Adversarial examples have attracted significant attention in machine learning, but the reasons
for their existence and pervasiveness remain unclear. We demonstrate that adversarial …

A recent line of work has uncovered a new form of data poisoning: so-called backdoor attacks.
These attacks are particularly dangerous because they do not affect a network's behavior …

The study of adversarial robustness has so far largely focused on perturbations bound in $\ell_p
$-norms. However, state-of-the-art models turn out to be also vulnerable to other, more …

The body naturally and continuously secretes sweat for thermoregulation during sedentary
and routine activities at rates that can reflect underlying health conditions, including nerve …

We show that simple spatial transformations, namely translations and rotations alone,
suffice to fool neural networks on a significant fraction of their inputs in multiple image …

Conventional plain-film radiographiy may be used as a screening method for various
pathological conditions of the sinonasal cavities. However, CT scanning remains the study of …

An important goal in deep learning is to learn versatile, high-level feature representations of
input data. However, standard networks' representations seem to possess shortcomings that…

We show that the basic classification framework alone can be used to tackle some of the most
challenging tasks in image synthesis. In contrast to other state-of-the-art approaches, the …

Clonal hematopoiesis of indeterminate potential (CHIP) describes a widespread expansion
of genetically variant hematopoietic cells that increases exponentially with age and is …

BACKGROUND Central nervous system Langerhans cell histiocytosis (CNS‐LCH) brain
involvement may include mass lesions and/or a neurodegenerative disease (LCH‐ND) of …