Research
The main research areas of Prof. Kecman and his students at LAAL are:- OLLA, A unique fast and novel (stochastic gradient descent (SGD) based) online learning algorithm (OLLA) for nonlinear i.e., kernelized, support vector machines (SVMs) and other alike models in primal domain. OLLA paper can be DOWNLOADED from DOWNLOADS page in the bar above.
- Learning algorithms for ultralarge datasets a.k.a. Big Data - on standalone computers, clouds, clusters, GPUs, grids,...
- Parallelization of machine learning algorithms
- Locally linear support vector machines and other local models for highly nonlinear dependencies
- Features extraction i.e., dimensionality reduction for very high dimensional datasets
- Transformation of regression problems into multiclass classification tasks and corresponding algorithms developments
- Semisupervised algorithms and software
- Applications of all the mentioned algorithms everywhere where the data sets are available
- Fighting AIDS i.e., SIDA - HIV infection models development. Biology systems modeling