Once again, great responsibility comes with great force, especially when it comes to the application of AI in the financial sector
CNA, March 19, 2021
According to recent research, face recognition can improve ATM performance by preventing credit card fraud and speeding up transactions. However, like any other technology, the face recognition system also has its drawbacks: for example, it can entail certain security risks. OCBC, one of the most influential lenders in Singapore, is starting to introduce a facial recognition system into its ATMs. So far, the system has been implemented in eight ATMs, but it will soon expand to the remaining 550. The system allows individuals to check the account balance, however the functions available with the face recognition system will also be expanded in the future.
The introduction of a facial recognition system into ATMs is nothing new. For example, such a system was used by some Chinese banks back in 2015. According to experts, such a system solves some security issues, since clients do not have to use a plastic card to work with the banking system.
Such systems also entail certain risks, because attackers will definitely try to deceive the systems. This can be done, for example, using deepfakes. However, experts believe that this should not stop the spread of the technology, since developers, together with banks, use a number of tactics to prevent attacks. In addition, there is still a lot to be done in the future, for example dual factor authentication can be implemented.
dezeen, March 19, 2021
Gucci has developed special digital-only trainers. It will be fashionable to try on shoes using augmented reality. For example, the new neon sneakers can be worn on social media photos. The model was named The Gucci Virtual 25 in honor of the popular number of Gucci creative director Alessandro Michele. This development of the fashion house is unique, since the famous company has developed for the first time a fashion accessory available to customers only in the virtual world.
Since last fall, Gucci has also developed a range of designs for a variety of virtual wearable devices in a number of video games – for example, one of the games in the Sims series.
Progress in the development of virtual clothing is an extremely important element in the development of virtual reality, since such developments can significantly affect the development and creation of a variety of adversarial clothes. Such clothes are already created at the moment and serve to deceive smart systems in the detection and recognition of objects.
Eurek Alert!, March 11, 2021
Deep neural networks used for image and video recognition still have a number of vulnerabilities. In particular, they are vulnerable to adversarial perturbations, tiny changes in the input image that are unnoticeable to the human eye, but affect the output.
Skoltech researchers have demonstrated that the patterns that cause errors in neural networks when recognizing images are similar to Turing patterns, which are essentially found throughout the world around them. In the future, such data may be useful in developing protection against attacks on neural networks.
The possibility of such attacks poses major security concerns. Such risks are especially relevant for autonomous vehicles, which, for example, can be deceived in recognizing road signs.
Professor Ivan Oseledets leading the Skoltech Computational Intelligence Lab at the Center for Computational and Data-Intensive Science and Engineering (CDISE), and his colleagues, studies a theory connecting such universal adversarial perturbations (UAPs) and Turing patterns. The nature of adversarial perturbations is still a mystery to researchers, and the lack of theory complicates research. In contrast, natural Turing patterns have a large theoretical basis, and with its help in the future it is possible to build a theory for adversarial perturbations.
It turns out that even complex systems can sometimes be deceived with an ordinary pen and paper OpenAI’s state-of-the-art machine vision AI is fooled by handwritten notes The Verge, March ...
