Publications

Research contributions in machine learning, computational biology and artificial intelligence for scientific discovery.

Research Focus

My research explores how modern machine learning methods can extract meaningful representations from complex biological and chemical data.

Current interests include foundation models for biology, self-supervised learning, molecular representation learning, and AI systems capable of supporting scientific workflows.

Machine Learning Computational Biology Foundation Models Drug Discovery Self-Supervised Learning

Published Work

A Self-Supervised Transfer Learning Approach for Collision Cross-Section Prediction

IWANN 2025
18th International Work-Conference on Artificial Neural Networks

Computational prediction of molecular properties is an important tool for supporting metabolite annotation. This work introduces a self-supervised transfer learning approach that learns molecular representations from large chemical databases and adapts them to collision cross-section prediction tasks.

The proposed framework improves cross-database generalization compared with traditional fingerprint-based approaches.

Self-Supervised Learning Transfer Learning Molecular Representation Learning

Poster

Training Deep Learning Neural Networks for Predicting CCS Using the METLIN-CCS Dataset

IWBBIO 2024
International Conference on Bioinformatics and Biomedical Engineering

This work investigates deep learning approaches for predicting collision cross-section values from molecular information using the METLIN-CCS dataset containing 61,863 experimentally measured values.

Different molecular representations were evaluated, including fingerprints, descriptors and combined feature representations, with deep neural networks achieving competitive prediction performance.

Deep Learning Computational Chemistry Biomedical AI

Paper

Under Review / In Preparation

Chemical Foundation Models Evaluation Project

Manuscript under peer review

Evaluation of chemical foundation models for molecular representation learning and downstream prediction tasks.

The work investigates how pretrained molecular models capture chemical information and how their representations transfer to scientific machine learning problems.

Foundation Models Chemical AI Representation Learning

Code Repository

Siamese Transfer Learning for Collision Cross-Section Prediction

Manuscript in preparation

Research extension exploring Siamese architectures and transfer learning approaches for improving molecular property prediction across different chemical datasets.

Siamese Networks Transfer Learning PyTorch

Research Profiles

More information about my publications and academic activity: