SimTech is set to welcome Sarah Baskin, a highly promising second-year student at Brandeis University in Waltham, Massachusetts (USA), as part of the prestigious DAAD RISE Fellowship program. Baskin, who arrives at the end of May, 2024, will spend around three months working on interpretability methods for contrastive learning in neural networks.
The DAAD RISE (Research Internships in Science and Engineering) program provides opportunities for undergraduate students from North America, the UK, and Ireland to undertake research internships at universities and research institutions across Germany. The program is funded by the German Federal Foreign Office and through various cooperative partnerships.
Baskin, who is double-majoring in Computer Science and Linguistics, was chosen from a competitive pool of applicants by Konstantin Nikolaou, a doctoral researcher at the Institute for Computational Physics (ICP) under the guidance of Professor Christian Holm. Nikolaou selected Baskin because of her background in Natural Language Processing (NLP) and her enthusiasm for exploring the inner workings of neural networks, making her an ideal candidate for the program.
“I am so excited to be participating in this program, as it will give me the chance to be involved in real academic research. I've been studying neural networks in my classes and have produced some personal projects related to classification, but I haven't yet seen what that work looks like in a research setting. Outside of my interest in the computational sphere, I also like to crochet and go on long bike rides”, says Sarah Baskin.
During her internship, Baskin will be working alongside Konstantin Nikolaou and other experts at the ICP, contributing to ongoing research projects aimed at enhancing the interpretability of neural networks, a critical area of study in the broader field of computational physics. Her work will focus on developing methods to better understand how neural networks learn and make decisions - an essential step toward creating more transparent and reliable artificial intelligence systems.
Baskin's arrival marks the second time a DAAD RISE Fellow joins SimTech. Mario Gaimann from the research group led by Miriam Klopotek also recruited a DAAD RISE Fellow, demonstrating the program's significance in fostering international collaboration and bringing new talent to research institutions in Germany.
Konstantin Nikolaou, who was instrumental in selecting Baskin for the DAAD RISE Fellowship, has been working on innovative approaches to neural network theory. His master's thesis, titled "Collective Variables in Data-centric Neural Network Training," examined entropy and other interpretable properties to analyze the training process of neural networks. Nikolaou's ongoing research seeks to understand the underlying processes that drive neural networks' learning capabilities, with the goal of understanding and improving current neural network models.
The DAAD RISE program plays a vital role in promoting academic exchange and fostering the development of emerging researchers like Baskin. As she begins her internship in Stuttgart, the research community eagerly anticipates the contributions she will make to the ongoing exploration of neural networks and their applications.