George Atia

Rediscovering Deep Neural Networks Through Finite-State Distributions

2018 · arXiv (Cornell University)

We propose a new way of thinking about deep neural networks, in which the linear and non-linear components of the network are naturally derived and justified in terms of principles in probability theory. In particular, …

A Differentiable Approach to Combinatorial Optimization using Dataless\n Neural Networks

2022 · arXiv (Cornell University)

The success of machine learning solutions for reasoning about discrete\nstructures has brought attention to its adoption within combinatorial\noptimization algorithms. Such approaches generally rely on supervised learning\nby leveraging datasets of the combinatorial structures of interest drawn …

BOSS: Bidirectional One-Shot Synthesis of Adversarial Examples

2021 · arXiv (Cornell University)

The design of additive imperceptible perturbations to the inputs of deep classifiers to maximize their misclassification rates is a central focus of adversarial machine learning. An alternative approach is to synthesize adversarial examples from scratch …

Explainable Adversarial Attacks on Coarse-to-Fine Classifiers

2025 · arXiv (Cornell University)

Traditional adversarial attacks typically aim to alter the predicted labels of input images by generating perturbations that are imperceptible to the human eye. However, these approaches often lack explainability. Moreover, most existing work on adversarial …