Tue PM1.L4.2
Improved Auto-Encoding using Deterministic Projected Belief Networks and Compound Activation Functions
Paul Baggenstoss, Fraunhofer, Germany
Session:
Tue PM1.L4: Learning Theory and Algorithms I Lecture
Track:
SiG-DML - Signal and Data Analytics for Machine Learning
Location:
Nautica
Presentation Time:
Tue, 5 Sep, 14:50 - 15:10 Finland Time (UTC +3)
Session Chair:
Alexander Jung, Aalto University
Presentation
Discussion
Resources
No resources available.
Session Tue PM1.L4
Tue PM1.L4.1: Assessment of a Two-step Integration Method as an Optimizer for Deep Learning
Paul Rodriguez, PUCP, Peru
Tue PM1.L4.2: Improved Auto-Encoding using Deterministic Projected Belief Networks and Compound Activation Functions
Paul Baggenstoss, Fraunhofer, Germany
Tue PM1.L4.3: DATA-FREE BACKBONE FINE-TUNING FOR PRUNED NEURAL NETWORKS
Adrian Holzbock, Ulm University, Germany; Achyut Hegde, Karlsruhe Institute of Technology, Germany; Klaus Dietmayer, Ulm University, Germany; Vasileios Belagiannis, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany
Tue PM1.L4.4: A STATISTICAL MODEL FOR PREDICTING GENERALIZATION IN FEW-SHOT CLASSIFICATION
Yassir Bendou, Vincent Gripon, Bastien Pasdeloup, Giulia Lioi, IMT ATLANTIQUE, France; Lukas Mauch, Stefan Uhlich, Fabien Cardinaux, Ghouthi Boukli Hacene, Javier Alonso Garcia, Sony Europe, Germany
Tue PM1.L4.5: Multiclass Minimax Learning for Deep Neural Networks
Cyprien Gilet, Université de Technologie de Compiègne, France; Marie Guyomard, Barbosa Susana, Lionel Fillatre, Université Côte d'Azur, France