Ionatan Kuperwajs

Theory, Cognition, and AI
PhD Candidate in Neural Science
New York University

About

Welcome to my personal website! I'm currently a PhD student in the Center for Neural Science at New York University working with Wei Ji Ma. I received my B.A. in neuroscience, computer science, and mathematics from Macalester College, a liberal arts school located in St. Paul, MN. My research interests lie at the intersection of cognitive science and computer science. I primarily use tools from artificial intelligence and reinforcement learning to infer what algorithms people use to plan sequences of actions in complex environments. Previously in my undergraduate and graduate career, I've been fortunate enough to have worked on projects in a wide range of areas within neuroscience, including visual neuroscience and psychophysics, statistical analysis and visualization of fMRI data, connectomics and machine learning, and network science.

Curriculum Vitae (Updated September 2019)

As an aspiring computational neuroscientist coming from a liberal arts background, I've been interested in common identity issues that many scientists, especially in academia, confront. Taking a social/developmental psychology course at Macalester has allowed me to explore these issues utilizing a distinct appraoch. I wrote my final paper on identity development in academia, with the main goal being to take editorial pieces that have been written by scientists detailing their experiences with impostor syndrome, gender and racial disparity in academia, and student-advisor relationships, and integrate theoretical frameworks and experiments from the literature that describe these shared experiences by many researchers through a social identity lens. I strongly believe that promoting discourse around these issues is crucial within the academic community in order to find collaborative solutions for these problems. I hope to continue to be involved in this in some way in graduate school and beyond, and would like to thank Wei Ji Ma's Growing Up in Science effort as the source of inspiration for my work.

Identity Development in Academia (Paper)

Roots

My family history is a bit convoluted, but it's an important part of understanding who I am. I was born in Madrid, Spain, and my family moved to Seattle when I was 5. My dad's parents are originally Polish, but my grandfather moved to Argentina right before the start of WWII. Growing up in Buenos Aires, my dad decided to move to Israel at the age of 18, a common migration for Jews at the time as a part of the Zionist movement. Meanwhile, my mom's family is from Morocco, and my mother was born on the ship from Meknes to Tel-Aviv (she's actually named after the boat!). My parents met in Israel, after my dad had already been married once and had two kids. For reasons that are too long to go into here, my parents moved to Madrid and lived there for 18 years, where my sister and I were both born. The main result of all of this is that I'm a dual Spanish and American citizen, fluent in three languages (English, Spanish, and Hebrew), and have wonderful extended family all over the globe.

Why Neuroscience?

Understanding what algorithms the brain employs to navigate complex environments remains one of the largest challenges in contemporary science. As I began my education at Macalester College, I became increasingly interested in neuroscience not only for the scientific foundation of the field, but for its unique interdisciplinary standing among the hard sciences. At a fundamental level, conducting research in computational neuroscience allows me to apply quantitative methods to understand and characterize the computations that underlie perception and behavior. I have continually found myself compelled by problems that are ecologically relevant, but have an elegant theoretical solution. To be a successful neuroscientist, one must have the ability to think in a holistic manner, and that's something that's always appealed to me.

Oustide of Academia

Throughout my life, I've always played soccer competitively, starting from a young age at the club level all the way to the varsity team at Macalester. While I've been with the program, we've consistently been ranked at the national level, and have won the MIAC conference title as well as hosted in the NCAA tournament. Apart from playing soccer, I'm an enormous Real Madrid fan, and follow many other sports pretty intently, such as basketball and tennis. I'm also an aspiring landscape photographer who loves to travel and hike (check out my Instagram profile, linked on the last page, for some of my favorite shots). I'll eventually get around to creating a portfolio section on this site!

Academic Portfolio

Ph.D. in Neural Science (Systems, Cognition, and Computation Track)
I am currently pursuing my doctoral degree in neuroscience in the lab of Wei Ji Ma where I develop computational models of human planning in combinatorial games. In my first year, I rotated with Eero Simoncelli and completed coursework in mathematics, behavioral and cognitive neuroscience, and cellular neuroscience.

