Morteza Mahdiani
Machine Learning Researcher
building multimodal systems.
I work on machine learning across vision, language, and audio, with an interest in how intelligent systems perceive, represent, and understand the world.
-
Our study of imagery vividness was accepted as a spotlight at NeurIPS 2026. We found that vision models captured much of the agreement across people about which scenes were easiest to imagine vividly.
-
We introduced VISGate at the NeurIPS UniReps workshop. The model uses image features to predict both brain responses and caption embeddings.