Skip Navigation Links

The lab is focused on developing efficient learning algorithms for data mining with applications in biology and medicine. The lab developed a wide range of algorithms from clustering, through discretization (CAIM and ur-CAIM, currently the best discretization algorithms for supervised balanced and unbalanced data), visualization, and supervised inductive machine learning algorithms for single-instance (DataSqueezer), multiple-instance (mi-DS) and one-class (OneClass-DS) learning. All are implemented in open-source software platforms and frequently used. The other area of research is in artificial neural networks, in particular in networks of spiking neurons. The later are used for modeling of organization and reorganization of the somatosensory cortex, glutamate release mechanism in epileptic hippocampus, cortex multisensory processing, and cortex multi-layer multi column inhibition. Outside of the brain modeling realm they are used for recognition of partially occluded and rotated face images without the need of any preprocessing, as they operate in a deep learning mode. The lab also developed a comprehensive chromosome 21 database and accompanying analytical tools, such as for protein-protein interactions, More about the lab’s activities can be found at here

Chromosome 21 gene function and pathway database

HMM predictor for protein protein interaction sites

GAKREM: clustering algorithm that generates a number of clusters on its own 

Computational models of multisensory processing

VCU Center for Clinical and Translational Research, Diabetes data and their Analysis

Multi-column multi-layer model of neocortex

Protein expression levels of 77 proteins measured in the nuclear fraction of cortex from control and Down syndrome mice.

401 West Main Street   Room E2264  Richmond, Virginia 23284