DIFS - DO IT FROM SCRATCH

Everything I have learned, from data structures & algorithms to machine learning & robotics, is coded solely from scratch.

The bare mininum packages used here are just numerical toolboxes such as PyTorch or Numpy. Some tasks, e.g., control theory, are done in MATLAB/Simulink.

Codes are saved here publicly for my future review. Each package contains a README.md file that describes more details about things inside:

Traditional machine learning algorithms

• Linear regression (extend) => Compare regression with interpolation
• Logistic regression
• Decision tree
• Softmax regression
• k-Nearest Neighbor
• DBSCAN (Density-Based Spatial Clustering of Applications with Noise)
• Support vector machine

Deep learning

• Fully-connected network
• Convolutional network
• Long short-term memory
• Transformer

Signal Processing (SSP)

• Estimators
• FIR/IIR Filters
• Hypothesis testing & classical inference algorithms
• Optimizers

Control Theory

• Model-based fault detection using parity method
• Fault-tolerance control with virtual sensor and virtual actuator
• Model-free fault detection using statistics