NEWS
ELLIS unit Amsterdam Celebrates Academic Achievements of Two Recently Graduated Honours Students
ELLIS unit Amsterdam community acknowledges the remarkable achievements of two of our MSc Honours students whose thesis research endeavours have made impacts in their respective fields of study within the AI landscape, Larissa de Ruijter and Rahel Beloch.
Equivariant Amortized Inference of Poses for Cryo-EM
Under the supervision of Taco Cohen and Gabriele Cesa (ELLIS Unit Amsterdam) and Ozan Öktem (ELLIS Unit Stockholm), Larissa de Ruijter conducted her thesis on cryogenic electron microscopy (cryo-EM). Her research, entitled “Equivariant Amortized Inference of Poses for Cryo-EM,” delves into the application of geometric deep learning to the cryo-EM reconstruction problem.
Through the ELLIS Unit Amsterdam MSc Honours Programme, Larissa had the privilege to collaborate and learn at the prestigious KTH – Royal Institute of Technology in Stockholm. Larissa expanded her knowledge on the cryo-EM reconstruction problem and related topics, thanks to interesting discussions with Ozan Öktem and other researchers at the KTH.
Reflecting on her experience, she emphasized the significance of networking and exploring diverse research topics, highlighting how conversations with researchers from differing- or related fields can shape ones own research trajectory and aid in solving questions within ones own research field. With an open mind, she remains undecided about her future path, contemplating both further academic pursuits and opportunities in industry.
Interactive: On Model and Data Relevance for Curricula in Teacher-Student Setups
Rachel Beloch, a dedicated honours student embarked on a journey of self-discovery and scholarly exploration under the guidance of Dieuwke Hupkes and Maartje ter Hoeve from the University of Amsterdam (ELLIS unit Amsterdam), alongside Emmanuel Dupoux from Ecole des Hautes Etudes en Sciences Sociales (EHESS) in Paris (ELLIS unit Paris). Titled “Interactive: On Model and Data Relevance for Curricula in Teacher-Student Setups,” her master’s thesis delved into revealing underlying curricula when teaching a language interactively to language models in a teacher-student framework to reveal how various data characteristics influence a language model’s understanding of language.
During the honours journey, Rahel had the privilege to collaborate with Emmanuel and his lab at Ecole Normale Supérieure (ENS) in Paris and to network with other researchers who work on similar topics. Her most important takeaway from the program and research visit is the competence to break her research topic down into the essential components and to present it convincingly and comprehensibly to others. Through this immersive experience, she gained invaluable insights into the relevance of her research in both academic and industrial settings, ultimately crystallising her decision to pursue a career in industry.
Now, Rahel has embarked on a new chapter as a Data Scientist at REWE Group, a prominent German retail and tourism company where she will be working on use cases involving generative AI for language.