Arash A. Amini

I work on problems in statistics, machine learning, optimization, etc. Some of my current research interests are:

High-dimensional statistics
Network models
Unsupervised and semisupervised learning
Clustering and community detection
Causal graphical models
Kernel methods
Representation learning
Optimization
Quantitative Finance

I did my PhD at UC Berkeley with Martin Wainwright and a post-doc at University of Michigan with Liza Levina and Long Nguyen. You can learn more about me from my CV.

Do you want to ask about research opportunities? Please refer to this FAQ first.

My office is at 8105F Math Sciences Building.
Office Hours: No office hours (Summer-Fall 2024)

Email:

Considering sending an email to request a PTE for class enrollment? Please make sure to review this information first.

news

Dec 11, 2023	Excited to be a panelist on Reconsidering Overfitting in the Age of Overparameterized Models at NeurIPS 2023!
Mar 4, 2022	The `bcdc` package for Bayesian community detection with covariates is out. Check out the paper for more details.
Feb 5, 2021	Check out the `hsbm` package for hierarchical network clustering.
Dec 30, 2020	Check out the `nett` package for community detection!
Dec 30, 2020	New paper out: Adjusted chi-square test for degree-corrected block models.

selected publications

Preprint

Sharp Bounds for Poly-GNNs and the Effect of Graph Noise

Vinas, Luciano, and Amini, Arash A.

Preprint 2024

Abs arXiv

We investigate the classification performance of graph neural networks with graph-polynomial features, poly-GNNs, on the problem of semi-supervised node classification. We analyze poly-GNNs under a general contextual stochastic block model (CSBM) by providing a sharp characterization of the rate of separation between classes in their output node representations. A question of interest is whether this rate depends on the depth of the network k, i.e., whether deeper networks can achieve a faster separation? We provide a negative answer to this question: for a sufficiently large graph, a depth k>1 poly-GNN exhibits the same rate of separation as a depth k=1 counterpart. Our analysis highlights and quantifies the impact of “graph noise” in deep GNNs and shows how noise in the graph structure can dominate other sources of signal in the graph, negating any benefit further aggregation provides. Our analysis also reveals subtle differences between even and odd-layered GNNs in how the feature noise propagates.
AOS

Adjusted Chi-Square Test for Degree-Corrected Block Models

Zhang, Linfan, and Amini, Arash A.

The Annals of Statistics 2023

HTML PDF Code
NeurIPS

Target alignment in truncated kernel ridge regression

Amini, Arash A, Baumgartner, Richard, and Feng, Dai

In NeurIPS 2022

arXiv Code
AOS

Concentration of kernel matrices with application to kernel spectral clustering

Amini, Arash A., and Razaee, Zahra S.

The Annals of Statistics 2021

arXiv HTML PDF
JMLR

Optimal bipartite network clustering

Zhou, Zhixin, and Amini, Arash A.

Journal of Machine Learning Research 2020

arXiv HTML PDF
NeurIPS

Globally optimal score-based learning of directed acyclic graphs in high-dimensions

Aragam, Bryon, Amini, Arash, and Zhou, Qing

In Advances in Neural Information Processing Systems 2019

HTML PDF Supp
JMLR

Analysis of spectral clustering algorithms for community detection: the general bipartite setting

Zhou, Zhixin, and Amini, Arash A.

Journal of Machine Learning Research 2019

arXiv HTML PDF
AOS

On semidefinite relaxations for the block model

Amini, A. A., and Levina, E.

The Annals of Statistics 2018

arXiv HTML PDF Code