TSVMs

A collection of Twin Support Vector Machine (TSVM) implementations for pattern classification.

Implementations

• TSVM - Twin Support Vector Machine
• UTSVM - Unsupervised Twin Support Vector Machine
• NewtonUTSVM - Newton Method-based Unsupervised Twin Support Vector Machine
• DTSVM - Deep Twin Support Vector Machine
• LSTSVM - Least Squares Twin Support Vector Machine
• DLSTSVM - Deep Least Squares Twin Support Vector Machine
• RUTSVM - Robust Unsupervised Twin Support Vector Machine
• RULSTSVM - Robust Unsupervised Least Squares Twin Support Vector Machine

Usage

Each implementation follows a similar interface:

from TSVM import TSVM

# Prepare your data
# X: features (n_samples, n_features)
# y: labels (n_samples, 1) with values +1 or -1
# C1, C2: regularization parameters

# Create and train the model
model = TSVM(X, y, C1=1.0, C2=1.0)
model.fit()

# Make predictions
model.predict(x_test)
predictions = model.get_preds()

Requirements

• numpy
• cvxopt

References

• TSVM: IEEE Paper
• UTSVM: ScienceDirect
• DTSVM: IEEE Paper
• LSTSVM: ScienceDirect
• RUTSVM: ScienceDirect

Author

Saeed Khosravi