In the sphere of AI, the best is yet to come, but we need to help it
TechTalks, December 16, 2020
Adversarial examples or adversarial attacks have gathered a lot of attention from both researchers and hackers lately. These are modified images, audio or text files that can mislead the AI system, thereby affecting its operation. Despite a large number of theoretical studies and discussions, the fight against adversarial attacks remains quite difficult in real non-theoretical life. What distinguishes an adversarial attack from other attacks is that there is no uniform pattern of behavior once it is detected. During the attack, the malefactor studies the system for a long time, changing pixel by pixel until he achieves the desired result, which makes each adversarial attack unique in its own way. This makes finding and patching adversarial attacks quite a difficult task.
The development and construction of the necessary defenses, as well as methods of combating adversarial attacks, which would work well and safely in practice, is a top priority task for the future.
The Hacker News, December 14, 2020
The Sophos and ReversingLabs cybersecurity companies have recently released the first-of-its-kind publically available malware research dataset. The project aims to benefit the industry in building applicable defenses improving security detection and response.
The dataset is called “SoReL-20M” (which stands for Sophos-ReversingLabs – 20 Million) and it contains information on 20 million Windows Portable Executable (.PE) files, among them there are 10 million deactivate malware samples. The selection is supposed to improve malware detection capabilities of ML-based approaches.
“Open knowledge and understanding about cyber threats also leads to more predictive cybersecurity,” Sophos AI group representatives commented. “Defenders will be able to anticipate what attackers are doing and be better prepared for their next move.”
TechRepublic, December 18, 2020
New Year is already very close, which means it’s time to make assumptions about what it can bring. According to many experts, the growth in the importance of artificial intelligence in various fields will not stop. Cybersecurity sphere is no exception.
Marcus Fowler, director of strategic threat at Darktrace, revealed some of his predictions for cybersecurity in 2021 mentioning the impact of artificial intelligence on the sphere as well. The expert believes that smart technologies can make life much easier for cybersecurity companies by solving detection issues for them. AI can really help the human team to a degree with those tasks that can be automated, for example, autonomous sorting, autonomous investigation.
IoTFuse has developed into Minnesota’s largest IoT conference and more recently the world’s first IoT-powered digital experience. Attendees from all over the world watch live workshops and speaker sessions from ...
