Topographic models
Deep neural networks that simulate the spatial and functional organization of the brain.
Nhut Truong
Computational Cognitive Science
Object Vision Group
University of Trento
I am a postdoctoral researcher working in the Object Vision Group led by professor Stefania Bracci at the University of Trento.
My research is at the intersection of AI and cognitive science, aiming to use knowledge from one domain to benefit the other. Specifically, I use deep learning to model aspects of human cognition, such as cortical topography and the number sense, as well as to study the alignment between models and humans.
Before that, I completed my PhD at the same university, advised by professor Uri Hasson, with a thesis titled Modeling Cognition by Pruning and Topography-Learning in Deep Neural Networks.
News
3 Aug 2026
I started my postdoc with professor Stefania Bracci, using topographic models and fMRI to study action representation in the visual cortex.
12 Jul 2026
New paper — Beyond topography: Topographic regularization improves robustness and reshapes representations in convolutional neural networks — was accepted in Neurocomputing.
24 Apr 2026
I successfully defended my PhD thesis, Modeling Cognition by Pruning and Topography-Learning in Deep Neural Networks!
Research (Google Scholar)
Human-AI alignment
Using pruning and explainable AI to identify the visual features that align deep neural network representations with human similarity judgments.
Numerosity in AI models
How deep neural networks encode human number sense.
Cognitive Computational Neuroscience (CCN) 2025, NeurIPS Workshop 2024