Publications de Igor Krawczuk | EPFL Graph Search

Graph generative deep learning models with an application to circuit topologies

Modern integrated circuits are tiny yet incredibly complex technological artifacts, composed of millions and billions of individual structures working in unison. Managing their complexity and facilitating their design drove part of the co-evolution of mode ...

EPFL2024

Going beyond Compositions, DDPMs Can Produce Zero-Shot Interpolations

Volkan Cevher, Grigorios Chrysos, Igor Krawczuk

Denoising Diffusion Probabilistic Models (DDPMs) exhibit remarkable capabilities in image generation, with studies suggesting that they can generalize by composing latent factors learned from the training data. In this work, we go further and study DDPMs t ...

2024

Distributed Extra-Gradient With Optimal Complexity And Communication Guarantees

Volkan Cevher, Kimon Antonakopoulos, Igor Krawczuk

We consider monotone variational inequality (VI) problems in multi-GPU settings where multiple processors/workers/clients have access to local stochastic dual vectors. This setting includes a broad range of important problems from distributed convex minimi ...

2023

Finding Actual Descent Directions For Adversarial Training

Volkan Cevher, Thomas Michaelsen Pethick, Igor Krawczuk, Leello Tadesse Dadi, Fabian Ricardo Latorre Gomez

Adversarial Training using a strong first-order adversary (PGD) is the gold standard for training Deep Neural Networks that are robust to adversarial examples. We show that, contrary to the general understanding of the method, the gradient at an optimal ad ...

2023

DiGress: Discrete Denoising diffusion for graph generation

Pascal Frossard, Volkan Cevher, Igor Krawczuk, Clément Arthur Yvon Vignac, Bohan Wang

This work introduces DiGress, a discrete denoising diffusion model for generating graphs with categorical node and edge attributes. Our model utilizes a discrete diffusion process that progressively edits graphs with noise, through the process of adding or ...

2023

Proximal Point Imitation Learning

Volkan Cevher, Luca Viano, Igor Krawczuk, Angeliki Kamoutsi

This work develops new algorithms with rigorous efficiency guarantees for infinite horizon imitation learning (IL) with linear function approximation without restrictive coherence assumptions. We begin with the minimax formulation of the problem and then o ...

2022

A Computational Turn in Policy Process Studies: Coevolving Network Dynamics of Policy Change

Igor Krawczuk

The past three decades of policy process studies have seen the emergence of a clear intellectual lineage with regard to complexity. Implicitly or explicitly, scholars have employed complexity theory to examine the intricate dynamics of collective action in ...

WILEY-HINDAWI2022

Filling gaps in trustworthy development of AI

Igor Krawczuk

American Association for the Advancement of Science2021

Multi-ReRAM synapses for artificial neural network training

Elmira Shahrabi, Igor Krawczuk, Yusuf Leblebici, Irem Boybat Kara, Carlo Ricciardi, Cecilia Giovinazzo, Evangelos Eleftheriou, Abu Sebastian

Metal-oxide-based resistive memory devices (ReRAM) are being actively researched as synaptic elements of neuromorphic co-processors for training deep neural networks (DNNs). However, device-level non-idealities are posing significant challenges. In this wo ...

IEEE2019

Effect of metal buffer layer and thermal annealing on HfOx-based ReRAMs

Elmira Shahrabi, Igor Krawczuk, Yusuf Leblebici, Jury Sandrini, Behnoush Attarimashalkoubeh

In this paper, we investigate different methods and approaches in order to improve the electrical characteristics of Pt/HfOx/TiN ReRAM devices. We discuss the improvement of the ReRAM electrical characteristics after the insertion of a Hf and Ti buffer lay ...

Ieee2016