https://github.com/roma-glushko/kaggle-digit-recognizer
🔢 Digit Recognition Competition
https://github.com/roma-glushko/kaggle-digit-recognizer
digit-recognition kaggle-competition kaggle-digit-recognizer machine-learning mnist-handwriting-recognition
Last synced: 27 days ago
JSON representation
🔢 Digit Recognition Competition
- Host: GitHub
- URL: https://github.com/roma-glushko/kaggle-digit-recognizer
- Owner: roma-glushko
- Created: 2020-12-28T18:47:50.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2021-01-28T07:34:26.000Z (over 4 years ago)
- Last Synced: 2025-02-12T11:53:15.534Z (3 months ago)
- Topics: digit-recognition, kaggle-competition, kaggle-digit-recognizer, machine-learning, mnist-handwriting-recognition
- Language: Jupyter Notebook
- Homepage:
- Size: 10.5 MB
- Stars: 0
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: readme.md
Awesome Lists containing this project
README
# Kaggle: Digit Recognizer
This is a solution for Kaggle Digit Recognizer competition.
It's based on legendary MNIST dataset which contains 70,000 handwritten digits in 28x28 format.
The notebook includes 3 solutions:
- SVM on PCA (0.97610)
- MLP (0.97614)
- Ensamble of 10x CNN with soft voting prediction (0.99657, top 6%)
Competition Link: https://www.kaggle.com/c/digit-recognizer/overview
## Installation
Install project dependencies with Poetry:
```bash
poetry install
```
Install dataset:
```bash
cd data
kaggle competitions download -c digit-recognizer
```
Project requires Python 3.8.6 as it uses TensorFlow2.
## T-SNE Visualization
