泛化能力和泛化误差_个性化或泛化是接管活动行业的老大哥

泛化能力和泛化误差

The center of any business is the customer. Understanding customer behavior and finding the most effective and appropriate way to accommodate their needs and expectations is the single most important aspect of business planning. Event management is no different. As event managers, we put the attendee in the center of any planning strategy and do in-depth research of methods to engage them through every touchpoint.

吨他中心的任何业务的是客户。 了解客户的行为并找到最有效和最合适的方式来满足他们的需求和期望是业务计划的最重要方面。 事件管理没有什么不同。 作为活动的管理者，我们将参加者放在任何计划策略的中心，并对方法进行深入研究，以使他们通过每个接触点参与其中。

Managing to please and engage the event attendees results in high satisfaction rates, and, ultimately, in good ROI and profit. And what a more secure way to engage someone than by making them feel like the event is specifically crafted for them. They feel acknowledged, you have saved them time, and have directly addressed something that they are interested in.

设法取悦并吸引活动参加者可以提高满意度，并最终带来良好的投资回报率和利润。 与让某人感觉像事件一样的一种比某人更安全的互动方式是专门为他们设计的。 他们感到认可，您节省了他们的时间，并直接解决了他们感兴趣的问题。

A decade ago, such levels of personalization were a tale of fiction. But as we have officially entered an era of commonly used artificial intelligence systems, personalization is not only possible but expected.

十年前，这种水平的个性化是一个虚构的故事。 但是，随着我们正式进入常用人工智能系统的时代，个性化不仅是可能的，而且是预期的。

人工智能实际上可以为您的活动做什么？ (What can AI actually do for your event?)

Here is what the industry experts say.

这是行业专家所说的。

1.个性化建议 (1. Personalized recommendations)

The biggest victory of AI in almost every industry, including events, is personalizing recommendations based on users’ preferences, background, etc. In the event context, that feature allows events to offer a much more effective networking experience. AI is offering personalized recommendations for attendees, exhibitors, and products on a colossal scale. An application that connects attendees is nothing out of the ordinary and is very standard for most exhibitions and conferences. However, scanning through thousands of people, selecting the ones that you want to connect with, and reaching out to each one individually proves to be time-consuming for the attendees and expensive for event planners to manage, says Tim Groot, co-founder, and CEO at Grip, a UK-based event networking platform powered by AI. Some applications, on the other hand, allow the user to fill a form with main interests and be matched with others based on similarities, but that match is not always a good fit. So how is an AI matchmaking engine any better? Well, as soon as you give it access, it creates an in-depth profile of your activity on social media platforms such as LinkedIn and Facebook, and based on the information it extracts, it recommends people to meet, sessions to attend, or products to buy that best suit your business interests. You can then indicate whether you are happy with the recommendations or not, and the machine will know what it needs to improve on.

在几乎包括事件在内的每个行业中，AI的最大胜利就是根据用户的喜好，背景等来个性化推荐 。在事件的上下文中，该功能使事件能够提供更加有效的网络体验。 AI正在为与会者，参展商和产品提供巨大的个性化建议。 连接与会者的应用程序与众不同，并且对于大多数展览和会议来说都是非常标准的。 不过，联合创始人Tim Groot表示 ，扫描成千上万的人，选择您想要与之联系的人，并分别与每个人联系，这对参与者来说是很耗时的，而对于活动策划者来说，这是很昂贵的。总部位于英国，由AI驱动的活动联网平台Grip的首席执行官。 另一方面，某些应用程序允许用户填写具有主要兴趣的表格，并根据相似性与其他人进行匹配，但是这种匹配并不总是很合适。 那么，人工智能对接引擎会更好吗？ 好吧，一旦您授予它访问权限，它就会在LinkedIn和Facebook等社交媒体平台上创建您的活动的深入概况，并基于其提取的信息，向人们推荐会议，参加会议或产品购买最适合您的商业利益的产品。 然后，您可以指示对建议是否满意，机器将知道需要改进的地方。

2.情感认同 (2. Emotional recognition)

Yes, it is just as scary as it sounds. In an interesting article on the topic, EventMB explores this emerging trend in the event industry. Although still shaky and quite controversial, emotional recognition, powered by artificial intelligence, is already happening. What is it exactly? It is an AI engine that uses facial recognition to analyze your facial expressions, and, based on the data, determines how you feel at any given moment.

