Linda Petrini

AI Resident

Google Brain Montreal


I'm currently a Google AI Resident in the Brain team in Montreal. My mentor is Laurent Dinh and we are working on generative models for unspervised object discovery.

In the past, I complited my master's at University of Amsterdam, where I worked with Rianne van den Berg (AMLAB) on priors for VAEs and semi-supervised learning, and wrote my thesis during an internship at Mila, where I worked with Devon Hjelm on fundamental principles in Representation Learning for Zero Shot Learning.

In my free time I practice yoga, acroyoga and read a lot.


  • Machine Learning
  • Generative Models
  • Representation Learning


  • MSc in Artificial Intelligence, 2019

    University of Amsterdam

  • BSc in Mathematics, 2017

    Universita' di Milano-Bicocca

Recent Publications

Locality and Compositionality in Zero-Shot Learning

In this work we study locality and compositionality in the context of learning representations for Zero Shot Learning (ZSL). In order …

Mutual Information heatmaps as a tool for interpretability

Interpretability is a major concern when applying deep convolutional neural networks to image data. In this work we introduce a tool …