About
I am currently pursuing my PhD in computer science under the supervision of Prof. Alex Bronstein and Prof. Reuven Cohen. My research focuses on applying deep learning techniques to temporal data, with two primary areas of interest: understanding physiological responses to exercise and optimizing communication networks.
In the field of physiology, I develop deep learning methods to estimate VO2 and other metabolic parameters typically measured using expensive lab-based equipment, such as the Cosmed K5, by leveraging smartwatch data and video signals. My work involves two key tasks: First, modeling latent spaces for wearable data (e.g., smartwatches) and lab-based equipment (e.g., Cosmed K5) using neural network-parameterized ordinary differential equations (ODEs), and then building a “translation bridge” to align these spaces, enabling accurate estimation of physiological states. Second, modeling physiological changes through video signals of runners on the track, where we extract skeletal data and derive features such as running economy to gain deeper insights into metabolic states.
Additionally, I focus on predicting injuries and recognizing diverse activities using smartwatch data, even in the absence of explicitly labeled activity sessions. This facilitates valuable insights into performance, recovery, and overall physiological health. I also develop physiologically informed imputation techniques to handle missing data in wearable-derived datasets. Finally, I work with medical data to tackle critical health challenges, such as predicting COPD readmission probabilities and exploring the relationship between continuous glucose monitoring (CGM) and accelerometer data, uncovering meaningful patterns in patient health.
In communication networks, my research spans QUIC traffic analysis and teleoperations, focusing on optimizing network performance using real-world data. This includes developing machine learning models to predict cellular network behavior for teleoperated driving and analyzing encrypted QUIC traffic to understand server-client interactions.
My passion lies in using real-world datasets to solve practical problems. I merge my love of running with data science by modeling physiological trends using wearable data, while in communications, I emphasize leveraging real network traffic to enhance teleoperations and encrypted traffic analysis. Whether working with physiological, medical, or network data, my goal is to drive innovation through actionable insights and practical applications.
Outside of work
I am married to Shir Cohen, who also holds a PhD in computer science. I am a proud father to Romy, and we share our home with Zeus, our loyal German Shepherd. When I’m not spending time with my family, I can usually be found running, coaching amateur athletes in long-distance running, or reading books about the sport. As a passionate marathon runner, I am a proud member of the ASICS Front Runner team and hold a personal best of 2:43 from the Valencia Marathon.