[Matlab code]

Binary Classification

-- is to learn a separation rule (classifier) from training dataset.

Quadratically Constrained Maximum A Posteriori (QCMAP) classifiers

-- It obtains binary classifiers based on the QCMAP criterion.

-- It can be formalized as "second-order cone programming".

We proposed four types of QCMAP classifiers

1. QCMAP with Gaussian weight

2. QCMAP with Difference of Gaussian (DOG) weight

3. QCMAP with Kernel Density Estimation (KDE) weight

4. QCMAP with DOG and KDE weight