Assignment Information

There will be four assignments, each of which is designed for testing your understanding of the taught materials. It could be either programming or written analysis.
All students are expected to follow the FSU Academic Honor Code.
All assignments follow the "no-late" policy; That is, assignments received after the due time will receive zero credit.

• Assignment 1
  • Topic: Data, Data Preprocessing
  • Type: written analysis and programming
  • Due time/date: 11:59pm Wednesday 2/10
  • Submission: one submission package (zip or tar.gz format) including the written homework (pdf) and the software package (including C/C++ source code, makefile, and readme). Submit via Canvas.
• Assignment 2
  • Topic: Data Preprocessing, Frequent Pattern Mining
  • Type: written analysis
  • Due time/date: 11:59pm Friday 3/5
  • Submission: one pdf file submitted via Canvas.
• Assignment 3
  • Topic: Classification
  • Type: machine code
  • Due time/date: 11:59pm Sunday 3/28
  • Submission: one .zip package file submitted via Canvas.
• Assignment 4
  • Topic: Classification, Clustering
  • Type: written analysis
  • Due time/date: 11:59pm Friday 4/16
  • Submission: one pdf file submitted via Canvas.

Project Information

The semester-long project involves a systematic study for a data mining research topic, by reading and understanding scientific publications, and writing a survey-like summary for that topic;
The project needs to be done individually;
The deliverables include (1) Project proposal (1-to-2 page): 10%; (2) Project presentation (15-20 minutes video): 30%; (3) Project report (5-8 pages, single column, Latex-preparation preferred): 60%.
Some recommended topics (and readings) are as follows:

1. Tree-based ensemble learning
   • XGBoost: a Scalable Tree Boosting System. KDD'16
   • LightGBM: a Highly Efficient Gradient Boosting Decision Tree. NeurIPS'17
   • CatBoost: Unbiased Boosting with Categorical Features. NeurIPS'18
2. Frequent graph pattern mining
   • gSpan: Graph-based Substructure Pattern Mining. ICDM'03
   • A Quickstart in Frequent Structure Mining Can Make a Difference. KDD'04
   • Flexible and Feasible Support Measures for Mining Frequent Patterns in Large Labeled Graphs. SIGMOD'17
3. Generative Adversarial Nets (GAN)
   • Generative Adversarial Nets. NeurIPS'14
   • Wasserstein GAN. Arxiv'17
   • Are GANs Created Equal? NeurIPS'18
4. Graph Embedding
   • DeepWalk - Online Learning of Social Representations (KDD'14)
   • LINE - Large-scale Information Network Embedding (WWW'15)
   • Node2vec - Scalable Feature Learning for Networks (KDD'16)
5. Data Sketching for Data Streams
   • Mergeable Summaries (TODS'13)
   • Efficient Frequent Directions Algorithm for Sparse Matrices (KDD'16)
   • A high-performance algorithm for identifying frequent items in data streams (IMC'17)