Learning to spot analogies and differences within/across visual categories is an arguably powerful approach in machine learning and pattern recognition which is directly inspired by human cognition. In this thesis, we investigate a variety of approaches which are primarily driven by correlation and tackle several computer vision applications.
|Titolo della tesi:||Learning by correlation for computer vision applications: from Kernel methods to deep learning|
|Data di discussione:||28-mar-2018|
|Appare nelle tipologie:||Tesi di dottorato|