是的，它听起来一样可怕。 在关于该主题的有趣文章中 ，EventMB探索了事件行业中的这种新兴趋势。 尽管仍然动摇不定且颇具争议，但由人工智能驱动的情感识别已经在发生。 究竟是什么？ 这是一个AI引擎，它使用面部识别来分析您的面部表情 ，并根据数据确定您在任何给定时刻的感觉。

Scientists have developed techniques to try and replicate the way humans detect and analyze emotions. They have then fed an AI machine with their algorithms, and allowed it to observe people using facial recognition software, much similar to the one we use to unlock our phones. It even goes one step further by attempting to predict behavior based on detected emotion. In the context of events, that would essentially indicate that event managers have the opportunity to understand people’s feelings not only through a survey at the end of the event but at any given point throughout.

科学家已经开发出技术，尝试并复制人类检测和分析情绪的方式。 然后，他们向AI机器提供了算法，并允许它使用面部识别软件观察人，这与我们用于解锁手机的软件非常相似。 通过尝试根据检测到的情绪来预测行为，它甚至更进一步。 就事件而言，这实质上表示事件管理者不仅有机会通过事件结束时的调查，而且可以通过整个过程中的任何特定点来了解人们的感受。

Image courtesy of Zenus. Source: EventMB

图片由Zenus提供。 资料来源： EventMB

3.通过计算机视觉映射事件 (3. Mapping events through computer vision)

The technology behind emotional recognition is computer vision powered by an AI machine learning engine. It is essentially one camera that can access images of the whole event, and use them to draw conclusions on various event aspects. Here are some examples of how the technology could be used in an event context, summarized by Zenus.

情感识别背后的技术是由AI机器学习引擎提供动力的计算机视觉。 本质上，它是一台摄像机，可以访问整个事件的图像，并使用它们得出事件各个方面的结论。 下面是一些例子中如何应用该技术可以在一个事件背景下，Zenus总结使用。

· Computer vision allows event managers to maintain an accurate real-time count on attendance. That would effectively show the strongest and weakest points of attendance.

·计算机视觉使事件管理器可以保持准确的实时出席人数。 这将有效地显示出出席会议的最强点和最弱点。

· Through computer vision, event managers can assess dwell times, and create accurate reports on how much time people spend at any given area. This can then be very valuable to sponsors, as well as allow improvement of event layout.

·通过计算机视觉，事件管理者可以评估停留时间，并创建有关人们在任何给定区域花费多少时间的准确报告。 这样对赞助商来说可能非常有价值，并且可以改善活动的布局。

· Computer vision also provides accurate real-time data on demographics, which can later be used for more targeted marketing and event design.

·计算机视觉还提供有关人口统计信息的准确实时数据，以后可用于更有针对性的营销和活动设计。

The general idea is personalization for the attendee and data for the event organizer. Ideally, that should be a ground-breaking technology that changes the industry radically. And, admittedly, to some extent it is.

总体思路是为与会者提供个性化设置，并为活动组织者提供数据。 理想情况下，这应该是一项能够从根本上改变行业的突破性技术。 而且，诚然，在某种程度上是这样。

但这真的是个性化还是泛化？ (But is it really personalization or generalization?)

The AI knows algorithms. To explain it to a five-year-old, it knows that if X and Y happen, then Z is a fact. It is likely that an IT professional would confront that statement with a much more complex explanation of the processes, but the reality is, the biggest barriers for AI have been and still are cultural and situational nuances. According to research, facial recognition is prone to misidentifying people of color and people who’s gender representation doesn’t quite fall under the binary norm. In other words, despite its efforts to be as human as possible, the AI struggles to recognize the beauty of human differences in appearance.

