Feb. 9, 2024, 5:44 a.m. | Beatrix M. G. Nielsen Anders Christensen Andrea Dittadi Ole Winther

cs.LG updates on arXiv.org arxiv.org

Diffusion models may be viewed as hierarchical variational autoencoders (VAEs) with two improvements: parameter sharing for the conditional distributions in the generative process and efficient computation of the loss as independent terms over the hierarchy. We consider two changes to the diffusion model that retain these advantages while adding flexibility to the model. Firstly, we introduce a data- and depth-dependent mean function in the diffusion process, which leads to a modified diffusion loss. Our proposed framework, DiffEnc, achieves a statistically …

advantages autoencoders computation cs.cv cs.lg diffusion diffusion model diffusion models encoder flexibility generative hierarchical improvements independent loss process stat.ml terms variational autoencoders

Research Scholar (Technical Research)

@ Centre for the Governance of AI | Hybrid; Oxford, UK

HPC Engineer (x/f/m) - DACH

@ Meshcapade GmbH | Remote, Germany

Data Science Advisor

@ Blue Yonder | Hamburg

Data Analyst

@ Sinch | São Paulo, State of São Paulo, Brazil - Remote

Data Engineer - Híbrido

@ SGS | Callao, Peru

Senior Analytics Engineer Brazil

@ Hiflylabs | Blumenau, Hungary