Explainable & Safe AI

Farzaneh Heidari

PhD Student · Université de Montréal

Mila – Quebec AI Institute · Advised by Guillaume Rabusseau

Explainable and safe AI, through tensor networks and multilinear methods.

Farzaneh Heidari

About

The multilinear thread

Many objects we care about in machine learning — feature attributions, cooperative-game values, learned representations, quantum circuits — can be written as multilinear maps. Tensor networks give a compact, tractable handle on them, and that idea runs through my work on explainability and AI safety.

More about me →

Research interests

    Selected work

    Featured projects

    All projects →

    Publications

    Recent papers

    All publications →

    News

    Latest updates

    All news →