TMNIST-Digits-Identification

Technology Used

1. CNN (Deep Learning)
2. TensorFlow
3. Pandas
4. Matplotlib
5. Python

Summary

We created sequential CNN model with 2 layer and after 5 rounds of training with 64 sets of data in each step we got accuracy of 99% which is pretty good for 2 layer CNN

The accuracy of the model during training reached 99.38%, and the test data set reached 99.10%

At each iteration the accuracy was increasing and loss was seen to be reduced marginally

Convolutional Neural Network

A Convolutional Neural Network (CNN) is a deep learning technique which can take in an input image, assign importance to various aspects in the image through learnable weights and biases. Finally, the trained model is able to differentiate one image from the other. One of the major advantages of using CNN over other deep learning techniques is that the pre-processing required is much lower.

Training Sample Images

Model Learning Curve

Model Accuracy & Confusion Matrix