This is an old revision of the document!
Suggestions on Machine Learning Course Projects
This webpage will be updated.
(1) Solve open problems (data) posted in (domestic/international) data science competitions
| Competition | Type | Download | Webpage |
|---|---|---|---|
| http://contest.kbig.kr/ | Movie Attendance Prediction | Data Request to Professor | |
| Insurance Fraud Detection | Data Request to Professor | ||
| Baseball Prediction | Data Request to Professor | ||
| http://crowd.kbig.kr/ | Disease Prediction | Data Request to Professor | webpage |
| Optimized Advertisement | Data Request to Professor | webpage | |
| https://www.kaggle.com/ | Kaggle Competitions | webpage | |
| Kaggle Datasets | webpage | ||
| https://gallery.cortanaintelligence.com/competitions | Decoding Brain Signal | webpage | |
| Womens Health Risk Assessment | webpage | ||
| Real-time Crime Forecasting Comeptition | webpage | ||
| BTW 2017 Data Science Challenge (Calls → Data Science Challenge) | webpage |
(2) Teach agents using OpenAI Gym
You can download reinforcement learning library from https://gym.openai.com/
- Develop and compare reinforcement learning algorithms
- You can use several environments supported from the library. Please see the list from https://gym.openai.com/envs
(3) Game
VizDoom AI Competition ( http://vizdoom.cs.put.edu.pl/ )
Project Malmo
https://www.microsoft.com/en-us/research/project/project-malmo/
You can use reinforcement learning to train “unit control” of StarCraft AI players on usemap settings (without terrain and with small number of units).
You can use open game data (from STEAM API, BWREPLAYs.com and LoL API).
(4) Financial Forecasting
Stock Market Index (Up/Down, Binary Classification) Prediction
- https://www.kaggle.com/c/the-winton-stock-market-challenge
Exhange Rate (Up/Down, Binary Classification) Prediction
(5) Art
Learn to compose music Learn to draw/style images (e.g., neural styling for art)
