Search Results - "Mitliagkas, Ioannis"

Joint Power and Admission Control for Ad-Hoc and Cognitive Underlay Networks: Convex Approximation and Distributed Implementation by Mitliagkas, Ioannis, Sidiropoulos, Nicholas D., Swami, Ananthram

“…Power control is important in interference-limited cellular, ad-hoc, and cognitive underlay networks, when the objective is to ensure a certain quality of…”
Get full text

Asynchrony begets momentum, with an application to deep learning by Mitliagkas, Ioannis, Ce Zhang, Hadjis, Stefan, Re, Christopher

“…Asynchronous methods are widely used in deep learning, but have limited theoretical justification when applied to non-convex problems. We show that running…”
Get full text

Feature learning as alignment: a structural property of gradient descent in non-linear neural networks by Beaglehole, Daniel, Mitliagkas, Ioannis, Agarwala, Atish

“…Understanding the mechanisms through which neural networks extract statistics from input-label pairs through feature learning is one of the most important…”
Get full text

An Empirical Study of Pre-trained Model Selection for Out-of-Distribution Generalization and Calibration by Naganuma, Hiroki, Hataya, Ryuichiro, Mitliagkas, Ioannis

“…In out-of-distribution (OOD) generalization tasks, fine-tuning pre-trained models has become a prevalent strategy. Different from most prior work that has…”
Get full text

Performative Prediction with Neural Networks by Mofakhami, Mehrnaz, Mitliagkas, Ioannis, Gidel, Gauthier

“…Performative prediction is a framework for learning models that influence the data they intend to predict. We focus on finding classifiers that are…”
Get full text

Generating Tabular Data Using Heterogeneous Sequential Feature Forest Flow Matching by Akazan, Ange-Clément, Jolicoeur-Martineau, Alexia, Mitliagkas, Ioannis

“…Privacy and regulatory constraints make data generation vital to advancing machine learning without relying on real-world datasets. A leading approach for…”
Get full text

Compositional Risk Minimization by Mahajan, Divyat, Pezeshki, Mohammad, Mitliagkas, Ioannis, Ahuja, Kartik, Vincent, Pascal

“…In this work, we tackle a challenging and extreme form of subpopulation shift, which is termed compositional shift. Under compositional shifts, some…”
Get full text

Optimal transport meets noisy label robust loss and MixUp regularization for domain adaptation by Fatras, Kilian, Naganuma, Hiroki, Mitliagkas, Ioannis

“…It is common in computer vision to be confronted with domain shift: images which have the same class but different acquisition conditions. In domain adaptation…”
Get full text

Additive Decoders for Latent Variables Identification and Cartesian-Product Extrapolation by Lachapelle, Sébastien, Mahajan, Divyat, Mitliagkas, Ioannis, Lacoste-Julien, Simon

“…We tackle the problems of latent variables identification and ``out-of-support'' image generation in representation learning. We show that both are possible…”
Get full text

No Wrong Turns: The Simple Geometry Of Neural Networks Optimization Paths by Guille-Escuret, Charles, Naganuma, Hiroki, Fatras, Kilian, Mitliagkas, Ioannis

“…Understanding the optimization dynamics of neural networks is necessary for closing the gap between theory and practice. Stochastic first-order optimization…”
Get full text

Convergence Analysis and Implicit Regularization of Feedback Alignment for Deep Linear Networks by Girotti, Manuela, Mitliagkas, Ioannis, Gidel, Gauthier

“…We theoretically analyze the Feedback Alignment (FA) algorithm, an efficient alternative to backpropagation for training neural networks. We provide…”
Get full text

Solving Hidden Monotone Variational Inequalities with Surrogate Losses by D'Orazio, Ryan, Vucetic, Danilo, Liu, Zichu, Kim, Junhyung Lyle, Mitliagkas, Ioannis, Gidel, Gauthier

“…Deep learning has proven to be effective in a wide variety of loss minimization problems. However, many applications of interest, like minimizing projected…”
Get full text

Gradient penalty from a maximum margin perspective by Jolicoeur-Martineau, Alexia, Mitliagkas, Ioannis

“…A popular heuristic for improved performance in Generative adversarial networks (GANs) is to use some form of gradient penalty on the discriminator. This…”
Get full text

Empirical Analysis of Model Selection for Heterogeneous Causal Effect Estimation by Mahajan, Divyat, Mitliagkas, Ioannis, Neal, Brady, Syrgkanis, Vasilis

“…We study the problem of model selection in causal inference, specifically for conditional average treatment effect (CATE) estimation. Unlike machine learning,…”
Get full text

A Reproducible and Realistic Evaluation of Partial Domain Adaptation Methods by Salvador, Tiago, Fatras, Kilian, Mitliagkas, Ioannis, Oberman, Adam

“…Unsupervised Domain Adaptation (UDA) aims at classifying unlabeled target images leveraging source labeled ones. In this work, we consider the Partial Domain…”
Get full text

Improving Gibbs Sampler Scan Quality with DoGS by Mitliagkas, Ioannis, Mackey, Lester

“…The pairwise influence matrix of Dobrushin has long been used as an analytical tool to bound the rate of convergence of Gibbs sampling. In this work, we use…”
Get full text

YellowFin and the Art of Momentum Tuning by Zhang, Jian, Mitliagkas, Ioannis

“…Hyperparameter tuning is one of the most time-consuming workloads in deep learning. State-of-the-art optimizers, such as AdaGrad, RMSProp and Adam, reduce this…”
Get full text

Towards efficient representation identification in supervised learning by Ahuja, Kartik, Mahajan, Divyat, Syrgkanis, Vasilis, Mitliagkas, Ioannis

“…Humans have a remarkable ability to disentangle complex sensory inputs (e.g., image, text) into simple factors of variation (e.g., shape, color) without much…”
Get full text

Understanding Adam Requires Better Rotation Dependent Assumptions by Maes, Lucas, Zhang, Tianyue H, Jolicoeur-Martineau, Alexia, Mitliagkas, Ioannis, Scieur, Damien, Lacoste-Julien, Simon, Guille-Escuret, Charles

“…Despite its widespread adoption, Adam's advantage over Stochastic Gradient Descent (SGD) lacks a comprehensive theoretical explanation. This paper investigates…”
Get full text

Gradient Descent Is Optimal Under Lower Restricted Secant Inequality And Upper Error Bound by Guille-Escuret, Charles, Ibrahim, Adam, Goujaud, Baptiste, Mitliagkas, Ioannis

“…Advances in Neural Information Processing Systems 35 (2022): 24893-24904 The study of first-order optimization is sensitive to the assumptions made on the…”
Get full text