Towards Transferable Adversarial Attacks on Image and Video Transformers.

The transferability of adversarial examples across different convolutional neural networks (CNNs) makes it feasible to perform black-box attacks, resulting in security threats for CNNs. However, fewer endeavors have been made to investigate transferable attacks for vision transformers (ViTs), which achieve superior performance on various computer vision tasks. Unlike CNNs, ViTs establish relationships between patches extracted from inputs by the self-attention module.

Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses.

As machine learning systems grow in scale, so do their training data requirements, forcing practitioners to automate and outsource the curation of training data in order to achieve state-of-the-art performance. The absence of trustworthy human supervision over the data collection process exposes organizations to security vulnerabilities; training data can be manipulated to control and degrade the downstream behaviors of learned models. The goal of this work is to systematically categorize and discuss a wide range of dataset vulnerabilities and exploits, approaches for defending against these threats, and an array of open problems in this space.

Absence of amusia and preserved naming of musical instruments in an aphasic composer.

M.M., a right-handed, 74 year old professional musician and composer, presented with a progressive aphasia with a severe anomia.

The influence of semantic and perceptual encoding on recognition memory in Alzheimer's disease.

Previous results from a population of patients with Alzheimer's disease (Dalla Barba and Goldblum, 1996) demonstrated that the ability of patients to make a semantic association between two items was significantly and positively correlated to their performance on a yes/no recognition task for the same items and that patients who were impaired on the semantic task did significantly worse on the recognition task than patients who were unimpaired on the semantic task. These findings gave support to a hierarchical model of organization of human memory in which episodic memory depends on the integrity of semantic memory. The present study further investigates the relationship between semantic memory deficits and episodic recognition memory in 15 patients with Alzheimer's disease and 15 controls, as a function of their semantic and perceptual encoding abilities and of their cognitive impairment in other domains.