Grigorios Chrysos

Cette personne a quitté l’EPFL

À propos
Confidentialité
Mentions légales

Graph Chatbot

Publications associées (29)

Page 2 sur 2

Robustness in deep learning: The good (width), the bad (depth), and the ugly (initialization)

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Zhenyu Zhu

We study the average robustness notion in deep neural networks in (selected) wide and narrow, deep and shallow, as well as lazy and non-lazy training settings. We prove that in the under-parameterized setting, width has a negative effect while it improves ...

2022

Cluster-guided Image Synthesis with Unconditional Models

Grigorios Chrysos

Generative Adversarial Networks (GANs) are the driving force behind the state-of-the-art in image generation. Despite their ability to synthesize high-resolution photo-realistic images, generating content with on-demand conditioning of different granularit ...

IEEE COMPUTER SOC2022

Single-pass Detection of Jailbreaking Input in Large Language Models

Volkan Cevher, Grigorios Chrysos, Yongtao Wu, Elias Abad Rocamora

Defending aligned Large Language Models (LLMs) against jailbreaking attacks is a challenging problem, with existing approaches requiring multiple requests or even queries to auxiliary LLMs, making them computationally heavy. Instead, we focus on detecting ...

2025

Federated Learning under Covariate Shifts with Generalization Guarantees

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Thomas Michaelsen Pethick

This paper addresses intra-client and inter-client covariate shifts in federated learning (FL) with a focus on the overall generalization performance. To handle covariate shifts, we formulate a new global model training paradigm and propose Federated Impor ...

Journal of Machine Learning Research Inc.2023

Controlling the Complexity and Lipschitz Constant improves Polynomial Nets

Volkan Cevher, Grigorios Chrysos, Fabian Ricardo Latorre Gomez, Zhenyu Zhu

While the class of Polynomial Nets demonstrates comparable performance to neural networks (NN), it currently has neither theoretical generalization characterization nor robustness guarantees. To this end, we derive new complexity bounds for the set of Coup ...

2022

Robust NAS under adversarial training: benchmark, theory, and beyond

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Yongtao Wu

Recent developments in neural architecture search (NAS) emphasize the significance of considering robust architectures against malicious data. However, there is a notable absence of benchmark evaluations and theoretical guarantees for searching these robus ...

2024

Efficient local linearity regularization to overcome catastrophic overfitting

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Elias Abad Rocamora

Catastrophic overfitting (CO) in single-step adversarial training (AT) results in abrupt drops in the adversarial test accuracy (even down to 0%). For models trained with multi-step AT, it has been observed that the loss function behaves locally linearly w ...

2024

Generalization Properties of NAS under Activation and Skip Connection Search

Volkan Cevher, Grigorios Chrysos, Fanghui Liu, Zhenyu Zhu

Neural Architecture Search (NAS) has fostered the automatic discovery of stateof- the-art neural architectures. Despite the progress achieved with NAS, so far there is little attention to theoretical guarantees on NAS. In this work, we study the generaliza ...

2022

Regularization of polynomial networks for image recognition

Volkan Cevher, Grigorios Chrysos, Bohan Wang

Deep Neural Networks (DNNs) have obtained impressive performance across tasks, however they still remain as black boxes, e.g., hard to theoretically analyze. At the same time, Polynomial Networks (PNs) have emerged as an alternative method with a promising ...

2023