Dr. Laufer-Goldshtein’s PhD research in Electrical Engineering at Bar-Ilan was in the emerging fields of data science and machine learning. Her work presented several novel paradigms for solving classical speech processing problems while exploiting a broad range of statistical methods and data-driven models. She developed novel high-performance methods for audio source separation and localization in challenging noise and reverberation conditions, based on geometric learning over manifolds and simplexes. Her research achievements were acknowledged by several awards, including the Adams Fellowship and the Wolf Award, both given to only few PhD students annually from the entire country.
As a postdoctoral researcher in the Computer Science and Artificial Intelligence Lab at MIT, Dr. Laufer-Goldshtein addresses new problems of highly structured data analysis, including representation learning, domain adaptation and multi-source fusion. She uses the tools she gained in manifold learning and statistical inference, combined with new theoretical and analytical techniques of graph theory, geometric deep learning and optimization.
Dr. Laufer-Goldshtein is applying these mathematical tools to molecular modeling for drug discovery, which deals with graph-structured data with complex dependencies and interactions between the atoms comprising the molecules. She aims to utilize these models for molecular property prediction and generation, developing viable, highly effective methods for speeding up drug discovery.