Adrian Vladu

I am a Postdoctoral Associate working with Alina Ene and Lorenzo Orecchia. I received my PhD from MIT Math in 2017, where I benefited from the brilliant advising of Jonathan Kelner and Aleksander Mądry.

I work on multiple aspects of convex (and non-convex) optimization. My work so far attempted to combine tools from convex geometry and continuous optimization in order to obtain improved algorithms for classical discrete problems.

I am also interested in various aspects of Deep Learning, particularly understanding what enables models to generalize, and how the training methods we use affect generalization.


News
  • (6/24) I talked at STOC about the adaptive complexity of submodular maximization. Three (!) different groups of researchers obtained similar results using amazingly different techniques. To get an idea, take a look at the slides.
  • (6/12) I talked at ICML about a provably efficient version of IRLS, which finally explains this method for approximating max flow in a more principled (and slightly more efficient) way.
Publications
Improved Convergence for ℓ and ℓ1 Regression via Iteratively Reweighted Least Squares
Alina Ene, Adrian Vladu
International Conference on Machine Learning (ICML 2019)
[talk video] [slides] [poster] [code]
Submodular Maximization with Matroid and Packing Constraints in Parallel
Alina Ene, Huy L. Nguyễn, Adrian Vladu
ACM SIGACT Symposium on Theory of Computing (STOC 2019)
[slides] [poster]
Towards Deep Learning Models Resistant to Adversarial Attacks
Aleksander Mądry, Aleksandar Makelov, Ludwig Schmidt, Dimitris Tsipras, Adrian Vladu
International Conference on Learning Representations (ICLR 2018)
Oral presentation at the Principled Approaches to Deep Learning workshop, ICML 2017
Matrix Scaling and Balancing via Box Constrained Newton's Method and Interior Point Methods
Michael B. Cohen, Aleksander Mądry, Dimitris Tsipras, Adrian Vladu
Symposium on Foundations of Computer Science (FOCS 2017)
[slides]
Multidimensional Binary Search for Contextual Decision-Making
Ilan Lobel, Renato Paes Leme, Adrian Vladu
ACM Conference on Economics and Computation (EC 2017)
Invited to the special issue
Appears in Operations Research
[slides] [poster]
Almost-Linear-Time Algorithms for Markov Chains and New Spectral Primitives for Directed Graphs
Michael B. Cohen, Jonathan A. Kelner, John Peebles, Richard Peng, Anup B. Rao, Aaron Sidford, Adrian Vladu
ACM SIGACT Symposium on Theory of Computing (STOC 2017)
Invited to the special issue
Appears in Highlights of Algorithms 2018
Negative-Weight Shortest Paths and Unit Capacity Minimum Cost Flow in Õ(m10/7 log W) Time.
Michael B. Cohen, Aleksander Mądry, Piotr Sankowski, Adrian Vladu
ACM-SIAM Symposium on Discrete Algorithms (SODA 2017)
Appears in Highlights of Algorithms 2017
[slides]
Tight Bounds for Approximate Carathéodory and Beyond.
Vahab S. Mirrokni, Renato Paes Leme, Adrian Vladu, Sam Chiu-wai Wong
International Conference on Machine Learning (ICML 2017)
Oral presentation at the Informs Optimization Society Conference 2016
[slides]
Faster Algorithms for Computing the Stationary Distribution, Simulating Random Walks, and More
Michael B. Cohen, Jonathan A. Kelner, John Peebles, Richard Peng, Aaron Sidford, Adrian Vladu
Symposium on Foundations of Computer Science (FOCS 2016)
[slides]
Improved Parallel Algorithms for Spanners and Hopsets.
Gary L. Miller, Richard Peng, Adrian Vladu, Shen Chen Xu
ACM symposium on Parallelism in Algorithms and Architectures (SPAA 2015)
How to Elect a Leader Faster than a Tournament
Dan Alistarh, Rati Gelashvili, Adrian Vladu
ACM Symposium on Principles of Distributed Computing (PODC 2015)
Online Ranking for Tournament Graphs
Claire Mathieu, Adrian Vladu
International Workshop on Approximation and Online Algorithms (WAOA 2010)
[slides]

Other Works
A Parallel Double Greedy Algorithm for Submodular Maximization
Alina Ene, Huy L. Nguyễn, Adrian Vladu
Phenotypic profiling reveals that Candida albicans opaque cells represent a metabolically specialized cell state compared to default white cells
(by contribution) Iuliana Ene, Matthew Lohse, Adrian Vladu, Joachim Morschhäuser, Alexander Johnson, Richard Bennett
mBIO (2016)


( * Per mathematical tradition, authors are written in alphabetic order unless stated otherwise.)