numpy无法转化为tensor

Question

这段代码报valid_freq=1，numpy无法转化为tensor，这里X集是numpy格式

import tensorflow as tf
import numpy as np
import wfdb
import pandas as pd
import ast
from matplotlib import pyplot as plt
import torch
from sklearn.preprocessing import LabelEncoder


def load_raw_data(df, sampling_rate, path):
    if sampling_rate == 100:
        data = [wfdb.rdsamp(path+f) for f in df.filename_lr]
    else:
        data = [wfdb.rdsamp(path+f) for f in df.filename_hr]
    data = np.array([signal for signal, meta in data])
    return data

path = ''
sampling_rate = 100

# load and convert annotation data
Y = pd.read_csv('ptbxl_database.csv', index_col='ecg_id')
Y.scp_codes = Y.scp_codes.apply(lambda x: ast.literal_eval(x))
# Load raw signal data
X = load_raw_data(Y, sampling_rate, path)

# Load scp_statements.csv for diagnostic aggregation
agg_df = pd.read_csv('scp_statements.csv', index_col=0)
agg_df = agg_df[agg_df.diagnostic == 1]

def aggregate_diagnostic(y_dic):
    tmp = []
    for key in y_dic.keys():
        if key in agg_df.index:
            tmp.append(agg_df.loc[key].diagnostic_class)
    return list(set(tmp))

# Apply diagnostic superclass
Y['diagnostic_superclass'] = Y.scp_codes.apply(aggregate_diagnostic)

# Split data into train and test
test_fold = 10
# Train
X_train = X[np.where(Y.strat_fold != test_fold)]
y_train = Y[(Y.strat_fold != test_fold)].diagnostic_superclass
# Test
X_test = X[np.where(Y.strat_fold == test_fold)]
y_test = Y[Y.strat_fold == test_fold].diagnostic_superclass
# tenserflow 训练
print(X_test , y_test)
print(type(y_test))
print(type(X_test))
# <class 'pandas.core.series.Series'>
# <class 'numpy.ndarray'>

#该数据类型
X_test=torch.from_numpy(X_test)
X_train=torch.from_numpy(X_train)

# artificial intelligence and the training
model = tf.keras.models.Sequential([
    # 卷积层
    tf.keras.layers.Conv1D(filters = 96,
                           kernel_size = 3,
                           padding = "same"),
    tf.keras.layers.BatchNormalization(),
    tf.keras.layers.Activation("relu"),
    tf.keras.layers.AveragePooling1D(pool_size = 4,
                                     padding = "valid"),
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128,activation = "sigmoid",
                          kernel_regularizer = tf.keras.regularizers.l1()),
    tf.keras.layers.Dense(5, activation = "sigmoid",
                          kernel_regularizer = tf.keras.regularizers.l2())
])

model.compile(optimizer="nadam",
              loss = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
              metrics = ["sparse_categorical_accuracy"])

model.fit(X_train,y_train,
          batch_size = 64,
          epochs = 10000,
          validation_data = (X_test, y_test),
          validation_freq = 1)

history = model.fit(X_train,y_train,
          batch_size = 64,
          epochs = 10000,
          validation_data = (X_test, y_test))
model.summary()

# 可视化
acc=history.history['sparse_categorical_accuaracy']
val_acc = history.history['val_sparse_categorical_accuracy']
loss=history.history['loss']
val_loss = history.history['val_loss']

plt.subplot(1,2,1)
plt.plot(acc, label='Training Accuracy')
plt.plot(val_acc, label='Validation Accuracy')
plt.title('Training and Validation Accurcy')
plt.legend()

plt.subplot(1,2,1)
plt.plot(loss, label='Training Loss')
plt.plot(val_loss, label='Validation Loss')
plt.title('Training and Validation Loss')
plt.legend()
plt.show()

Answer 1

引自免费微信小程序：皆我百晓生

你遇到的问题是在将numpy数组转换为TensorFlow的tensor时，发现数据类型不能被正确转换。你的X_test和X_train是numpy数组，你试图将它们转换为TensorFlow的tensor。但是，你的目标变量y_test和y_train是Pandas的Series类型，不能直接作为模型的标签输入。你需要将它们转换为numpy数组。以下是解决这个问题的步骤：

首先，确保你的y_test和y_train是numpy数组格式。你可以使用.values或者.to_numpy()方法将Pandas Series转换为numpy数组。修改如下：

y_train = Y[(Y.strat_fold != test_fold).values].diagnostic_superclass
y_test = Y[Y.strat_fold == test_fold].diagnostic_superclass.values

然后，你可以尝试再次将numpy数组转换为TensorFlow的tensor。这里需要注意的是，由于你的目标变量是分类问题（具有多个类别），你需要确保你的标签是整数类型（integer）。如果你的标签是字符串类型（string），你需要使用.astype(int)将其转换为整数类型。如果你的类别标签已经是整数类型，则可以跳过这一步。然后你可以使用tf.convert_to_tensor()函数将numpy数组转换为TensorFlow的tensor。修改如下：

y_train = y_train.astype(int)  # 如果你的类别标签是字符串类型，则需要此行转换
y_test = y_test.astype(int)  # 同上

X_train = tf.convert_to_tensor(X_train)
X_test = tf.convert_to_tensor(X_test)

另外，你在训练模型时使用的损失函数是SparseCategoricalCrossentropy，这意味着你的目标变量应该是整数类型的类别标签。请确保你的处理方式是正确的。如果你在处理过程中遇到其他问题，请随时向我询问。