I work primarily in the field of statistical pattern recognition, plus many of its application areas.
The fundamental theme of my research is on observation and modeling of complex systems and phenomena in the physical world. This theme develops into algorithms, tools, and applications of pattern recognition, data mining, performance monitoring, and computational modeling and simulation.
I seek to discover and represent knowledge embedded in large, high-dimensional data sets by algorithmic processes. More specifically, I pursue methods for natural partitioning of data, systematic search for correlations, characterizing complexity of dependences, generation and evaluation of feature transformations, dynamic adjustment of established models, and above all, classification. On classification, I explored methods for multiple classifier systems, random decision forests, and more recently, data complexity analysis.
To facilitate these, I also explore methods and tools for interactive data visualization and analysis. For my special interests in data coming from sensors and imaging devices, my methods emphasize joint explorations with the raw data and all levels of abstraction resulting from pre-processing, feature extraction, and decision making algorithms.
Besides pattern recognition, I pursue an interest in computational modeling and simulation of complex systems. I have built models of reading processes, sensor networks, and optical transport systems. These include 7 years (2001-2007) of work on computational simulation, statistical monitoring, and diagnostics of the product LambdaXtreme Transport.