In development

Flow Matching

Researching novel flow matching techniques for scientific data in collaboration with SciLifeLab

Home/Projects/Flow Matching

Problem & Impact

Researching novel flow matching techniques to improve generative modeling for scientific applications.

Tech Stack

Python
PyTorch

Key Features

  • Development of novel continuous normalizing flow architectures for scientific data
  • Strategic research partnership with SciLifeLab leveraging their infrastructure and datasets

Timeline

Start Date

September 2025

Current Phase

Testing and benchmarking phase

Upcoming Milestones

  • Complete flow matching architecture implementation
  • Validate flow matching models on diverse datasets (images, S&P data, and air quality data)
  • Benchmark performance against established scientific evaluation metrics

Affiliations

SciLifeLab

Maintenance & Deployment

Research project with active development by KTH AI Society team in collaboration with SciLifeLab

Contributors

FN

Felix Nannesson Meli

Research Lead

DD

Divyayan Dey

Researcher