Project
Please see a sampled project report from here
The project may be completed either individually or as a team; both approaches are acceptable. For team-based projects, only one team member should submit the final report and clearly specify all contributing teammates. Bonus Points will apply if you consider doing projects in the following fields with (*) or any domain beyond the following:
1. Background and Problem Formulation - 10%
- Background - 5%:
- What is the general background of the problem you are working on?
- I want to develop a better paper categorization system
- What is the general background of the problem you are working on?
- Problem Formulation - 5%:
- Under the general topic, what specific problem is your project addressing?
- I want to develop a machine learning model/algorithm to take input of the paper, output the paper topic (machine learning, computer system, human-computer collaboration, etc.)
- Under the general topic, what specific problem is your project addressing?
2. Data Mining Stage - 35%
- Data Collection and Store - 15%:
- What data are you looking to kick off your project? How do you collect them? What data structure do you use to represent them?
- I collect Cora/Citeseer/Pubmed Data from somewhere (e.g., a paper, a GitHub repository, Hugging Face, etc.), and I use an adjacency list to store their connection and a matrix to store their node feature
- What data are you looking to kick off your project? How do you collect them? What data structure do you use to represent them?
- Data Mining - 20%:
- What kind of data mining problem do you need to do and why?
- I need to analyze the network homophily/heterophily since leveraging this property might help me develop a better machine learning model for paper classification.
- How do you do it?
- I calculate for every edge, the two ending points, whether they are in the same class or not, and quantify the average ratio as a homophily ratio
- What kind of pattern do you find? How do you present your findings/analysis?
- I find that in many paper citation networks, the homophily is pretty high. Using Number/Table/Figure, etc.
- What kind of data mining problem do you need to do and why?
3. Machine Learning Stage - 35%
- Machine Learning Model Design:
- Based on your targeted problem, what kind of machine learning model do you want to build and why?
- I want to build a graph neural network to fully exploit the discovered homophily principle.
- How do you build your machine learning model, and what is the key design?
- The key design is to aggregate neighborhood information together, and I use the PyTorch Geometric Package, and within this package, certain functions can help me achieve my design philosophy
- What kind of experiments do you have to achieve your task?
- I conduct paper classification by taking the paper as input through my graph neural network, and it helps me achieve very high node classification performance.
- Based on your targeted problem, what kind of machine learning model do you want to build and why?
4. Discussion - 10%
- Discussion - 5%:
- Any further discussion on your exploration of this project?
- I take networks of different homophily and derive a clear positive relation between node classification and homophily level.
- Any further thinking and interesting questions you can ask about this project?
- How about the network performance on a heterophily network like dating networks? If we maliciously upload some papers that are meaningless, would that compromise the performance?
- Any further discussion on your exploration of this project?
- Feeling of this class - 5%:
- Feeling of this class.
* Infrastructure Mining and Machine Learning
- Water/Power/Gas network
- Transportation network
- Networking system
* Neural Biology Learning
- Brain Network
- Protein/Molecule
- Gene
Social Mining and Machine Learning
- Bluesky
- Amazon Recommender System