Your model has an output of shape (10,), however, your outputs have dimension (1,). You probably want to convert your y_train to categorical one-hot vectors, ie, via keras.utils.np_utils.to_categorical.

Thanks for your answer @kgrm , I used sparse_categorical_crossentropy to solve my problem

@yazidabdulsalam @hafizurcse My issue got solve by using loss =“sparse_categorical_crossentropy” Thanks.

@yazidabdulsalam Hey mate,

Not sure if you still need any help.

Just wanna explain your message and then explain what to do. By your error message " Error when checking target: expected dense_15 to have shape (None, 8) but got array with shape (43314, 1)" means that you supposed to feed a 2D array of shape(x, 8) such as (1, 8), or (2, 8) or (3, 8) or … … or (n, 8). That clears that x is a variable and you can have any number of rows but you must provide 8 columns. That is what “None” stands for.

Having same issue as ganav on Win7 anaconda setup. Keras version 2.0.9.

My code is inspired by this https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html just that I am trying multi class classification with optimizer = ‘adam’ or ‘rmsprop’ loss = ‘categorical_crossentropy’ batch_size = 16 number of classes = 10

ValueError: Error when checking target: expected activation_5 to have shape (None, 1) but got array with shape (16, 10)

P.S. This worked perfectly for binary classificaiton with optimizer = ‘adam’ or ‘rmsprop’ loss = ‘binary_crossentropy’

EDIT: My issue is resolved, it was a mistake by not making Dense=number of classes in my final layer

Importing the libraries

import numpy as np import matplotlib.pyplot as plt import pandas as pd

Importing the dataset

train_dataset = pd.read_csv(‘train.csv’) test_dataset = pd.read_csv(‘test.csv’) X = train_dataset.iloc[:, 1:94].values y = train_dataset.iloc[:, 94].values

Encoding categorical data

from sklearn.preprocessing import LabelEncoder, OneHotEncoder labelencoder_y_1 = LabelEncoder() y = labelencoder_y_1.fit_transform(y)

Splitting the dataset into the Training set and Test set

from sklearn.model_selection import train_test_split X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.3, random_state = 0)

Feature Scaling

from sklearn.preprocessing import StandardScaler sc = StandardScaler() X_train = sc.fit_transform(X_train) X_test = sc.transform(X_test)

#Part 2 -Lets make the ANN

#importing the keras libraries and keras import keras from keras.models import Sequential from keras.layers import Dense

#initializing the ANN classifier = Sequential()

#adding an input layer and the first hidden layer classifier.add(Dense(output_dim = 50,init =‘uniform’,activation = ‘relu’,input_dim=93))

#adding the second hidden layer classifier.add(Dense(output_dim =50,init =‘uniform’,activation = ‘relu’))

#adding the third hidden layer classifier.add(Dense(output_dim =50,init =‘uniform’,activation = ‘relu’))

#adding the forth hidden layer classifier.add(Dense(output_dim =50,init =‘uniform’,activation = ‘relu’))

#Adding the output layer #if you dependent variable is more than 1 categorie use softmax as the activation function, the ouput dim will also be your number of features classifier.add(Dense(output_dim = 8 ,init =‘uniform’,activation = ‘softmax’))

#Compiling the ANN #categorical_crossentropy classifier.compile(optimizer = ‘adam’ , loss = ‘categorical_crossentropy’, metrics = [‘accuracy’])

#Fitting the ANN to the traning set classifier.fit(X_train,y_train,batch_size = 10 ,nb_epoch=100)

def baseline_model(Xtrain,):
        # create model
        model = Sequential()
        model.add(Dense(50, input_dim=X_train.shape[1], init='normal'))
        model.add(PReLU())
        model.add(Dropout(0.2))
        model.add(Dense(output_dim=22, init='normal', activation='softmax'))
        # Compile model
        model.compile(loss='categorical_crossentropy', optimizer='adadelta', metrics=['accuracy'])  #logloss
        return model

You can try to use categorical_crossentropy instead of sparse_categorical_crossentropy .

Can anyone please tell me while solving a multiclass classification problem, why do we write n+1 variables in the Dense layer in Keras and 1 in case of binary classification not 2 or 3? Like if there are 3 classes to be predicted we write model.add(Dense(4,activation='softmax')) but incase of binary classification we write model.add(Dense(1,activation='sigmoid'))? Is there something like softmax uses more layers than sigmoid function or something? What is the reason?

@yazidabdulsalam what you did to solve your issue? I am also facing same issue in R but not getting what @hafizurcse trying to say.