Speaker:
Dr. Nicola Strisciuglio (DMB)
Short bio:
Nicola Strisciuglio received the PhD degree 'cum laude' in Computer Science from the University of Groningen (Netherlands) in 2016, and the PhD degree in Information Engineering from the University of Salerno (Italy) in 2017.
He is an Assistant Professor at the Faculty of Electrical Engineering, Mathematics and Computer Science of the University of Twente, The Netherlands. He has been general co-chair of the 1st and 2nd International Conference on Applications of Intelligent Systems (APPIS) in 2018 and 2019. His research interests include (multi-modal) machine learning, signal processing and computer vision.
Title:
Bio-inspired trainable features for image and sound processing
Abstract:
Representation learning is a fundamental aspect of Pattern Recognition and consists of learning features from training samples instead of engineering hand-crafted data representations, which usually require domain knowledge. Recent deep learning approaches require very large amount of training data to learn effective features.
In this talk, we will discuss the trainable feature extractors COSFIRE and COPE, used to learn effective representations from a small number of training examples in applications of image and audio processing. Their design is inspired by some functions of the visual and auditory systems of the brain. We will focus on the use of COSFIRE features for the delineation of elongated structures in images (e.g. blood vessels, rivers and roads in aerial images) and of the COPE features for the detection and classification of audio events in noisy environments.