B.A. with Honors in Neuroscience, Computer Science, and Mathematics
I recently finished up my undergraduate degree, where I split my time between the neuroscience and mathematics, statistics, and computer science departments. Being at a liberal arts school allowed me to draw courses from these different departments that contributed in unique ways to my ability to move forward into my graduate work and attempt to answer the types of research questions that I find interesting. My undergraduate coursework largely consisted of neuroscience and neurobiology (Brain, Mind, and Behavior, Behavioral Neuroscience), mathematics/statistics (Linear Algebra, Probability, Combinatorics), and computer science (Algorithms, Machine Learning). I want to especially thank my outstanding advisor, Andrew Beveridge, for the mentorship and instillment of a love for mathematics.

Pre-Graduate Research Projects

Computational Visual Neuroscience Lab


As an undergraduate, I worked as a research assistant under Kendrick Kay at the University of Minnesota's Center for Magnetic Resonance Research during the academi year. The lab's primary goal is to understand how the human brain represents visual images and makes perceptual decisions about these images using a combined experimental and computational approach that seeks to develop models that characterize the stimulus transformations perfomed by the brain, mostly using fMRI measurements. I began by mainly testing common statistical processes in their data analysis pipeline for pre-processed high-resolution fMRI data in MATLAB. I also completed my honors thesis in conjunction with this lab, developing MATLAB tools for visualization and analysis of high-resolution fMRI data. We hope to make our code open source and available sometime in 2018. Stay tuned for updates!

Lab Website | Thesis Manuscript (PDF) | Thesis Defense Slides (PDF)

Turaga Lab


I spent the summer of 2017 at Janelia Research Campus working under Srini Turaga as part of the Janelia Undergraduate Scholars Program. Broadly speaking, the Turaga Lab develops machine learning methods to map the structure and function of neural circuits. My project was to further develop an existing framework for mapping neural connectivity in vivo from population activity measurements by calcium imaging combined with cellular resolution optogenetic activity perturbations. This is a fully Bayesian approach based on utilizing variational autoencoders to model spiking activity using discrete latent variables, low-dimensional latent common input, and sparse spike-and-slab generalized linear coupling between neurons. This model is adaptive to different optogenetic datasets as well as spike inference and fitting methods. Additionally, I worked on extensive data science techniques for interpreting the resultant model's output.

Lab Website | Interview | Slides (PPT) | Slides (PDF) | Poster (PDF)

LINK-Group


During the spring of my junior year I studied abroad in Budapest, Hungary, and took courses at Aquincum Institute of Technology, a program for North American undergraduates studying computer science, mathematics, and software engineering. One of the courses I took there was "Structure and Dynamics of Complex Networks", where P├ęter Csermely was a professor. He quickly noted my interest in interdisciplinary research, and invited me to join LINK-Group. Hosted at Semmelweis University, LINK-Group is a multidiscplinary cohort of investigators working to discover the topology and dynamics of different complex networks. Quite conveniently, they had a connectomics project that I became an integral part of. We worked on a dynamic network model to simulate global activity states of the C. elegans nervous system, which is well-defined in terms of neural connectivity within neuroscience literature. The long-term goal of the study was to find stable attractor states of the C. elegans "mind," and map those to distinct behavioral patterns. I focused on the mating behavior of the male connectome (posterior neurons), a smaller cortical sub-network that dictates known behaviors of the worm, in order to test parameters for the larger system on a manageable scale. I'm still involved in this project remotely, and will be contributing to the set of publications.

Lab Website | Slides (PPT) | Slides (PDF)

Computational Neuroimaging Lab


In the summer of 2016, I worked at New York University under David Heeger as part of their Summer Undergraduate Research Program (NSF REU). The lab generally focuses on visual perception and neuroscience by developing computational theories of neural processing in the visual cortex of the brain and testing predictions of those theories using psychophysical (perceptual psychology) measurements of human vision and neuroimaging (fMRI) measurements of human brain activity. I specifically implemented mathematical and perceptual models of optic flow, including a novel version based on motion without movement. This allowed us to determine that the human visual system estimates heading direction and angular velocity from the evolution of the optic flow field over time in conjunction with the instantaneous velocity of each vector in the flow field, a revision of the currently established theory within neuroscience and vision literature. The academic paper for this work is currently being written with collaboration of other lab members.

Lab Website | Interview | Slides (PPT) | Slides (PDF) | Poster (PDF)