Andreas Maggiori

I am a Postdoctoral Researcher at the Data Science Institute (DSI) at Columbia University, where I am working with Will Ma and Eric Balkanski. I am interested in the intersection of Online Decision Making, Machine Learning and Theoretical Computer Science with a particular focus on learning augmented algorithms.

Prior to that, I earned my PhD at EPFL, advised by Rudiger Urbanke and Ola Svensson. During my PhD, I visited the Simons Institute, UC Berkeley for 2 months for the Data-Driven Decision Processes program. I also interned twice at Google Zurich, hosted by Nikos Parotsidis and Ehsan Kazemi accordingly.

Before joining EPFL, I did my undergrad in Greece, in the Electrical and Computer Engineering department of National Technical University of Athens.

Email  /  Google Scholar  /  CV

• (October, 2023) Happy to share that I started my post-doc at Columbia University!
• (September, 2023) I successfully defended my PhD thesis, titled Beyond worst-case analysis, with or without predictions.
• (June, 2023) I gave an invited talk at INFORMS Applied Probability Society Conference in Nancy, France.
• (September, 2022) I was invited to give two talks at the theory seminar of UMass about the papers: The Primal-Dual Method for Learning Augmented Algorithms and Online and Consistent Correlation Clustering.
• (September, 2022) I am visiting the Simons Institute, UC Berkeley for the Data-Driven Decision Processes program.
• (May, 2022) I started a research internship in Google Zurich, hosted by Ehsan Kazemi. I will be working on efficient active learning for graphs.
• (April, 2022) Our paper Online and Consistent Correlation Clustering has been accepted to appear in ICML 2022! This is the result of my internship in Google in Summer 2021.

Online and Consistent Correlation Clustering
Vincent Cohen-Addad*, Silvio Lattanzi*, Andreas Maggiori*, Nikos Parotsidis*
ICML, 2022
slides/ talk

An Improved Analysis of Greedy for Online Steiner Forest∗
Etienne Bamas, Marina Drygala, Andreas Maggiori
SODA, 2022
arxiv/ slides

The Primal-Dual method for Learning Augmented Algorithms
Etienne Bamas*, Andreas Maggiori*, Ola Svensson*
NeurIPS, 2020 (oral talk)
arxiv/ talk/ code

Learning Augmented Energy Minimization via Speed Scaling
Etienne Bamas*, Andreas Maggiori*, Lars Rohwedder*, Ola Svensson*
NeurIPS, 2020 (spotlight)
arxiv/ slides/ code

Online Matching with General Arrivals
Buddhima Gamlath, Michael Kapralov, Andreas Maggiori, Ola Svensson, David Wajc
FOCS, 2019
arxiv/ David's talk

• I co-organized the ALPS (ALgorithms with PredictionS) workshop in EPFL in May 2022, along with Etienne Bamas and Adam Polak.

• I organized a study-group on how continuous optimization methods can be used to tackle combinatorial problems. The website of the study-group with notes and recorded lectures can be found here. (If you do not have an ETH account and you want to have access to the lecture videos, please drop me an email)

• I am/was teaching assistant for the following courses:
  • NTUA: Algorithms and Complexity, Discrete Mathematics
  • EPFL: Theory of Computation, Machine Learning, Learning Theory, Algorithms, Advanced Probability and Applications, Foundations of data science

• I completed my undergraduate thesis under the supervision of Dimitris Fotakis. Thesis: Using Machine Learning Techniques to Infer Players' Valuations in Online Ad Auctions.

• I am from Athens, Greece and enjoy gelato, rod fishing, kayaking, snowboarding and basketball. When I am in Athens, your chances of finding me at Amorgiano listening to Thanassis are pretty high.