Handwritten_digit_recognizer

Using Tensorflow creating a handwritten digit recognizer So using the tensorflow and the keras API installed in i have created this handwritten digit recognizer it consists of creating the sets of data from the data set being downloaded into it then the next step involves creating the layers

The layers are as follows

• Flatten layer- i have added this layer because here in this model the whole image is being sent as one unit into the input of 28*28 units
• Dense layer is for the operation of input and ouput
• Dropout layer is used for solving the problem of overfitting as my model was facing that issue
• Dense layer is using the softmax function here which gives us the probability here

The next step was compiling

• The model using the adam compiler and the loss function as sparse_categorical_crossentropy

Training the data using the parameters:

• Epoch = number of repetitions on the set of data
• Validation split = to validate the training a fractional amount of data is specified for validation
• Batchsize =to divide the data in a significant amount of batches
• Verbose = to define the representation of training
• Data and its labels were the first two ones

Testing the data using the data given

Build Instructions

Enter this command into terminal/command prompt:

git clone https://github.com/AKA2501/digit_recognizer.git 

Contributions

Anyone who's intrested can contribute to this project