Cyprus University of Technology · Limassol
Generative models, built on probability.
The GenML Laboratory studies how machines learn to generate data and how they can say how sure they are. We work on diffusion and flow models, Bayesian deep learning and stochastic inference, and we take them to autonomous driving, video and language.
Twelve years of statistical machine learning in Cyprus.
The lab was founded in 2014 as the Statistical Machine Learning Research Laboratory and is now the GenML Laboratory. It sits in the Department of Electrical Engineering and Computer Science and Engineering at the Cyprus University of Technology.
Our method is Bayesian at heart. We design models whose uncertainty is part of the answer, and we build the inference algorithms that make them scale to modern data and modern hardware.
- Founded2014
- VenuesNeurIPS · ICML · ICCV · AAAI · AISTATS
- Compute4 × NVIDIA H100
- FundingHorizon Europe · MSCA
Six lines of work that share one toolkit.
Scalable generative modelling
Diffusion, flow matching and few-step generators for images, video and trajectories.
Bayesian deep learning
Priors, posteriors and nonparametric models that make deep networks report what they do not know.
MCMC and Langevin inference
Sampling algorithms for posterior inference and for steering pretrained generators at test time.
Stochastic and sparse attention
Attention operators with random-measure foundations, built to be efficient and uncertainty-aware.
Trustworthy AI
Calibration, faithfulness testing and hallucination detection, backed by statistical guarantees.
Multi-agent LLM systems
Orchestration of language-model agents that plan, critique and improve each other.
- Autonomous driving
- Video understanding and generation
- Natural language processing
- Recommendation systems
- Financial inference
- Engineering surrogates
European research projects.
Human-aligned decision-making for autonomous vehicles. The lab develops a dual architecture that pairs a slow vision-language model with a fast diffusion-based planner, together with work on standards compliance.
Remote attendance of live-streamed performing arts with a stronger sense of presence: gaze-contingent video super-resolution, an LLM-based virtual companion and a VR virtual foyer.
Trustworthy AI infrastructure for Cyprus. The lab contributes the project's AI Registry tooling, built on multi-agent orchestration.
Circular power-electronics converter design. The lab hosts a doctoral candidate working on world models for power-module health monitoring and lifetime estimation.
Twenty years in generative machine learning, with a background in Bayesian statistics from the University of Miami and a former Senior Research Associate post at Imperial College London. Area Chair at ICLR, Senior Program Committee member at AAAI, and member of the National AI Taskforce of the Republic of Cyprus.
Artist-specific generative models and controllable style.
Parameter-efficient adapters for video generation.
Neural surrogates for ship-hull hydrodynamics, with Dromon Bureau of Shipping.
Human-aligned decision-making for autonomous vehicles (Task 4.2).
Collaborative perception for connected autonomous vehicles (Task 3.3).
Agentic AI and the Indian Buffet of Agents (IBoA) framework.
Anytime-valid statistical testing of whether deep-network explanations are faithful.
Flow matching with per-channel Gamma scale-mixture noise.
Join us. We welcome PhD and MSc applicants with a strong background in probability, statistics or deep learning. Write to the Lab Director with a CV and a short note on the problems that interest you.
Limassol, Cyprus