Using lightweight electroencephalography (EEG) to exploit the fringe P3 response

长期以来，警察素描师一直被用来建立嫌疑人的侧写。他们利用了证人陈述和受害者报告，但是，这种方法可能不可靠，因为证人可能会隐瞒信息或在他们的陈述中不诚实。近年来，随着电子面部识别技术（E-FIT）的广泛使用，甚至最近引入人工智能来帮助创建这些E-FIT配置文件，已经取得了长足的进步。这些方法仍然受到证人偏见和对抗技术的影响，因为它们需要证人的证词来创建。有时这些e - fit会因为其滑稽的外观而受到新闻和社交媒体的关注，从而对调查产生不利影响[6-8]。尽管如此，E-FITs已被全球30多个国家使用，成功率在40-55%之间，这就提出了如何使这些图像防篡改以实现更高的准确性和更快速的图像生成[5]的问题。目前，VisionMetric有限公司正在研究如何提高E-FITs的精度，该研究使用快速连续视觉呈现（RSVP）来测量大脑对与嫌疑人相似的图像的反应。RSVP和其他基于脑电图（EEG）的隐藏信息测试，结合edge - p3方法，已被证明是一种很有前途的法医应用方法。例如，Frings-P3方法已被用于在地址流中正确检测受试者的电子邮件地址，从而将个人的在线身份与他们自己联系起来b[1]。它通过在屏幕上快速地向受试者展示一系列图像或类似的刺激，使受试者没有足够的时间有意识地思考所显示的信息，但大脑确实识别出熟悉程度。当受试者看到熟悉的物品时，研究人员能够检测到P3反应，通过这种方式，你可以绕过受试者可能尝试和部署的对抗技术。2014年，研究人员能够使用这种RSVP方法来证明你可以成功绕过对抗技术[3]。在这个例子中，研究人员在屏幕上显示了受试者的名字，并告诉一些受试者在显示熟悉的样本时尽量抑制反应的技巧，而一些受试者被告知要努力提高对控制刺激的反应。这项研究表明，通过使用这种方法，你可以绕过用来欺骗测谎仪等系统的反措施技术，这已经成为执法机构几十年来的一个问题。尽管这些方法显示出了希望，但主要的缺点是进行测试所需的脑电图设备。在研究环境中，传统设备是合适的，但是有很长的准备时间，需要对每个受试者进行培训，设备体积大，佩戴者不舒服，通常需要在受试者的头皮上放置许多电极，因此不容易在现实世界中部署。为了解决这个问题，研究人员开发了一种名为SKINTRONICS的新型脑电图。这种设备是无线的，重量轻，而且受到的电磁干扰少得多，已经被证明与传统的台式脑电图系统[4]具有相同的准确性，如果不是更好的话。虽然该系统已被证明是有效的，但它从未被用于探测P3波。如果我们能开发出一种方法来准确地结合这些方法，并证明你可以根据大脑P3反应生成越来越准确的图像，这将是一项有价值的研究，并将帮助全球的执法机构。

For a long time, police sketch artists have been used to build a profile of a suspect. They have utilized witness statements and victim reports, however, this method for identifying suspects can be unreliable since the witnesses can hold information back or not be truthful in their statements. Strides have been made in recent years with the more widespread use of the electronic facial identification technique (E-FIT) and even more recently the introduction of artificial intelligence to aid the creation of these E-FIT profiles. These methods still suffer from witness bias and countermeasure techniques since they require a testimony from the witness to create. Sometimes these E-FITs get attention from the news and social media for their comical appearance which can adversely affect the investigation [6-8]. Despite this, E-FITs are used by over 30 countries around the world and have success rates between 40-55% which raises the question of how these images can be made tamper-proof to enable better accuracy and more rapid image generation [5].Current research into improving the precision of the E-FITs by VisionMetric Ltd uses rapid serial visual presentation (RSVP) to measure the brains responses to images that bear resemblance to the suspect. RSVP and other electroencephalography (EEG) based concealed information tests, in combination with the Fringe-P3 method, have proven to be a promising method for forensic applications. For example, the Frings-P3 method has been used to detect subject's email addresses correctly in a stream of addresses thus linking an individual's online identity to themselves [1]. It does this by showing the subject a series of images or similar stimuli on a screen fast enough that the subject does not get enough time to consciously think about the information they have been shown, but the brain does recognise familiarity. Researchers were able to detect a P3 response when the subject was shown a familiar item, in this way you can bypass countermeasure techniques that a subject may try and deploy. In 2014, researchers were able to use this RSVP method to show that you can successfully bypass countermeasure techniques [3]. In this example, the researchers showed the subjects names on the screen and some of the subjects were told techniques to try and dampen the response when a familiar sample was shown, and some subjects were told to try and elevate the response measured to the control stimulus. This research demonstrates that by using this method you are able to bypass countermeasure techniques employed to trick systems like lie detectors which has been a problem for law enforcement agencies for decades.Although these methods show promise, the major drawback is the EEG equipment needed in order to carry out the tests. In a research environment, the traditional equipment is suitable however there is a long preparation time, requires training on a per subject basis and the equipment is bulky and uncomfortable for the wearer, often requiring many electrodes to be placed on the scalp of the subject, therefore not easily deployable in a real-world scenario. To address this issue, researchers have developed a new EEG called SKINTRONICS. This device, which is wireless, lightweight and suffers from a lot less electromagnetic interference has been shown to have the same if not better accuracy than traditional benchtop EEG systems [4]. Although the system has proven to work, it has never been applied to detecting P3 waves. If we could develop a method to accurately combine these methods and prove you can generate increasingly more accurate images based on the brains P3 response it would be an valuable piece of research and would aid law enforcement agencys around the globe.