UE Machine Learning for Computer Vision and Audio Processing

User information

Please note that you are curently looking at the ongoing Academic Programs. Applications are now closed for this academic year (2020-2021) for licences, professional licences, masters, DUT and regulated health training. If you are interested for an application in 2021-2022, please click on this link for the appropriate Academic Programs.

Degrees incorporating this pedagocial element :


In this course we present recent state-of-the-art methods for visual object category representation and recognition, and the techniques that underpin these methods.   Methods will include so called "bag of  features" approaches, Fisher vectors, and convolutional neural networks for tasks such as instance-level image retrieval, image classification, object localization, semantic segmentation, image
caption generation and action recognition in videos.
On the machine learning side we consider clustering methods (k-means, mixutre of Gaussians), classification techniques (SVM, logistic discriminant), and kernels to obtain non-linear classifiers, as well as the principles underlying neural networks (multi-layer perceptron, back-propagation, convolutional networks, recurrent networks).