Core Research Activities

Machine Learning

Development of novel machine learning algorithms, especially concerning connectionist approaches and neural network architectures.

Pattern classification and recognition

Development of neural network based constrained learning algorithms for classification and object recognition. Also clustering algorithms with applications in several fields. A platform for solving pattern classification problems (Billnet) using connectionist and statistical methods has been developed by our group and is available on-line.

Knowledge representation

Development of decision support neural or knowledge based systems in many application areas, some of which are listed below. Feature selection and extraction methods for the efficient encoding of information.

Multimedia processing and retrieval

Experience and published work in the fields of segmentation, image understanding, optical character recognition, content-based image retrieval.