Image
Prof. Dmitry Vetrov
Professor of Computer Science
School of Computer Science & Engineering
Email Address
dvetrov@constructor.university
Research Interests
Prof. Dmitry Vetrov's research focuses on combining Bayesian framework with Deep Learning models.
Selected Publications
Recent publications
- To Stay or Not to Stay in the Pre-train Basin: Insights on Ensembling in Transfer Learning (2023)
- Star-Shaped Denoising Diffusion Probabilistic Models (2023)
- Entropic Neural Optimal Transport via Diffusion Processes (2023)
- StyleDomain: Efficient and Lightweight Parameterizations of StyleGAN for One-shot and Few-shot Domain Adaptation (2023)
- UnDiff: Unsupervised Voice Restoration with Unconditional Diffusion Model (2023)
- HIFI++: A Unified Framework for Bandwidth Extension and Speech Enhancement (2023)
- MARS: Masked Automatic Ranks Selection in Tensor Decompositions (2023)
- HyperDomainNet: Universal Domain Adaptation for Generative Adversarial Network (2022)
- Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes (2022)
- FFC-SE: Fast Fourier Convolution for Speech Enhancement (2022)
- Variational Autoencoders for Precoding Matrices with High Spectral Efficiency (2022)
- On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay (2021)
- Leveraging Recursive Gumbel-Max Trick for Approximate Inference in Combinatorial Spaces (2021)
- On Power Laws in Deep Ensembles (2020)
Publications on Scopus