Simple and easy to understand (hopefully) deep learning by using Tensorflow in Python.
The code is used to solve the competition problem from Kaggle name Digit Recognizer.
The code will seperated in 3 files
- verify_dataset.ipynb (not finish yet) This files intent to verify dataset to make sure that we understand dataset we have.
- multilayer_perceptron.ipynb This files show you how to solve the problem by using MLP(Multilayer Perceptron) which is simple enough to understand the basic concept of Deep Learning and try to understand code.
- convolution.ipynb This files show you how to solve the problem by using CNN(convolution neural network) which is the suitable solution for image classification problem.
Please consider the following detail
- Repository don't provide you the dataset which you could download from kaggle directly and put it in the folder name "dataset"
- The code don't provide you the best result or cover appropriate technique for win the competition.
I hope this code would be useful for you. Any advises are welcome so please email to me.
Best Regards,
Chet Chetchaiyan
Kaggle : Chetchaiyan