Scalpel: Customizing DNN pruning to the underlying hardware parallelism

As the size of Deep Neural Networks (DNNs) continues to grow to increase accuracy and solve more complex problems, their energy footprint also scales. Weight pruning reduces DNN model size and the computation by removing redundant weights. However, we implemented weight pruning for several popular n...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	2017 ACM/IEEE 44th Annual International Symposium on Computer Architecture (ISCA) S. 548 - 560
Hauptverfasser:	Jiecao Yu, Lukefahr, Andrew, Palframan, David, Dasika, Ganesh, Das, Reetuparna, Mahlke, Scott
Format:	Tagungsbericht
Sprache:	Englisch
Veröffentlicht:	ACM 01.06.2017
Schlagworte:	Computational modeling Graphics processing units Hardware hardware parallelism Matrix converters Microcontrollers multiple data neural network pruning Parallel processing single instruction Sparse matrices
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!