Signal Analysis and machine learning for biomedical applications
In many real-life situations, it is important to automatically classify a given signal based on its characteristics, e.g. detection of seizures in EEG signals. This problem of signal classification is usually solved by a combination of signal processing techniques and machine learning algorithms. The signal analysis methods like time-frequency analysis and wavelet transform are used in the first stage to extract features that can better separate one class of signals from others. The second stage employs machine learning/pattern recognition techniques for classifying the extracted features to achieve the final decision. This interdisciplinary research group focuses combines signal processing expertise from the Biomedical Engineering Technology domain and machine learning expertise from the Information Technology domain to develop and efficiently implement algorithms for the classification of real-life signals such as biomedical signals. The focus research areas are:
1) Design of advanced signal analysis techniques for the analysis of non-stationary signals.
2) Extraction of features from joint time-frequency representations.
3) Application and enhancement of existing machine learning algorithms for the classification of non-stationary signals.
4) Real-time implementation of algorithms on hardware and software platforms such as GPUs/FPGAs/DSPs.
Dr. Nabeel Ali Khan
Dr. M. Ishtiaq (CS Department)
Dr. Ateeq Ur Rehman
Engr. Saad Habib Qureshi
Mr. Aqib Rehman