Home

Selected Roles

2023- Postdoctoral Research Fellow,
Mila AI Institute,
Montreal, Canada

2021-23 Postdoctoral Research Fellow,
Harvard University,
Boston, United States

2021 AI Research Scientist (visiting),
Naver Labs,
Seoul, South Korea

2020-21 Postdoctoral Research Associate,
University of Cambridge,
Cambridge, UK

Education

2017-20 PhD in Artificial Intelligence
and Robotics,
University of Cambridge,
Cambridge, UK

2013-17 BEng in Artificial Intelligence
and Software Engineering,
University of Edinburgh,
Edinburgh, UK

Biography

I am currently a Postdoctoral Research Fellow in Yoshua Bengio‘s group, at the Mila AI Institute. My current work focuses on generative AI, learning representations, domain understanding, and generalization in deep learning, as well as applications in drug discovery and robotics.

I received my BEng in Artificial Intelligence and Software Engineering in 2017 from the University of Edinburgh, where I graduated summa cum laude, obtaining the Class Award and Howe Prize for the best performance in Artificial Intelligence.

During my Ph.D. and Postdoctoral research at the University of Cambridge‘s Biologically Inspired Robotics Laboratory (BIRL) under the supervision of Prof. Fumiya Iida, I worked on the application of machine learning models for sensory perception and action in robotics systems.

In 2021 I joined Never Corp. as a visiting AI Research Scientist, pursuing research in the topics of learning representations, domain understanding, robustness, and generalization.

Finally, I am a former Postdoctoral Fellow at Harvard University and Dana Farber, where I worked closely with Dr. Ming-Ru Wu to boost computational biomedical research and to use machine learning to answer basic questions in tumor biology and immunotherapy, with the ultimate goal of improving cancer treatment quality and efficiency.

Research Interests

I am interested in the interplay of foundation AI research and its applications to critical domains. I am particularly interested in the topics of generative AI, multi-modal perception, and learning representations as well as the downstream implications of these research domains in fairness, explicability, and (safe) inference. I am also very enthusiastic about the impact of machine learning in the sciences and engineering. I am fortunate to be involved in several of these domains including Vision, Computational Biology (deep RL, bio-sequence design), and robotics research (Visual and Tactile action/perception, multi-modal time series).

For collaborations please contact me at luca.scimeca@mila.quebec.

Highlights

Shortcut Learning in Deep Neural Networks. Which cues will your model choose to learn?

Luca Scimeca |
July 31, 2023|

One of our publications in ICLR 2022 investigates the issue of shortcut cue …

Luca Scimeca |
November 25, 2020|

Robotics Palpation (University of Cambridge, 2019) This paper proposes a framework to investigate …

Luca Scimeca |
July 15, 2020|

Bayesian Exploration for Morphology-Action Co-optimization (University of Cambridge, 2019) Morphology been shown to …

Luca Scimeca |
June 20, 2020|

Dynamic Robot Control for Expressive Piano Playing (University of Cambridge, 2019) Piano is …

Luca Scimeca |
August 30, 2019|

Coding with Neural Networks (Massachusetts Institute of Technology, 2019) The project was developed …

Luca Scimeca |
November 19, 2018|

Multiscale Object Recognition with Inception Neural Networks (University of Cambridge, 2018) Multiscale object …