Muhammad Mobeen · the RingMuhammad Mobeen

Research

Where the loop starts.

The first arc of the loop. Papers only, each linked to its venue.

IMVIP 2024September 2024

Leveraging GANs for Synthetic Tabular Data Generation: A Platform-Centric Solution to Data Scarcity

Muhammad MobeenAwais AfzalReeda SaeedTayyaba ArshadMuhammad Asad

26th Irish Machine Vision and Image Processing Conference
IET Conference Proceedings, Volume 2024, Issue 10

Introduces Synthium AI, a platform-centric approach to data scarcity in machine learning. The system uses generative adversarial networks (CTGAN for tabular data, DGAN for time series) inside a modular, automated workflow that produces high-fidelity synthetic datasets for teams that cannot access the real thing, whether for reasons of privacy, regulation, or simple availability.

DOI 10.1049/icp.2024.3311

Read the paperThe venture it became
IMVIP 2024 conference identity
The conference call in gallery view during the session. The largest tile is the room camera at the University of Limerick, showing a raked lecture theatre with attendees seated across several rows.
The talk in progress, screen shared over the conference call. The slide is titled Methodology and diagrams the conditional tabular GAN: a sampled value for a column conditions both a lookup into the data set and a conditional generator, while a discriminator, marked as used only in training, judges real records against generated ones. Participant thumbnails run down the right side.

The room at the University of Limerick, joined over the call

  • GANs
  • CTGAN
  • DGAN
  • Synthetic data
  • Tabular data
  • Time series

Everything indexed, including whatever lands after this page, is on Google Scholar. The author record is ORCID 0000-0002-4269-1909.

Working on something adjacent?

Synthetic data, agentic pipelines, generative video. If your problem lives near this work, comparing notes is cheap.

Book a callSee the ventures