检查输入时出错：预期 lstm_input 有 3 个维度，但得到形状为 (160, 1000) 的数组

Question

当我尝试将 RNN 模型与我的数据集一起使用时出现此错误。这是我的模型和数据集的精简版，但它产生的错误完全相同 - dropbox.com/sh/1a2guzsk4afp94r/AACkOQ1UibLZAhzGoyZJtVcqa?dl=0

训练数据由每 1000 个浮点数长的浮点数列表组成。例如，

 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.55 0.2  0.5  0.85 0.8  0.2
 0.2  1.   0.2  0.2  0.5  0.75 0.3  0.5  0.5  0.3  0.85 0.65 0.15 0.4
 0.3  0.6  0.05 0.15 0.85 0.5  0.45 0.45 0.05 0.25 0.5  0.45 0.05 0.1
 0.45 0.9  0.35 0.35 0.65 0.15 0.45 0.3  0.3  0.6  0.2  0.2  0.05 0.55
 0.45 0.75 0.5  0.5  1.   0.5  1.   0.6  0.45 0.05 0.45 0.4  0.05 0.05
 0.8  0.05 0.85 0.85 0.8  0.7  0.9  0.65 0.2  1.   0.3  0.85 0.8  0.7
 0.95 0.2  0.7  0.95 0.95 0.2  0.2  0.25 0.6  0.7  0.3  0.25 0.15 0.2
 0.7  0.15 0.5  0.35 0.1  0.15 0.2  0.2  0.5  0.2  0.65 0.8  0.15 0.6
 0.2  0.2  0.2  0.6  0.65 0.05 0.3  0.8  0.7  0.05 0.65 0.3  0.8  0.7
 0.05 0.85 0.65 0.65 0.45 0.45 0.65 0.75 0.85 0.8  0.65 0.15 0.25 0.65
 0.2  0.9  0.5  0.45 0.2  1.   0.9  0.8  0.6  0.05 0.5  0.25 0.85 0.6
 0.75 0.85 1.   0.6  0.1  0.25 0.4  0.4  0.25 0.85 0.85 0.05 0.2  0.45
 0.3  0.45 0.2  0.5  1.   0.65 0.1  0.4  0.7  0.05 0.05 1.   0.45 0.1
 0.85 0.25 0.4  0.05 0.5  1.   0.55 1.   0.6  0.3  0.15 0.8  0.9  0.5
 1.   0.4  0.4  0.85 0.9  0.3  0.45 0.35 0.75 0.9  0.6  0.05 0.55 0.2
 0.6  0.5  0.1  0.8  0.45 0.45 0.1  0.85 0.9  0.8  0.25 0.5  0.7  0.05
 0.45 0.3  0.9  0.5  0.45 0.65 0.7  0.2  0.05 1.   0.6  0.9  0.1  0.1
 0.85 0.25 0.2  0.5  0.4  0.8  0.7  0.6  0.4  0.7  0.3  0.3  0.5  0.65
 0.8  0.8  0.25 0.25 0.6  0.65 0.9  0.7  0.2  0.2  0.2  0.45 0.8  0.9
 0.6  0.95 0.45 0.5  0.4  0.8  0.2  0.25 0.05 0.1  0.8  0.4  0.45 0.1
 0.85 0.15 0.65 0.5  0.5  0.5  0.55 0.05 0.5  1.   0.5  0.2  0.25 0.85
 0.85 0.05 0.65 0.2  0.05 0.1  0.45 0.9  0.1  0.15 0.6  0.65 0.75 0.75
 0.5  0.2  0.35 0.75 0.75 0.35 0.35 0.85 0.45 0.15 0.35 0.85 0.5  0.6
 0.05 0.5  0.5  0.25 0.7  0.15 0.8  0.45 0.85 0.7  0.45 0.85 0.4  0.1
 0.6  0.7  0.05 0.1  0.15 0.85 0.9  0.5  0.05 0.45 0.75 0.75 0.5  0.15
 0.55 0.45 0.85 0.5  0.85 0.75 0.6  0.2  0.5  0.5  0.9  0.7  0.75 0.95
 0.7  0.3  0.5  0.25 0.7  0.2  0.55 0.2  0.15 0.5  0.25 0.3  0.05 0.75
 0.3  0.2  0.2  0.35 0.5  0.05 0.35 0.75 0.55 0.05 0.05 0.9  0.55 0.95
 0.5  0.6  0.05 0.5  0.35 0.65 0.6  0.55 0.65 0.15 0.9  0.4  0.25 0.6
 1.   0.4  0.45 0.55 0.9  0.9  0.2  0.6  0.45 0.65 0.45 0.7  0.75 1.
 0.5  0.5  0.5  0.45 0.3  0.65 0.9  0.6  0.1  0.3  0.45 0.85 0.85 0.9
 0.05 0.05 0.3  0.5  0.4  0.3  0.5  0.1  0.3  0.3  0.05 1.   0.5  0.6
 0.4  0.6  0.1  0.65 0.65 0.2  0.75 0.5  0.05 0.25 0.2  0.5  0.3  0.55
 0.05 0.4  0.15 0.7  0.25 0.85 0.9  0.9  0.25 0.2  0.15 0.9  0.45 0.3
 0.45 0.45 0.8  0.8  0.45 0.8  0.8  0.5  0.7  0.85 0.3  0.4  0.3  0.25
 0.2  0.6  0.5  0.15 0.6  0.5  0.75 0.15 0.35 0.5  0.15 0.3  0.05 0.9
 0.65 0.9  0.6  0.5  0.2  0.75 0.45 0.35 0.7  0.6  0.45 0.9  0.85 0.7
 0.4  0.25 0.65 0.65 0.3  0.4  0.9  0.85 0.1  0.6  0.2  1.   0.15 0.4
 0.65 0.5  0.85 0.9  0.45 0.4  0.75 0.55 1.   0.7  0.45 0.9  0.2  0.5
 0.5  0.35 0.6  1.   0.6  0.5  1.   0.45 0.8  0.5  0.45 0.6  0.85 0.2
 0.2  0.9  0.3  0.45 0.45 0.75 1.   0.5  0.7  0.8  0.3  0.4  0.85 0.95
 0.5  0.5  0.5  0.5  0.4  1.   0.25 0.75 0.8  0.9  0.15 0.15 0.25 0.85
 0.2  0.45 0.5  0.7  0.2  0.1  0.9  0.9  0.45 0.95 0.45 0.2  0.75 0.4
 0.2  0.85 0.2  0.9  0.3  0.15 0.55 0.95 0.5  0.5  0.85 0.55 0.45 0.2
 0.6  0.4  0.2  0.7  0.3  0.45 0.6  0.4  0.5  0.2  0.45 0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.   0.
 0.   0.   0.   0.   0.   0.  ]

