Results 31 to 40 of about 273,497 (278)
Revisiting Knowledge Distillation via Label Smoothing Regularization
Computer Vision and Pattern Recognition, 2020 Knowledge Distillation (KD) aims to distill the knowledge of a cumbersome teacher model into a lightweight student model. Its success is generally attributed to the privileged information on similarities among categories provided by the teacher model ...
Li Yuan +4 more
semanticscholar +1 more source
Directional Regularity and Metric Regularity
SIAM Journal on Optimization, 2007 For general constraint systems in Banach spaces, we present the directional stability theorem based on the appropriate generalization of the directional regularity condition, suggested earlier in [A. V. Arutyunov and A. F. Izmailov, Math. Oper. Res., 31 (2006), pp. 526-543].
Arutyunov A.V. +2 more
openaire +3 more sources
Proceedings of the American Mathematical Society, 1972 Mathematics Technical ...
openaire +4 more sources
Non-uniform refinement: adaptive regularization improves single-particle cryo-EM reconstruction
Nature Methods, 2019 Cryogenic electron microscopy (cryo-EM) is widely used to study biological macromolecules that comprise regions with disorder, flexibility or partial occupancy.
A. Punjani, H. Zhang, David J. Fleet
semanticscholar +1 more source
Regular Homomorphisms and Regular Maps
European Journal of Combinatorics, 2002 The authors study regular automorphisms of oriented maps. Their main concern is the problem of lifting and projecting map automorphisms. Regular homomorphisms of oriented maps essentially arise from a factorization by a subgroup of automorphisms. In this paper, this kind of map automorphisms is studied in detail.
Malnič, Aleksander +2 more
openaire +1 more source
The Little Engine That Could: Regularization by Denoising (RED) [PDF]
SIAM Journal of Imaging Sciences, 2016 Removal of noise from an image is an extensively studied problem in image processing. Indeed, the recent advent of sophisticated and highly effective denoising algorithms has led some to believe that existing methods are touching the ceiling in terms of ...
Yaniv Romano, Michael Elad, P. Milanfar
semanticscholar +1 more source
Regular Operator Equations: Conditions for Regularity [PDF]
Proceedings of the American Mathematical Society, 1982 Regular operator equations are causal equations admitting unique solutions and have the property that all of their limiting equations along solutions admit unique solutions. Sufficient conditions which guarantee that an operator equation x = T x x = Tx is regular are given in case T T is
openaire +2 more sources
Machine Learning with Membership Privacy using Adversarial Regularization [PDF]
Conference on Computer and Communications Security, 2018 Machine learning models leak significant amount of information about their training sets, through their predictions. This is a serious privacy concern for the users of machine learning as a service.
Milad Nasr, R. Shokri, Amir Houmansadr
semanticscholar +1 more source
Regularized CNN Feature Hierarchy for Hyperspectral Image Classification
Remote Sensing, 2021 Convolutional Neural Networks (CNN) have been rigorously studied for Hyperspectral Image Classification (HSIC) and are known to be effective in exploiting joint spatial-spectral information with the expense of lower generalization performance and ...
Muhammad Ahmad +2 more
doaj +1 more source
Error Estimations for Total Variation Type Regularization
Mathematics, 2021 This paper provides several error estimations for total variation (TV) type regularization, which arises in a series of areas, for instance, signal and imaging processing, machine learning, etc.
Kuan Li, Chun Huang, Ziyang Yuan
doaj +1 more source