Artur Back de Luca
Ph.D. Student in Computer Science
University of Waterloo
About
I am a fourth-year Ph.D. student in Computer Science at the University of Waterloo and a member of the WatCL lab. I am advised by Kimon Fountoulakis, and my research explores reasoning in neural networks by focusing on their ability to learn algorithmic tasks. I am also broadly interested in theoretical machine learning and graphs. I previously studied AI and Robotics (M.Sc.) at Sapienza University of Rome, and Mechanical Engineering (B.Sc.) at the Federal University of Santa Catarina.
During Summer 2025, I joined Amazon New York as an Applied Scientist Intern on the SCOT team, where I worked with Ruijun Ma and Youxin Zhang on inbound event forecasting. In 2022, I interned at Huawei’s Noah’s Ark Lab, where I worked with Guojun Zhang on federated learning and domain generalization, and with Yingxue Zhang on invariant graph representations.
News
-
Apr 2026
I was awared the NSERC Canada Graduate Research Scholarship (CGRS D) and the President’s Graduate Scholarship.
-
Apr 2026
-
Mar 2026
I will be joining Amazon again for another summer internship in Applied Research at SCOT.
-
Feb 2026
Our paper on exact graph algorithm execution was accepted at the Workshop on Latent & Implicit Thinking at ICLR 2026.
-
Feb 2026
New paper on exact graph algorithm execution with graph neural networks.
-
Jan 2026
I was awarded the David R. Cheriton Graduate Scholarship.
-
Dec 2025
I was awarded the TD Layer 6 Graduate Scholarship in Data and AI.
-
Sep 2025
Our paper on exact execution of algorithmic instructions was accepted at NeurIPS 2025.
-
Jul 2025
Our paper on exact permutation learning was accepted at the HiLD workshop at ICML 2025.
-
May 2025
I received the Ontario Graduate Scholarship (OGS) and the President’s Graduate Scholarship.
-
Feb 2025
I am joining Amazon for a summer internship in Applied Research at SCOT
-
Feb 2025
New paper on Transformers and algorithmic computation.
-
Jun 25, 2024
I presented our work on local graph clustering at the Fields Institute for the Workshop on Complex Networks in Banking and Finance.
-
May 2024
Our paper on looped transformers for graph algorithms was accepted at ICML 2024.
-
Feb 2024
New paper on looped transformers for graph algorithms.
Publications
-
Learning to Execute Graph Algorithms Exactly with Graph Neural Networks
-
Learning to Add, Multiply, and Execute Algorithmic Instructions Exactly with Neural Networks
-
Exact Learning of Permutations for Nonzero Binary Inputs with Logarithmic Training Size and Quadratic Ensemble Complexity
-
Positional Attention: Expressivity and Learnability of Algorithmic Computation
-
Simulation of Graph Algorithms with Looped Transformers
-
Local Graph Clustering with Noisy Labels
-
Mitigating Data Heterogeneity in Federated Learning with Data Augmentation