Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Picture for Cengiz Pehlevan

Asymptotics of representation learning in finite Bayesian neural networks


Jun 11, 2021
Jacob A. Zavatone-Veth, Abdulkadir Canatar, Cengiz Pehlevan

* 12+28 pages, 2+1 figures; v2: fix typo in (8) and add references 

  Access Paper or Ask Questions

Learning Curves for SGD on Structured Features


Jun 09, 2021
Blake Bordelon, Cengiz Pehlevan

* Fixed Typo in A.7 

  Access Paper or Ask Questions

Out-of-Distribution Generalization in Kernel Regression


Jun 04, 2021
Abdulkadir Canatar, Blake Bordelon, Cengiz Pehlevan


  Access Paper or Ask Questions

Exact priors of finite neural networks


May 18, 2021
Jacob A. Zavatone-Veth, Cengiz Pehlevan

* 12+11 pages, 4 figures; v2: references and figures added 

  Access Paper or Ask Questions

Activation function dependence of the storage capacity of treelike neural networks


Jul 21, 2020
Jacob A. Zavatone-Veth, Cengiz Pehlevan

* 18 pages, 2 figures 

  Access Paper or Ask Questions

Statistical Mechanics of Generalization in Kernel Regression


Jul 07, 2020
Abdulkadir Canatar, Blake Bordelon, Cengiz Pehlevan

* 30 pages, 10 figures 

  Access Paper or Ask Questions

Associative Memory in Iterated Overparameterized Sigmoid Autoencoders


Jun 30, 2020
Yibo Jiang, Cengiz Pehlevan

* ICML 2020 

  Access Paper or Ask Questions

Blind Bounded Source Separation Using Neural Networks with Local Learning Rules


Apr 11, 2020
Alper T. Erdogan, Cengiz Pehlevan

* ICASSP 2020 

  Access Paper or Ask Questions

Supervised Deep Similarity Matching


Mar 10, 2020
Shanshan Qin, Nayantara Mudur, Cengiz Pehlevan


  Access Paper or Ask Questions

Spectrum Dependent Learning Curves in Kernel Regression and Wide Neural Networks


Feb 19, 2020
Blake Bordelon, Abdulkadir Canatar, Cengiz Pehlevan


  Access Paper or Ask Questions

A Closer Look at Disentangling in $β$-VAE


Dec 11, 2019
Harshvardhan Sikka, Weishun Zhong, Jun Yin, Cengiz Pehlevan

* Presented at the 53rd Asilomar Conference on Signals, Systems, and Computers 

  Access Paper or Ask Questions

Structured and Deep Similarity Matching via Structured and Deep Hebbian Networks


Oct 11, 2019
Dina Obeid, Hugo Ramambason, Cengiz Pehlevan

* Accepted for publication in NeurIPS 2019 

  Access Paper or Ask Questions

Neuroscience-inspired online unsupervised learning algorithms


Sep 06, 2019
Cengiz Pehlevan, Dmitri B. Chklovskii

* Accepted for publication in IEEE Signal Processing Magazine 

  Access Paper or Ask Questions

A Spiking Neural Network with Local Learning Rules Derived From Nonnegative Similarity Matching


Feb 18, 2019
Cengiz Pehlevan

* ICASSP 2019 

  Access Paper or Ask Questions

Efficient Principal Subspace Projection of Streaming Data Through Fast Similarity Matching


Aug 06, 2018
Andrea Giovannucci, Victor Minden, Cengiz Pehlevan, Dmitri B. Chklovskii

* 9 pages, 4 figures 

  Access Paper or Ask Questions

Why do similarity matching objectives lead to Hebbian/anti-Hebbian networks?


Jul 11, 2017
Cengiz Pehlevan, Anirvan Sengupta, Dmitri B. Chklovskii

* Accepted for publication in Neural Computation 

  Access Paper or Ask Questions

Blind nonnegative source separation using biological neural networks


Jun 01, 2017
Cengiz Pehlevan, Sreyas Mohan, Dmitri B. Chklovskii

* Accepted for publication in Neural Computation 

  Access Paper or Ask Questions

Self-calibrating Neural Networks for Dimensionality Reduction


Dec 11, 2016
Yuansi Chen, Cengiz Pehlevan, Dmitri B. Chklovskii

* 2016 Asilomar Conference on Signals, Systems and Computers 

  Access Paper or Ask Questions

A Normative Theory of Adaptive Dimensionality Reduction in Neural Networks


Jan 26, 2016
Cengiz Pehlevan, Dmitri B. Chklovskii

* Advances in Neural Information Processing Systems (NIPS), 2015 

  Access Paper or Ask Questions

Optimization theory of Hebbian/anti-Hebbian networks for PCA and whitening


Nov 30, 2015
Cengiz Pehlevan, Dmitri B. Chklovskii

* Annual Allerton Conference on Communication, Control, and Computing (Allerton) 2015 

  Access Paper or Ask Questions

A Hebbian/Anti-Hebbian Network for Online Sparse Dictionary Learning Derived from Symmetric Matrix Factorization


Nov 30, 2015
Tao Hu, Cengiz Pehlevan, Dmitri B. Chklovskii

* 2014 Asilomar Conference on Signals, Systems and Computers. v2: fixed a typo in equation 23 

  Access Paper or Ask Questions

A Hebbian/Anti-Hebbian Network Derived from Online Non-Negative Matrix Factorization Can Cluster and Discover Sparse Features


Mar 02, 2015
Cengiz Pehlevan, Dmitri B. Chklovskii

* 2014 Asilomar Conference on Signals, Systems and Computers 

  Access Paper or Ask Questions