Matthias Becker

Dr. Matthias Becker is a researcher at the German Center for Neurodegenerative Diseases (DZNE), leading projects at the intersection of artificial intelligence, molecular generation, and privacy-preserving data utilization. His work focuses on advancing generative modeling techniques and deploying them in collaborative initiatives across institutions.

• Developed DrugDiff, a latent diffusion model for controlled generation of drug-like molecules with adjustable molecular properties.
• Contributed to the PriSyn project, pioneering synthetic data generation via differentially private machine learning to enable privacy-compliant medical research.
• Active in Swarm Learning, applying decentralized AI to high-dimensional flow cytometry, medical imaging, and large language model fine-tuning.
• Leverages HPC/GPU clusters and evaluates FPGA technologies for energy-efficient computational pipelines.

His research emphasizes sustainability in AI by optimizing model architectures for reduced energy consumption and data efficiency, while addressing critical challenges in neurodegenerative disease studies and personalized medicine.

Publications include a 2024 study in Journal of Cheminformatics analyzing autoencoder architectures for molecular data. Collaborations span the CISPA Helmholtz Center, QuantPi, and Hewlett Packard Enterprise.