Huanhuan Chen – Softwares

Return to homepage Huanhuan Chen.

Probabilistic Classification Vector Machines

Description

PCVMs is a sparse learning algorithm. After analysing relevance vector machines (RVMs) for classification problems and observe that adopting the same prior for different classes may lead to unstable solutions. To tackle this problem, a signed and truncated Gaussian prior is adopted over every weights in PCVMs, where the sign of prior is determined by the class label, i.e. +1 or -1. Also in PCVMs, the kernel parameters can automatically optimistic during the training part.For more details, you may look at this paper.

Tutorail

Please download PCVM Matlab codes here.

Dr. rer. nat. Frank-Michael Schleif has nicely ported the Matlab code to C++ code. Please download the PCVM C++ codes here.

The demo will illustrate PCVM performance on Ripley's synthetic dataset. You may see this figure:



Two classes are label by red dot and blue X mark. The decision boundary is illustrated by the figure with a green dash line. Also, support vectors are mark circled by red circles. So, in this demo, PCVM only needs four support vectors to make classification decision.

Cluster Regularization

Description

Cluster-Based Regularization(ClusterReg) is a semi-supervised classification(SSC) algorithm. SSC learns from cheap unlabeled data and labeled data to predict the labels of test instances. In order to make use of the information from unlabeled data, there should be an assumed relationship between the true class structure and the data distribution. One assumption is that data points clustered together are likely to have the same class label. ClusterReg takes the partition given by a clustering algorithm as a regularization term in the loss function of an SSC classifier.

Please download ClusterReg codes here.