Estaremos recebendo a visita do Tommi, que vai estar brevemente conosco na próxima quinta 25/10, apresentando a seguinte palestra na H-319-11, às 13h30:

Title: On Extreme Minimal Learning Machine

Speaker: Prof. Tommi Kärkkäinen, Faculty of Information Technology, University of Jyväskylä, Finland

Abstract:

Even in the era of deep learning, methods based on shallow feedforward architecture with random basis provide a simple and scalable alternative for supervised machine learning. The most popular of these methods is the extreme learning machine (ELM), with many possible random feature maps. More recently, methods using distance-based random basis have been suggested: the minimal learning machine (MLM) and the extreme minimal learning machine (EMLM). The basic forms of these methods are presented with some recent empirical findings -work in progress- with sparsity-favoring regularization. As the last point, these activities are linked to the determination of nanostructures - the second main reason why the speaker is having a research visit in Brasil.