_std_dev_encoded,x_raw_min_encoded,x_raw_max_encoded,x_abs_min_encoded,x_abs_max_encoded,y_mean_encoded,y_median_encoded,y_std_dev_encoded,y_raw_min_encoded,y_raw_max_encoded,y_abs_min_encoded,y_abs_max_encoded,z_mean_encoded,z_median_encoded,z_std_dev_encoded,z_raw_min_encoded,z_raw_max_encoded,z_abs_min_encoded,z_abs_max_encoded,]
)

x = layers.Dense(32,activation="relu")(all_features)
x = layers.Dropout(0.5)(x)
output = layers.Dense(1,activation="sigmoid")(x)
model = keras.Model(all_inputs,output)
model.compile("adam","binary_crossentropy",metrics=["accuracy"])

# `rankdir='LR'` is to make the graph horizontal.
keras.utils.plot_model(model,show_shapes=True,rankdir="LR")

# model.summary()

# Train model
model.fit(train_ds,epochs=1,validation_data=validate_ds)

score = model.evaluate(x=test_labels,verbose=1)
#
print('Test loss: ',score[0])
print('Test accuracy: ',score[1])

CSV 如下所示

group_timestamp,label,x_mean,x_stdev,y_stdev,z_stdev,z_abs_max
2017-05-02 17:35:20,-8.40793368,-8.432378499999999,0.0812278134949539,-8.632295,-8.24563,8.24563,8.632295,-180900768.0,0.0,180900768.0,180900768.0
2017-05-02 17:19:40,1.96025263,1.9716251,0.0710845152401064,1.816002,2.112883,180900752.0,180900752.0
...

我该如何解决这个问题？

---

此外，欢迎任何其他输入修复/改进代码/模型！

解决方法

改进建议，查看174行命令：