IRIS-HEP Fellow: Paulius Balčiūnas



Fellowship dates: Feb – Apr, 2026

Home Institution: Vilnius University Faculty of Mathematics and Informatics


Project: Lossy Data Compression for Simulated CMS Pile-up Pixel Subdetector Datasets

CMS datasets pose significant challenges for data storage and processing due to their large size and complexity. The primary goal of this project is to investigate the possibilities of applying Vector Quantization Variational Autoencoder (VQ-VAE) model for CMS pileup dataset lossy compression. VQ-VAE model aims to encode the data into a compact latent representation, which can later be reconstructed using VQ-VAE learned codebooks with minimal information loss



More information: My project proposal

Mentors:
  • Valdas Rapševičius (Vilnius University)

  • Tomas Raila (Vilnius University)

Presentations and Publications
Current Status
A placeholder for status updates

Contact me: