LAS VEGAS, Aug. 13, 2018 /PRNewswire/ -- KEEN, a global provider of cloud and mobile security, today announced GeekPwn 2018, the world's leading platform for cybersecurity researchers was held at DEFCON, Las Vegas on August 11th. The event offered a window into the world of world-class technology hacker culture and spirit to security enthusiasts and fans. This year, the event focused on AI cybersecurity, and featured various contests and talks on the topic.
Highlights from these talks and contests include the first global CAAD CTF (Competition of Adversarial Attacks and Defense, Capture the Flag) event that featured six top AI security teams, CAAD Village keynotes that revealed top 10 AI security issues for the first time, the inaugural "break-the-wall" Robot Agent Contest, and the "Matrix and Zen" project Geek Pwn-ty.
This year's GeekPwn 2018 sessions was joined by prominent industry experts from institutes such as Tsinghua University, UC Berkeley, Cornell University, Yale University and JD.com, amongst others. Using GeekPwn as a platform, these experts showcased their findings and paid tribute to the prominence that the whitehat hacking industry has achieved over the years.
An industry First: The first ever Capture-the-Flag A&D (Attack & Defense) battle in the AI field
AI technology has gradually evolved and now is being applied to all aspects of people's daily life. This makes the potential threats of AI something that cannot be ignored. As the world's first global platform for exploring AI and professional security, GeekPwn has been focused on AI security since it was founded in 2014. GeekPwn has encouraged and guided industry experts and technicians to explore AI security for several years to verify possible vulnerabilities in AI. The most advanced adversarial sample on AI security battle has become the calling card of GeekPwn events, and this year it was held for the first time on American soil in Las Vegas. Through extending research of adversarial samples, GeekPwn offers cross-border integration between the AI community and the broader hacker community and to push healthy and sustainable growth of the whole AI industry.
DEFCON is widely regarded as the birthplace of the competition of cyber-attack and defense technology. CTF, the signature event of DEFCON, is known as the "World Cup" in the network security circle. During DEFCON, GeekPwn debuted the world's first CTF in machine learning (ML) field, surpassing the limitations of A&D of traditional cyber security competitions. For the first time, participants can view the complexities of conducting "Attack and Defense in the AI world."
Six teams from top tech firms and research institutions, including Tsinghua University, Cornell University, Yale University and JD.com, participated in the industry's first CAAD CTF. GeekPwn not only attracted teams which are deeply involved in the security field for many years which consist of top experts, like the Blade Team and the NorthWest Security team, but also teams such as YYZZ and UCNESL, which composed of young talented students or graduates from top organizations such as Yale and Cornell.
The contest utilized a random on-site matching mechanism which placed six top AI security research teams to create "Adversarial Samples" to defraud and attack other teams' defense algorithms. After a close contest, the TSAIL team from Tsinghua University emerged champions of the CAAD CTF contest. NorthWestTSec team won the second place and YYZZ team won the third place. UCNESL, Blade and JD-Omega teams had the winning prizes.
CAAD Village: 10 AI security topics released
In addition to the exciting AI attack and defense battles, GeekPwn2018 in Las Vegas brought 10 keynotes with tangible evidenced-based sharing of information technology enthusiasts from all over the world. GeekPwn's CAAD Village invited 10 world-class AI and hacker researchers, from Clemson University, University of California, Tsinghua University, NorthWestSec team, Blade Team, Tencent Security Yunding Lab, JD-Omege Team, Baidu X Lab, UCNESL team and YYZZ team, to share the latest research results on AI security and adversarial samples. Topics covered included autonomous vehicles, speech recognition systems, anti-virus software and neural networks, which are highly watched by AI entrepreneurs, practitioners and scholars.
Jose BOEY and Dr BO Li kickstarted the keynotes with two talks on the main stage. They addressed the issue of “Adversarial Attacks on Face Detectors using Neural Net based Constrained Optimization” and “Secure learning in adversarial environments” respectively. BOEY proposed a novel strategy to craft adversarial examples by solving a constrained optimization problem using an adversarial generator network (GAN) and demonstrated successfully on stage how it can be done. Breaking off from conventional attack strategies, BOEY showed that the same trained generator is capable of attacking new images without explicitly optimizing on them, which can lead to new implications for new forms of attack and defense. BO’s keynote covered extensively how generating evasion attacks in adversarial environments can prevent the undermining of the integrity of learning systems, and hence protecting the privacy of users' data. BO also provided some insights about how to leverage data properties, such as spatial consistency as potential detection approach against such adversarial attacks, which is what she is currently working on in her research
Some highlights at the GeekPwn CAAD Village include seeing, Weilin Xu, a PhD student at the University of Virginia and intern researcher at Baidu X-Lab brought a demo of "Magic Tricks for Self-driving Cars”, revealing how the use of adversarial ML (machine learning) technology can possibly deceive camera-based perception systems.
Another notable highlight was the presentation given by Rohan TAORI and Amog KAMSETTY, graduate students from the University of California at Berkeley, when they shared the topic of directional "black box" attack analysis for audio systems. They stated that the application of deep reproduction networks on audio dubbing has brought significant contribution to automatic speech recognition (ASR) systems. Many cases have shown that small antagonistic disturbances can confuse deep neural networks, making it impossible to achieve correct predictions of specific targets.
BO Shi, a researcher from the Tencent Security Yunding Lab, shared contents of Dogfight weapon GAN, explaining how malwares/viruses can be used to fight against anti-virus (AV) software. It was discovered that once malware appeared, a battle took place between the malware and AV, and therefore, it can be used to enhance more machine learning-based methods that are suitable for malware detection.
Robot Agent Contest: An exhibition of the true hacker spirit and cross-industry brainstorms
In addition to focusing on AI security, the return of the Robot Agent Contest was a highly popular feature at GeekPwn. The contest showcased many experts' participation in the robotics field, highlighting various security issues across physical and cyberspace worlds. At the unveiling part of Robot Agent contest in Las Vegas, researchers from the University of Nevada's DASL Lab showcased their self-made robots. Under control of the players, the tech-savvy robots completed the tasks of placing a covert listening device in the lab and openning safe to get secret information, which brought a delightful visual and intellectual stimulant to the audience.
Besides competition and ground-breaking keynotes, GeekPwn paid homage to the hacker spirit with industrial innovation, freedom of ideas and breakthrough research. In the off-site interactive session, GeekPwn set up a robot soccer competition to create a robotic version of the "World Cup". The Las Vegas edition ended with a themed party called "Matrix and Zen". The meeting of a grassroots hacker culture and traditional oriental culture is a breath of fresh air within global hacker circle.
Wang Qi, KEEN's CEO and GeekPwn's Founder, remarked: "GeekPwn is perhaps the first platform in the world to encourage 'PWN Everything'. Therefore, GeekPwn hopes to establish a bridge between white hat hackers and other technical fields, and let white hats explore wider horizon fields by adopting hacker-like thinking. Security risks are always making the world realize the importance of prevention and value of whitehat hackers."
The Shanghai edition of the GeekPwn 2018 Carnival will be held between the 24th and 25th October 2018. Various competitions such as the CAAD Competition, AI Data Tracker, Hacker Room Challenge, Robot Agent Challenge, Vulnerability-based Hack, Non-Vulnerability-based Hack, GAN or Recover Blurred Image Challenge and RFID Long Distance R&W Challenge will be on the stage of Shanghai.
More information on the Shanghai edition can be found on the website at geekpwn.org.
For more information, please contact: