We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computer Vision and Pattern Recognition

Title: Hybrid Pruning: Thinner Sparse Networks for Fast Inference on Edge Devices

Abstract: We introduce hybrid pruning which combines both coarse-grained channel and fine-grained weight pruning to reduce model size, computation and power demands with no to little loss in accuracy for enabling modern networks deployment on resource-constrained devices, such as always-on security cameras and drones. Additionally, to effectively perform channel pruning, we propose a fast sensitivity test that helps us quickly identify the sensitivity of within and across layers of a network to the output accuracy for target multiplier accumulators (MACs) or accuracy tolerance. Our experiment shows significantly better results on ResNet50 on ImageNet compared to existing work, even with an additional constraint of channels be hardware-friendly number.
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
Cite as: arXiv:1811.00482 [cs.CV]
  (or arXiv:1811.00482v1 [cs.CV] for this version)

Submission history

From: Xiaofan Xu [view email]
[v1] Thu, 1 Nov 2018 16:24:50 GMT (117kb,D)

Link back to: arXiv, form interface, contact.