elcaiseri / Digit-Recognizer-MNIST

Implement deep learning (CNN Network) and machine learning (Random Forest, KNN and Naive Bayes) to recognize digits from (0~9)

Digit Recognizer - MNIST

This repository focuses on digit recognition using two different approaches:

1. Deep Learning with Keras: Achieves an accuracy of approximately 99%.
2. Machine Learning with scikit-learn: Achieves an accuracy of about 95%.

The dataset used for this project is the Digit Recognizer Dataset from Kaggle.

MNIST: Introduction to Convolutional Neural Network (CNN) with Keras

Table of Contents:

1. Introduction
2. Data Preparation
   • Load data
   • Normalization
   • Reshape
   • Label encoding
   • Split training and validation set
3. Convolutional Neural Network
   • Define the model architecture
   • Data augmentation
   • Model training

4. Evaluate the Model
   • Training and validation curves
   • Confusion matrix
5. Prediction and Submission
   • Prediction validation results
   • Submission
6. References

MNIST: Simple Machine Learning Algorithms

Table of Contents:

1. Introduction
2. Machine Learning Algorithms
   • Random Forest Classifier
   • KNeighborsClassifier
   • Naive Bayes
3. Submission

---

You can copy and edit the code directly from my Kaggle notebooks:

1. MNIST: Simple CNN Keras (Accuracy: 0.99) => Top 1%
2. MNIST: Simple Machine Learning Algorithms

If you find this project helpful or interesting, please don't forget to upvote it!

© 2023 elcaiseri