Publications

Full list of papers, technical reports, and associated materials (code, slides, bib). See also Google Scholar.

Papers

Deep Multivariate Models with Parametric Conditionals

D. Schlesinger, B. Flach, A. Shekhovtsov · arXiv preprint 2026

Scalable Binary-Quantized Neural Networks for Energy-Efficient Vision

A. Shekhovtsov, Š. Obdržálek · ECML-PKDD 2026 · LNCS 16943, pp. 253–270

Generalized and Optimal Straight-Through Estimators

J. Hooper, A. Shekhovtsov · AISTATS · Spotlight 2026

RANSAC Scoring Functions: Analysis and Reality Check

A. Shekhovtsov · arXiv preprint 2025

Symmetric Equilibrium Learning of VAEs

B. Flach, D. Schlesinger, A. Shekhovtsov · AISTATS 2024

Cold Analysis of Rao-Blackwellized Straight-Through Gumbel-Softmax Gradient Estimator

A. Shekhovtsov · ICML 2023

Generalized Differentiable RANSAC

T. Wei, Y. Patel, A. Shekhovtsov, J. Matas, D. Barath · ICCV 2023

VAE Approximation Error: ELBO and Exponential Families

D. Schlezinger, A. Shekhovtsov, B. Flach · ICLR 2022

Bias-variance tradeoffs in single-sample binary gradient estimators

A. Shekhovtsov · GCPR 2021

Reintroducing straight-through estimators as principled methods for stochastic binary networks

A. Shekhovtsov, V. Yanush · GCPR 2021

Initialization and Transfer Learning of Stochastic Binary Networks From Real-Valued Ones

A. Livochka, A. Shekhovtsov · BiVision @ CVPR 2021

Path Sample-Analytic Gradient Estimators for Stochastic Binary Networks

A. Shekhovtsov, V. Yanush, B. Flach · NeurIPS 2020

Taxonomy of dual block-coordinate ascent methods for discrete energy minimization

S. Tourani, A. Shekhovtsov, C. Rother, B. Savchynskyy · AISTATS 2020

Belief propagation reloaded: Learning bp-layers for labeling problems

P. Knöbelreiter, C. Sormann, A. Shekhovtsov, F. Fraundorfer, T. Pock · CVPR 2020

Feed-forward Propagation in Probabilistic Neural Networks with Categorical and Max Layers

A. Shekhovtsov, B. Flach · ICLR 2019

Stochastic Normalizations as Bayesian Learning

A. Shekhovtsov, B. Flach · ACCV 2018

Normalization of Neural Networks using Analytic Variance Propagation

A. Shekhovtsov, B. Flach · CVWW 2018

MPLP++: Fast, Parallel Dual Block-Coordinate Ascent for Dense Graphical Models

S. Tourani, A. Shekhovtsov, C. Rother, B. Savchynskyy · ECCV 2018

Feed-forward Uncertainty Propagation in Belief and Neural Networks

A. Shekhovtsov, B. Flach, M. Bušta · CoRR 2018

Generative Learning for Deep Networks

B. Flach, A. Shekhovtsov, O. Fikar · CoRR abs/1709.08524 2017

Scalable Full Flow with Learned Binary Descriptors

G. Munda, A. Shekhovtsov, P. Knöbelreiter, T. Pock · GCPR 2017

Maximum Persistency via Iterative Relaxed Inference with Graphical Models

A. Shekhovtsov, P. Swoboda, B. Savchynskyy · PAMI 2017

End-to-End Training of Hybrid CNN-CRF Models for Stereo

P. Knöbelreiter, C. Reinbacher, A. Shekhovtsov, T. Pock · CVPR 2017

Complexity of Discrete Energy Minimization Problems

M. Li, A. Shekhovtsov, D. Huber · ECCV 2016

Joint M-Best-Diverse Labelings as a Parametric Submodular Minimization

A. Kirillov, A. Shekhovtsov, C. Rother, B. Savchynskyy · NIPS 2016

Solving Dense Image Matching in Real-Time using Discrete-Continuous Optimization

A. Shekhovtsov, C. Reinbacher, G. Graber, T. Pock · CVWW 2016

Maximum Persistency via Iterative Relaxed Inference with Graphical Models

A. Shekhovtsov, P. Swoboda, B. Savchynskyy · CVPR 2015

Higher Order Maximum Persistency and Comparison Theorems

A. Shekhovtsov · CVIU (SI on Inference & Learning of Graphical Models) 2014

Maximum Persistency in Energy Minimization

A. Shekhovtsov · CVPR 2014

Exact and Partial Energy Minimization in Computer Vision

A. Shekhovtsov · PhD Thesis, CTU in Prague 2013

Curvature Prior for MRF-Based Segmentation and Shape Inpainting

A. Shekhovtsov, P. Kohli, C. Rother · DAGM/OAGM 2012

A Distributed Mincut/Maxflow Algorithm Combining Path Augmentation and Push-Relabel

A. Shekhovtsov, V. Hlaváč · IJCV 2012

On Partial Optimality by Auxiliary Submodular Problems

A. Shekhovtsov, V. Hlaváč · Control Systems and Computers, 2011(2), special issue 2011

Joint Image GMM and Shading MAP Estimation

A. Shekhovtsov, V. Hlaváč · ICPR 2010

A Lower Bound by One-against-all Decomposition for Potts Model Energy Minimization

A. Shekhovtsov, V. Hlaváč · Computer Vision Winter Workshop 2008

On Partial Optimality in Multi-label MRFs

P. Kohli, A. Shekhovtsov, C. Rother, V. Kolmogorov, P. Torr · ICML 2008

A Discrete Search Method for Multi-modal Non-Rigid Image Registration

A. Shekhovtsov, J.D. Garcia-Arteaga, T. Werner · NORDIA workshop @ CVPR 2008

Efficient MRF Deformation Model for Non-Rigid Image Matching

A. Shekhovtsov, I. Kovtun, V. Hlaváč · Computer Vision and Image Understanding 2008

Unified Framework for Semiring-Based Arc Consistency and Relaxation Labeling

T. Werner, A. Shekhovtsov · Computer Vision Winter Workshop, St. Lambrecht, Austria 2007

Supermodular Decomposition of Structural Labeling Problem

A. Shekhovtsov · Control Systems and Computers #1, 2006, pp.39-48 (in Russian) 2006

A Higher Order MRF-Model for Stereo-Reconstruction

B. Flach, D. Schlesinger, A. Shekhovtsov · Pattern Recognition, LNCS vol. 3175, pp. 440–446 2004

Technical Reports

Reports not superseded by / merged into a published paper above.

LP-Relaxation of Binarized Energy Minimization

A. Shekhovtsov, V. Kolmogorov, P. Kohli, V. Hlaváč, C. Rother, P. Torr · Research Report CTU–CMP–2007–27, CTU in Prague 2008

Efficient MRF Deformation Model for Image Matching

A. Shekhovtsov, I. Kovtun, V. Hlaváč · Research Report CTU–CMP–2006–08, CTU in Prague 2006