https://github.com/yuhexiong/digit-recognition-cnn-python

https://github.com/yuhexiong/digit-recognition-cnn-python

adam-optimizer cnn cross-entropy data-augmentation deep-learning python tensorflow

Last synced: 9 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/yuhexiong/digit-recognition-cnn-python
• Owner: yuhexiong
• Created: 2023-10-17T15:27:22.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2023-12-28T13:23:53.000Z (almost 2 years ago)
• Last Synced: 2025-01-30T01:14:49.760Z (11 months ago)
• Topics: adam-optimizer, cnn, cross-entropy, data-augmentation, deep-learning, python, tensorflow
• Language: Python
• Homepage:
• Size: 19.1 MB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Digit Recognition CNN

### DataSet From [Kaggle Competition - Digit Recognizer](https://www.kaggle.com/competitions/digit-recognizer)

## Overview

- Language: Python v3.10.12

- Package: Tensorflow

- Model: CNN

- Loss Function: Cross Entropy

- Optimizer: Adam, Learning Rate = 0.0003

- data augmentation to reduce overfitting

## Data Distribution

![image](https://github.com/yuhexiong/digit-recognition-CNN-python/blob/main/image/data_distribution.png)

## Loss and Accuracy

![image](https://github.com/yuhexiong/digit-recognition-CNN-python/blob/main/image/loss_history.png)

![image](https://github.com/yuhexiong/digit-recognition-CNN-python/blob/main/image/accuracy_history.png)

## Layer

![image](https://github.com/yuhexiong/digit-recognition-CNN-python/blob/main/image/cnn_layers.png)

## Confusion Matrix - Accuracy Rate 98.66%

![image](https://github.com/yuhexiong/digit-recognition-CNN-python/blob/main/image/confusion_mtx.png)