Giambattista Parascandolo

(or "GB", Giovanni, ...)

I'm a Research Scientist at OpenAI, leading a team trying to solve reasoning in neural networks.

I did a PhD in machine learning at the Max Planck Institute for Intelligent Systems and ETH Zürich, as a doctoral fellow of the Center for Learning Systems, where I was co-supervised by Bernhard Schölkopf and Thomas Hofmann. The topic of my PhD was out-of-distribution generalization with deep learning.
I was also an Ellis student.

During my PhD, I spent time at DeepMind in London, and Google X in Mountain View (California).
Also, as the final step of a professorship interview process at MIT started in 2020, I gave an invited seminar talk. My research plan proposed to use GPT for reasoning [slides] and was called nonsense by a large part of the committee :)

I just want to understand the mind and create intelligence.


Selected Publications (for the full list Google Scholar)




Deep Learning Beyond the Training Distribution

Giambattista Parascandolo
Thesis committee: Bernhard Schoelkopf, Yoshua Bengio, Thomas Hofmann.

PhD Thesis - ETH Zurich and Max Planck Institute for Intelligent Systems Tuebingen (2021)


Neural Symbolic Regression that Scales

Luca Biggio*, Tommaso Bendinelli *, Alexander Neitz, Aurelien Lucchi, Giambattista Parascandolo

ICML 2021

[paper] [code]

Learning explanations that are hard to vary

Giambattista Parascandolo*, Alexander Neitz*, Antonio Orvieto, Luigi Gresele, Bernhard Schölkopf (* equal contribution)

ICLR 2021

[paper] [code]

A teacher-student framework to distill future trajectories

Alexander Neitz*, Giambattista Parascandolo*, Bernhard Schölkopf (* equal contribution)

ICLR 2021

[paper] [code]

A Seq2Seq approach to Symbolic Regression

Luca Biggio, Tommaso Bendinelli, Aurelien Lucchi, Giambattista Parascandolo

NeurIPS 2020 Workshop ”Learning Meets Combinatorial Algorithms”
NeurIPS 2020 Workshop ”Knowledge Representation & Reasoning Meets Machine Learning”


 Divide-and-Conquer Monte Carlo Tree Search For Goal-Directed Planning

Giambattista Parascandolo*, Lars Buesing*, Josh Merel, Leonard Hasenclever, John Aslanides, Jessica B. Hamrick, Nicolas Heess, Alexander Neitz, Theophane Weber (* equal contribution)


[paper] [supporting-website]

Adaptive Skip Intervals: Temporal Abstraction for Recurrent Dynamical Models

Alexander Neitz, Giambattista Parascandolo, Stefan Bauer, Bernhard Schölkopf

NeurIPS 2018

[paper]  [code]

Generalization in anti-causal learning

Niki Kilbertus*, Giambattista Parascandolo*, Bernhard Schölkopf (* alphabetical order)

NeurIPS 2018 Workshop on Critiquing and correcting trends in machine learning


Learning Independent Causal Mechanisms

Giambattista Parascandolo, Niki Kilbertus, Mateo Rojas-Carulla, Bernhard Schölkopf

ICML 2018

also at: NIPS 2017 Workshop on Learning Disentangled Representations - Spotlight presentation

[paper] [bibtex]

Tempered Adversarial Networks

M. S. M. Sajjadi, Giambattista Parascandolo, Arash Mehrjou, Bernhard Schölkopf

ICML 2018

also at: ICLR 2018 Workshop


Avoiding Discrimination Through Causal Reasoning

Niki Kilbertus, Mateo Rojas-Carulla, Giambattista Parascandolo, Moritz Hardt, Dominik Janzing, Bernhard Schölkopf

NeurIPS 2017

[paper] [bibtex] [poster] [in the press: MPI]

ConvWave: Searching for Gravitational Waves with Fully Convolutional Neural Nets

Timothy Gebhard*, Niki Kilbertus*, Giambattista Parascandolo, Ian Harry, Bernhard Schölkopf (* equal contribution)

NeurIPS 2017 Workshop on Deep Learning for Physical Sciences

[paper] [bibtex] [code] [poster]

Sound Event Detection in Multichannel Audio Using Spatial and Harmonic Features,

Sharath Adavanne, Giambattista Parascandolo, Pasi Pertilä, Toni Heittola, Tuomas Virtanen

Workshop on Detection and Classification of Acoustic Scenes and Events (DCASE) 2016 - 1st place in the challenge.


Acoustic Scene Classification Using Convolutional Neural Networks

Michele Valenti, Aleksandr Diment, Giambattista Parascandolo, Stefano Squartini, Tuomas Virtanen

International Joint Conference on Neural Networks (IJCNN) 2017 - Best Student Poster Intel Award

IEEE AASP Challenge on Detection and Classification of Acoustic Scenes and Events (DCASE2016)


Low-latency sound source separation using deep neural networks

Gaurav Naithani, Giambattista Parascandolo, Tom Barker, Niels Henrik Pontoppidan, Tuomas Virtanen

IEEE Global Conference on Signal and Information Processing (GlobalSIP) 2016


Convolutional Recurrent Neural Networks for Polyphonic Sound Event Detection

Emre Cakir*, Giambattista Parascandolo*, Toni Heittola, Heikki Huttunen, Tuomas Virtanen (*equal contribution)

IEEE/ACM Transactions on Audio, Speech, and Language Processing (Journal)


Recurrent neural networks for polyphonic sound event detection in real life recordings

Giambattista Parascandolo, Heikki Huttunen, Tuomas Virtanen

ICASSP 2016 - Best student paper of its special session




g b  [a t]  o p e n a i  [d o t]  c o m

Italy is a beautiful place. If you're going there on vacation, do yourself a favor and avoid Milan.