JEPA
I am interested in designing ways to learn more effective and dense representations in neural network training. I find Joint Embedding Predictive Architectures to be a promising approach to learning.
Homepage
Marco Petri
PhD Student in Information Technology and Software Engineer
I am a PhD student and deep learning researcher at Politecnico di Milano, and a professional software engineer. Currently, I am pursuing a Doctor of Philosophy, a research experience lasting three to four years. My interests span both basic and applied research, specifically targeting Technology Readiness Levels 2 to 4.
Regarding research, I am currently interested in representation learning and the use of embeddings for prediction. Specifically, I am exploring Joint Embedding Predictive Architectures and their applications to various domains, including robotics and computer vision.
Research Interests
Robotics
I am interested in how artificial intelligence can enable a robot to move and perform various tasks, from picking up objects to navigating freely through space. In this context, I find world models particularly interesting.
Time series
I am interested in studying various phenomena that exhibit strong temporal dependencies and patterns. Economic, industrial, and many other types of data, along with their underlying generation processes, are profoundly interesting.
Latest Posts
The 10 most recent posts, including both tech and opinions.