Final Project

The objective of the final project is to integrate various machine learning techniques to achieve the best performance. Final project is a group effort. Each group can have 2-4 members. You are required to apply ALL techniques learned in this semester, including those for feature extraction/selection, dimensionality reduction, classification/regression, and fusion. Your design needs to be well justified and pros/cons thoroughly analyzed through comprehensive experimental design and performance evaluation.

Schedule

• (5) Milestone 1 (Due 11/28): Group Formation and Topic selection. Submit through Canvas. Approval and comments will be returned in one day. The same topic cannot be chosen by more than 1 group. The topic follows the first-come first-served rule. So pick a topic as soon as possible.
• (5) Milestone 2 - Literature Survey (Due 12/05): Background study including references and state-of-the-art performance on the dataset (2-page report need to be submitted on Canvas).
• (5) Milestone 3 - Prototype 1 (Due 12/07): Prototype, preliminary results and task allocation among group members. Apply at least one learned technique successfully for each component in the pipeline on the chosen dataset and submit a 1-page report.
• (100) Final presentation (Due 12/13)(Presentation slides due after the presentation on 12/13. Submit through Canvas)
• (85) Final report (Due 12/14). Submit through Canvas. Implement at least two solutions to each component of the pipeline. Determine what metrics to use. Provide performance evaluation results.

Potential Topics

Requirement

General steps involved in a machine learning problem include

• Data collection (raw data)
• Feature extraction (how to extract features from the raw data)
• Feature reduction (dimensionality reduction - Fisher's linear discriminant or PCA or t-SNE)
• Classification/Regression methods need to be included
  • Supervised learning and Unsupervised learning
  • Baysian approaches and non-Baysian approaches
  • Parametric and Non-parametric density estimation in supervised learning
  • Fusion
• Performance evaluation
• Feedback system