New Machine Learning Algorithm Breaks Captchas More Effectively
Scientists from UK’s Lancaster University and China’s Northwest and Peking University recently developed a new machine learning algorithm which is reportedly able to solve Captcha systems more effectively.
The GAN (Generative Adversarial Network) based algorithm was able to break text captchas from sites like like Wikipedia, Microsoft, eBay, Baidu, Google, Alipay, JD, Qihoo360, Sina, Weibo, and Sohu. The researchers found that it takes up to 2 hours (less than 30 minutes for most of the scheme) to collect 500 captchas, and less than 2 hours to label them by one user, which means the effort and cost for launching the attack on a particular captcha scheme is low.
They added that the GAN’s solver component which they developed was also more efficient and cheaper than any other approaches and can solve a captcha within 0.05 of a second on a PC.