About me

My name is Sílvia Casacuberta Puig and I am a student from Barcelona pursuing a joint AB degree in Mathematics and Computer Science with a concurrent masters (SM) in Computer Science at Harvard University (Massachusetts, USA).

I am broadly interested in theoretical computer science, including matrix and graph algorithms, cryptography, data privacy, and algorithmic fairness.

Email: scasacubertapuig [at] college.harvard.edu. Google scholar.

Publications

Differential Privacy
Widespread Underestimation of Sensitivity in Differentially Private Libraries and How to Fix It, with Michael Shoemate, Salil Vadhan, and Connor Wagaman. Preliminary version at TPDP 2022 at ICML 2022. arXiv. Paper selected for one of the six spotlight talks.

Cryptography
SoK: Oblivious Pseudorandom Functions, with Julia Hesse and Anja Lehmann. In IEEE EuroS&P 2022. IACR ePrint. EuroS&P 2022 Distinguished Paper Award Finalist.

Matrix Algorithms
Faster Matrix Inversion and Rank Computation in Finite Fields, with Rasmus Kyng. In ITCS 2022. arXiv.

Natural Language Processing
Evaluating Word Embeddings with Categorical Modularity, with Karina Halevy and Damián Blasi. In Findings of ACL, 2021. arXiv, GitHub.

Interpretable Machine Learning
PCACE: A Statistical Approach to Ranking Neurons for CNN Interpretability, with Esra Suel and Seth Flaxman. In Responsible AI and DeepSpatial workshops at KDD 2021. arXiv.

Number Theory
On the Divisibility of Binomial Coefficients. Ars Math. Contemp. 19 (2020), 297-309. arXiv.

Manuscripts

Cryptography
Quantum and Classical Algorithms for Bounded Distance Decoding, with Richard Allen, Ratip Emin Berker, and Michael Gul. IACR ePrint, 2022.

Talks

Oblivious pseudorandom functions, EuroS&P 2022, Genoa (Italy), June 2022.

Widespread underestimation of sensitivities in DP libraries: vulnerabilities and solutions, Privacy Tools DP seminar in Boston (USA), May 2022. Joint with Connor Wagaman.

Faster sparse matrix inversion rank computation in finite fields, ITCS’22 (video), February 2022.

Faster sparse matrix inversion and rank computation in finite fields, Exact Computing Research Seminar at Université de Montpellier (France), January 2022.

Faster sparse matrix inversion and rank computation in finite fields, Google Zurich Algorithms & Optimization Group Seminar (virtual), December 2021.

Verification of differentially private algorithms, Harvard PRISE final presentation (virtual), August 2021.

PCACE: A Statistical Approach to Ranking Neurons for CNN Interpretability, Responsible AI workshop and DeepSpatial workshop at KDD 2021, August 2021.

Oblivious pseudorandom functions (OPRFs): origins and modern applications, IBM Research Security Seminar, Zurich (Switzerland), May 2021.

Why are p-adic numbers useful for fast linear algebra algorithms?, Harvard Math Table (poster, slides), December 2020.

On the divisibility of binomial coefficients, final presentation at the Research Science Institute at MIT (video), August 2017.

Teaching

AM 107: Graph Theory and Combinatorics, taught by Prof. Leslie Valiant (Spring 2022).

CS 120: Introduction to Algorithms and Their Limitations, taught by Prof. Salil Vadhan (Fall 2021).

CS 124: Data Structures and Algorithms, taught by Prof. Michael Mitzenmacher (Spring 2020).

Math 1b: Calculus, Series, and Differential Equations (Fall 2019).

Inclusivity

Co-president of Harvard Gender Inclusivity in Mathematics.

Co-academics director of Harvard Women in Computer Science.

Present and past affiliations

OpenDP (https://opendp.org/): June 2021 – present. Supervisor: Prof. Salil Vadhan.

IBM Research Zurich: January – May 2021. Supervisor: Dr. Julia Hesse.

ETH Zurich CS department: June – December 2020. Supervisor: Prof. Rasmus Kyng.

Harvard Radcliffe Institute: September 2019 – May 2020. Supervisor: Dr. Damián Blasi.

Imperial College Mathematics department: June – August 2019. Supervisor: Prof. Seth Flaxman.

Max Planck Institute for Quantum Optics: July 2018. Supervisor: Prof. Jordi Tura.

MIT Mathematics department: June – August 2017. Supervisor: Dr. Oscar Mickelin.