Zur Webseite der Uni Stuttgart
<  Delay Fault detection in FinFET technology using parameter variation
12.02.19

Fault Tolerance and Approximate Computing for Support Vector Machines

Kategorie: Open Seminar - Rechnerarchitektur

13:00 - 13:45, external place, M. Sc. Alexander Schöll, Institut für Technische Informatik


Object classification and outlier detection are essential applications 
of modern Machine Learning models. In supervised learning, such models 
approximate an unknown function based on known samples from a training 
data set. The approximation function is often implemented either by one 
of various types of Artificial Neural Networks or by a classifier using 
Support Vector Machines (SVM). Reduced-precision and approximate 
computing are rather often used for implementing efficient neural 
networks, but less often for the training of SVMs. This presentation 
discusses how Approximate Computing (AxC) techniques can be used for 
compute-intensive linear algebra operations in order to construct an 
efficient SVM training algorithm. A novel fault tolerance method is used 
to monitor and control errors induced by approximation.

Akzeptieren

Diese Webseite verwendet Cookies. Durch die Nutzung dieser Webseite erklären Sie sich damit einverstanden, dass Cookies gesetzt werden. Mehr erfahren, zum Datenschutz