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

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

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

Computer Science > Machine Learning

Title: Dichotomize and Generalize: PAC-Bayesian Binary Activated Deep Neural Networks

Abstract: We present a comprehensive study of multilayer neural networks with binary activation, relying on the PAC-Bayesian theory. Our contributions are twofold: (i) we develop an end-to-end framework to train a binary activated deep neural network, (ii) we provide nonvacuous PAC-Bayesian generalization bounds for binary activated deep neural networks. Our results are obtained by minimizing the expected loss of an architecture-dependent aggregation of binary activated deep neural networks. Our analysis inherently overcomes the fact that binary activation function is non-differentiable. The performance of our approach is assessed on a thorough numerical experiment protocol on real-life datasets.
Comments: 22 pages. Accepted for publication at NeurIPS 2019
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:1905.10259 [cs.LG]
  (or arXiv:1905.10259v4 [cs.LG] for this version)

Submission history

From: Benjamin Guedj [view email]
[v1] Fri, 24 May 2019 14:37:38 GMT (3812kb,D)
[v2] Wed, 29 May 2019 12:52:11 GMT (3797kb,D)
[v3] Tue, 12 Nov 2019 14:22:07 GMT (4237kb,D)
[v4] Wed, 13 Nov 2019 20:18:17 GMT (4237kb,D)

Link back to: arXiv, form interface, contact.