Dr. Stefano Bruno

I am a Postdoctoral Research Associate at UNIST (South Korea). I am part of the InnoCORE programme under the AI-driven Space Solar initiative.

My research interests lie at the interface of Machine Learning, Generative AI, and Probability. I design algorithms with provable guarantees to tackle fundamental challenges in deep learning and generative modelling.

Previously, I was a Postdoctoral Research Associate in the School of Mathematics at the University of Edinburgh (UK). During my time in Edinburgh, I secured a £500,000 UK–South Korea Collaborative R&D grant, funded by Innovate UK, to support my research in collaboration with Neurolabs. The project has also raised an additional £415,000 in South Korea, funded by the Korea Institute for Advancement of Technology.

From November 2024 to April 2025, I worked as a Senior Researcher at Neurolabs, a UK-based Computer Vision startup. In April 2025, Neurolabs secured a $7.8 million Series A funding round.

I completed my PhD in Mathematics at the University of Bath (UK), under the supervision of Prof. Hendrik Weber, as part of the Probability Laboratory (Prob-L@B) and the EPSRC Centre for Doctoral Training in Statistical Applied Mathematics. My doctoral research was in Stochastic Analysis, with particular emphasis on Stochastic Partial Differential Equations.

Email:

sbruno@unist.ac.kr (UNIST, South Korea)

Office:

301-2, 112 Building, 50, UNIST-gil, Ulsan, 44919 Republic of Korea.

Affiliation:

Preprint:

Flatness-Aware Stochastic Gradient Langevin Dynamics.

Stefano Bruno, Youngsik Hwang, Jaehyeon An, Sotirios Sabanis, Dong-Young Lim.

Available on ArXiv. First version: October 2025. Submitted. Code.

Publications:

Wasserstein Convergence of Score-based Generative Models under Semiconvexity and Discontinuous Gradients.

Stefano Bruno, Sotirios Sabanis.

Transactions on Machine Learning Research (TMLR), ISSN 2835-8856.Accepted for publication: September 2025. First version on ArXiv: May 2025.

On diffusion-based generative models and their error bounds: The log-concave case with full convergence estimates.

Stefano Bruno, Ying Zhang, Dong-Young Lim, Ömer Deniz Akyildiz, Sotirios Sabanis.

Transactions on Machine Learning Research (TMLR), ISSN 2835-8856. Accepted for publication: February 2025. First version on ArXiv: November 2023.

Optimal regularity in time and space for stochastic porous medium equations.

Stefano Bruno, Benjamin Gess, Hendrik Weber.

Annals of Probability, Vol. 50, No. 6, 2288-2343. November 2022.

Grants:

Innovate UK: circa £500,000. Additional funding (over £415,000) will be provided by the Korea Institute for Advancement of Technology as part of the UK-South Korea collaborative R&D grant.

Project: "AI innovation in the supply chain of consumer packaged-goods for recognising objects in retail execution, supply chain management and smart factories: using novel diffusion-based optimisation algorithms and diffusion-based generative models". A brief description of the project can be found here.

London Mathematical Society: £3,000. Support for organising the Hackathon Workshop on Generative Modeling.
UoE School of Mathematics Knowledge Exchange fund: £1,000. Support for organising the Hackathon Workshop on Generative Modeling.

Hackathon on Generative AI:

I was the lead organiser of the Hackathon Workshop on Generative Modeling during the Isaac Newton Institute-Alan Turing Institute satellite programme on "Diffusions in machine learning: Foundations, generative models and non-convex optimisation". Two industry-inspired challenges were provided by Amazon (Edinburgh).

Reviewer: Transactions on Machine Learning Research (TMLR).

PhD Defense Examiners: Prof. Franco Flandoli (SNS Pisa) and Prof. Alexander Cox (University of Bath, UK).

Profiles: LinkedIn, Google Scholar, ResearchGate.