English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper we describe how to plant novel types of backdoors in any facialrecognition model based on the popular architecture of deep Siamese neuralnetworks, by mathematically changing a small fraction of its weights (i.e.,without using any additional training or optimization). These backdoors forcethe system to err only on specific persons which are preselected by theattacker. For example, we show how such a backdoored system can take any twoimages of a particular person and decide that they represent different persons(an anonymity attack), or take any two images of a particular pair of personsand decide that they represent the same person (a confusion attack), withalmost no effect on the correctness of its decisions for other persons.Uniquely, we show that multiple backdoors can be independently installed bymultiple attackers who may not be aware of each other's existence with almostno interference.  We have experimentally verified the attacks on a FaceNet-based facialrecognition system, which achieves SOTA accuracy on the standard LFW dataset of$99.35\%$. When we tried to individually anonymize ten celebrities, the networkfailed to recognize two of their images as being the same person in $96.97\%$to $98.29\%$ of the time. When we tried to confuse between the extremelydifferent looking Morgan Freeman and Scarlett Johansson, for example, theirimages were declared to be the same person in $91.51 \%$ of the time. For eachtype of backdoor, we sequentially installed multiple backdoors with minimaleffect on the performance of each one (for example, anonymizing all tencelebrities on the same model reduced the success rate for each celebrity by nomore than $0.91\%$). In all of our experiments, the benign accuracy of thenetwork on other persons was degraded by no more than $0.48\%$ (and in mostcases, it remained above $99.30\%$).

Facial Misrecognition Systems: Simple Weight Manipulations Force DNNs to Err Only on Specific Persons