no code implementations • 13 Mar 2023 • Zhongwen Zhang, Yuri Boykov
We propose "collision cross-entropy" as a robust alternative to Shannon's cross-entropy (CE) loss when class labels are represented by soft categorical distributions y.
no code implementations • 26 Jan 2023 • Zhongwen Zhang, Yuri Boykov
Maximization of mutual information between the model's input and output is formally related to "decisiveness" and "fairness" of the softmax predictions, motivating these unsupervised entropy-based criteria for clustering.
no code implementations • CVPR 2021 • Zhongwen Zhang, Dmitrii Marin, Maria Drangova, Yuri Boykov
We are interested in unsupervised reconstruction of complex near-capillary vasculature with thousands of bifurcations where supervision and learning are infeasible.
no code implementations • CVPR 2019 • Zhongwen Zhang, Egor Chesakov, Dmitrii Marin, Yuri Boykov
We propose a new geometric regularization principle for reconstructing vector fields based on prior knowledge about their divergence.