数据用 0 填充，这样所有序列的长度都是 1000 个元素。

这是我正在使用的模型的代码（如果要下载，请使用 dropbox link）

    import numpy as np
        from sklearn.utils import shuffle
        import tensorflow as tf
        from tensorflow import keras
        from tensorflow.keras.models import Sequential
        from tensorflow.keras.layers import Dense, Dropout, LSTM
        import numpy as np


        def main():
            posData = createListOfSequences('PositiveData.txt')
            negData = createListOfSequences('NegativeData.txt')

            # Shuffle both
            np.random.shuffle(posData)
            np.random.shuffle(negData)

            # Create target lists
            PosTargets = [1.0] * len(posData)
            NegTargets = [0.0] * len(negData)

            # Combine PosData and NegData into 1 list
            Sequences = posData + negData
            Targets = PosTargets + NegTargets

            # shuffle sequences but maintain link to targets
            (Sequences, Targets) = shuffle(Sequences, Targets)

            # make all sequences 1000 in length
            for sequence in Sequences:
                sequence = padWithZero(1000, sequence)

            Sequences = np.array(Sequences)
            Targets = np.array(Targets)

            FitModel(Sequences, Targets)


        def FitModel(data, target):
            trainPercentage = 0.8
            splitValue = round(len(data) * trainPercentage)

            (x_train, y_train) = data[0:splitValue], target[0:splitValue]
            (x_test, y_test) = data[splitValue:len(
                data)], target[splitValue:len(target)]

            model = Sequential()

            model.add(LSTM(128, activation='relu',
                           input_shape=(1000, 1), return_sequences=True))
            model.add(Dropout(0.2))
            model.add(LSTM(128, activation='relu'))
            model.add(Dropout(0.2))
            model.add(Dense(32, activation='relu'))
            model.add(Dropout(0.2))
            model.add(Dense(1, activation='sigmoid'))

            opt = tf.keras.optimizers.Adam(lr=1e-3, decay=1e-5)

            model.compile(optimizer='rmsprop',
                          loss='binary_crossentropy',
                          metrics=['accuracy'])

            model.fit(x_train, y_train, epochs=3, validation_data=(x_test, y_test))


        def padWithZero(targetLength, sequence):
            i = 0
            while not(len(sequence) >= targetLength):
                if (i % 2 == 0):
                    sequence.append(0.0)
                else:
                    sequence.insert(0, 0.0)
                i += 1
            return sequence


        def createListOfSequences(dataPath):
            DataAsString = []
            for line in open(dataPath):
                x = line.rstrip()
                DataAsString.append((x.split(',')))

            ListOfSequences = []

            for seq in DataAsString:
                listOfFloats = []
                for val in seq:
                    if '.' in val:
                        listOfFloats.append(float(val))
                ListOfSequences.append(listOfFloats)

            return(ListOfSequences)


        main()

如有任何帮助，我们将不胜感激。谢谢

Answer 1

正如错误信息所说，LSTM 需要 3 个维度的输入。你可以像这样重塑：

x_train = x_train.reshape(x_train.shape[0], 1, x_train.shape[1])

基本上，将形状 (a, b) 变成 (a, 1, b)。

检查输入时出错：预期 lstm_input 有 3 个维度，但得到形状为 (160, 1000) 的数组

Error when checking input: expected lstm_input to have 3 dimensions, but got array with shape (160, 1000)

python

numpy

lstm

keras

recurrent-neural-network