Gradient Descent (GD) v.s. Conjugate Gradient Descent (CGD) for 2-D Linear Regression
This is one part of the course assignment for HKUST-GZ MICS 6000I Physical Design Automation of Digital Systems. This project is alive, maintained by linfeng.du@connect.ust.hk. Any discussion or suggestion would be greatly appreciated!
- Python 3.9
- ply 3.11
- matplotlib 3.5.1
- logging 0.5.1.2
- numpy 1.21.5
-
Logs:
- in GD-CGD.log
-
Convergence trajectories:
- Gradient Descent (GD): in GD-cost-convergence.png
- Conjugate Gradient Descent (CGD): in CGD-cost-convergence.png
-
3-D Convergence Animation:
- Gradient Descent (GD): in GD-result-3d.gif
- Conjugate Gradient Descent (CGD): in CGD-result-3d.gif
-
Intermediate results (3-D figure) during convergence