S
Stephen J. Verzi
Researcher at Sandia National Laboratories
Publications - 61
Citations - 744
Stephen J. Verzi is an academic researcher from Sandia National Laboratories. The author has contributed to research in topics: Artificial neural network & Fuzzy logic. The author has an hindex of 15, co-authored 60 publications receiving 597 citations. Previous affiliations of Stephen J. Verzi include University of New Mexico.
Papers
More filters
Journal ArticleDOI
Potential Public Health Effects of Reducing Nicotine Levels in Cigarettes in the United States
Benjamin J. Apelberg,Shari P. Feirman,Esther Salazar,Catherine G. Corey,Bridget K. Ambrose,Antonio Paredes,Elise Richman,Stephen J. Verzi,Eric D. Vugrin,Nancy S. Brodsky,Brian L. Rostron +10 more
TL;DR: It is estimated that lowering the nicotine content in cigarettes to a minimally addictive level could substantially reduce tobacco-related mortality.
Journal ArticleDOI
Training deep neural networks for binary communication with the Whetstone method
TL;DR: Whetstone as mentioned in this paper is a method to bridge the gap by converting deep neural networks to have discrete, binary communication, where activation function at each layer is progressively sharpened towards a threshold activation, with limited loss in performance.
Journal ArticleDOI
Modeling the potential effects of new tobacco products and policies: a dynamic population model for multiple product use and harm.
Eric D. Vugrin,Brian L. Rostron,Stephen J. Verzi,Nancy S. Brodsky,Theresa J. Brown,Conrad J. Choiniere,Blair N. Coleman,Antonio Paredes,Benjamin J. Apelberg +8 more
TL;DR: A multi-state, dynamical systems population structure model that can be used to assess the effects of tobacco product use behaviors on population health shows that population health benefits are particularly sensitive to product risks and initiation, switching, and dual use behaviors.
Journal ArticleDOI
Whetstone: A Method for Training Deep Artificial Neural Networks for Binary Communication.
TL;DR: The Whetstone method achieves this by gradually sharpening activation functions during the training process by converting deep neural networks to have discrete, binary communication.
Proceedings ArticleDOI
Exemplar-based pattern recognition via semi-supervised learning
Georgios C. Anagnostopoulos,Madan Bharadwaj,Michael Georgiopoulos,Stephen J. Verzi,Gregory L. Heileman +4 more
TL;DR: It is shown that, especially for difficult pattern recognition problems featuring high class overlap, for exemplar-based classifiers implementing SSL the generalization performance improves, while the number of necessary exemplars decreases significantly, when compared to the original versions of the classifiers.