Dr. Kecman is tenured Professor with the Computer Science Department at the Virginia Commonwealth University (VCU) in Richmond, VA, USA, where he directs the Learning Algorithms and Applications Laboratory (LAAL).
Dr. Kecman was the Fulbright Professor at MIT, Cambridge, MA, USA; DFG Professor at TH Darmstadt; DAAD Konrad Zuse Professor at FH Heilbronn, FHTW Berlin and SWFH Soest; Research Fellow at Drexel University, Philadelphia, PA and at Stuttgart University, as well as the associate professor at both The University of Auckland and Zagreb University.
Dr. Kecman’s current research interests include: machine learning from experimental data (knowledge discovery, data mining) by support vector machines and neural networks, as well as modeling human knowledge by fuzzy logic systems. Theory, practice, philosophy and versatility of these soft computing tools is used in broad fields of different (nonlinear) regression and pattern recognition (classification, decision making) tasks in – e-commerce, bioinformatics, vision systems, computer graphics, data and signal processing, numerical mathematics, credit assignment problems, (financial) time series analysis, image compression, expert and decision systems and computer intelligence. Vojo’s other longlasting commitment was/is fighting AIDS (SIDA) disease by HIV infection modeling, which is a part of his modeling of biological systems devotion for many years.
Industry Expertise (1)
Areas of Expertise (5)
Machine learning and data mining
Bioinformatics and biomedical informatics
Fuzzy logic modeling
System dynamics modeling and analysis
SVM algorithms for large datasets
University of Zagreb: Ph.D. 1982
University of Zagreb: M.Sc. 1978
University of Zagreb: Dipl. Ing. 1972
- Graduate Committee
- SoE Faculty Council
- SoE P&T Committee
Selected Articles (5)
The adaptive local hyperplane (ALH) algorithm is a very recently proposed classifier, which has been shown to perform better than many other benchmarking classifiers including support vector machine (SVM), K-nearest neighbor (KNN), linear discriminant analysis (LDA), and K-local hyperplane distance nearest neighbor (HKNN) algorithms. Although the ALH algorithm is well formulated and despite the fact that it performs well in practice, its scalability over a very large data set is limited due to the online distance computations associated with all training instances. In this paper, a novel algorithm, called ALH-Fast and obtained by combining the classification tree algorithm and the ALH, is proposed to reduce the computational load of the ALH algorithm. The experiment results on two large data sets show that the ALH-Fast algorithm is both much faster and more accurate than the ALH algorithm.
The objective of this work was to optimize (minimize) the compressed air required to control the rate of ammonia removal in a commercially operated wastewater bioreactor, while still maintaining operation within environmental consent limits. In order to do this, a nonlinear dynamic model based on the International Association on Water Quality (IAWQ) activated sludge model No. 3 was developed, expressing the nitrification kinetics and hydraulic dynamics of the system. From this model a steady state representation of the plant was derived, and simulated for various load characteristics experienced at the facility, and as a result an optimal load profile was developed for the compressed air distribution to the four aerobic zones. The optimal load profile will ensure that the amount of compressed air required to control the rate of ammonia removal is optimized.
In this paper, a novel classifier, called adaptive local hyperplane, is proposed for pattern classification. The experimental results on 11 real data sets demonstrate that the proposed classifier outperforms, on average, all the other seven benchmarking classifiers. In particular, it is the best classifier in 10 out of 11 data sets, and it is the close second best for just one data set.
The paper introduces a novel adaptive local hyperplane (ALH) classifier and it shows its superior performance in the face recognition tasks. Four different feature extraction methods (2DPCA, (2D)2PCA, 2DLDA and (2D)2LDA) have been used in combination with five classifiers (K-nearest neighbor (KNN), support vector machine (SVM), nearest feature line (NFL), nearest neighbor line (NNL) and ALH). All the classifiers and feature extraction methods have been applied to the renown benchmarking face databases—the Cambridge ORL database and the Yale database and the ALH classifier with a LDA based extractor outperforms all the other methods on them. The ALH algorithm on these two databases is very promising but more study on larger databases need yet to be done to show all the advantages of the proposed algorithm.
This paper presents the algorithms and the results of multi-user detectors (MUD) on a synchronous chaos-based code division multiple access system (CDMA), which uses chaotic sequences as the spreading codes. Popular linear and non-linear MUD algorithms such as the decorrelator detector, minimum mean square error (MMSE) detector and parallel interference cancellation (PIC) detector are all considered in this paper. These conventional detectors are used to compare the BER performance with a novel blind-MUD receiver. The blind-MUD is achieved by a recently emerged learning technique called support vector machines (SVM). This method can be used to replace the conventional matched filter of the receiver and can be implemented on the forward link. All the MUD schemes are simulated over an AWGN channel and result shows that the blind-MUD compare favorably with other techniques.