AI知道算法。 为了向五岁的孩子解释，它知道如果X和Y发生，那么Z是事实。 IT专业人员可能会以更复杂的流程解释来面对该声明，但现实情况是，AI的最大障碍一直是，而且仍然是文化和情境上的细微差别 。 根据研究 ，面部识别很容易误认有色人种，而性别代表性还不太符合二元制的人。 换句话说，尽管AI尽了最大的努力，但AI仍在努力识别外观上的人类差异之美。

That inevitably brings another question. If the technology is not advanced enough to recognize our most human features such as our gender or the color of our skin, then could it really succeed in identifying how we feel at any given moment? By referencing new research on emotional recognition, conducted by The APS, EventMB is pointing out that using facial analysis alone to detect emotion could prove highly inaccurate. Our emotions are nuanced, and most importantly, expressed in various ways depending on the human’s personality. Analyzing someone’s emotional state by simply looking at their facial expressions and running the data through a list of possible conclusions such as “smiling = happy” is naive.

这不可避免地带来了另一个问题。 如果这项技术不够先进，无法识别我们最人类的特征，例如我们的性别或皮肤颜色，那么它真的可以成功地识别出我们在任何给定时刻的感觉吗？ 通过引用由APS进行的有关情感识别的新研究 ， EventMB指出，仅使用面部分析来检测情感可能会证明非常不准确。 我们的情感细微差别，最重要的是，取决于人的性格以各种方式表达。 仅通过查看某人的面部表情并通过一系列可能的结论(例如“微笑=快乐”)运行数据来分析某人的情绪状态就太幼稚了。

And then, on a simpler level, as EventMB put it, “What if I am sneezing? Does that mean I am not enjoying the session I am attending?”. What happens when our face reacts to something that is out of context? I see a picture of my ex on Facebook during lunch, for example. Does the AI know my expression is not a result of the way I feel towards the food?

然后，从一个更简单的层面上，如EventMB所说 ：“如果我打喷嚏怎么办？ 这是否意味着我不喜欢正在参加的会议？”。 当我们的脸对超出上下文的事物做出React时会发生什么？ 例如，午饭时我在Facebook上看到前任的照片。 AI是否知道我的表情不是我对食物的感觉的结果？

Finally, what about privacy? If you are into conspiracies or have recently watched any CSI series, you have already asked yourself “But what do they REALLY use all this data for?”. Yes, some people are concerned about where all of this data is stored and don’t find joy or comfort in knowing their emotions are being so carefully observed by a Master Computer.

最后，隐私如何？ 如果您是阴谋家或最近看过任何CSI系列节目，您已经问过自己“但是他们真的将所有这些数据用于什么吗？”。 是的，有些人担心所有这些数据都存储在哪里，而知道自己的情绪正在被主计算机如此仔细地观察时，他们并不会感到高兴或安慰。

Others worry about a more tangible issue: being spammed with offers based on an inaccurate analysis of their feelings. Similar to that one time your 14-year-old daughter played a Justin Bieber song from your YouTube account, and now you get countless suggestions for make-up tutorial videos. The AI read the data, but it didn’t know the context.

其他人则担心一个更具体的问题：由于对自己的感受的不准确分析而使他们被垃圾邮件所笼罩。 类似于您14岁的女儿从您的YouTube帐户播放贾斯汀·比伯(Justin Bieber)歌曲的那一次，现在，您收到了有关化妆教程视频的无数建议。 AI读取了数据，但它不知道上下文。

活动行业正在反击保护数字人权 (The event industry is fighting back to protect digital human rights)

40 of the world’s largest music festivals, including Burning Man, Coachella, and Lollapalooza, have declared war against facial recognition technology. The movement, initially started by musicians and activists, is now supported by many famous performers and has resulted in the largest event ticket distributor, Ticketmaster, taking a step back from its strategy to develop a facial recognition event entrance method.

世界上最大的40个音乐节 ，包括燃烧人，科切拉和Lollapalooza， 都宣布反对面部识别技术。 运动，最初是由音乐家和活动家开始，现在被许多著名演员的支持，并已造成了最大的活动门票总代理，票务， 退后一步 ，从它的战略开发面部识别事件入口方法。

Amanda Plamer, amongst many others, invokes on her Twitter account:

阿曼达·普拉默(Amanda Plamer)以及其他许多人都在她的Twitter帐户中调用：

“People should feel safe & respected at shows, not subject to surveillance, harassment, deportation, or arrest.”

“人们在表演时应该感到安全和受到尊重，不受监视，骚扰，驱逐出境或逮捕。”

Source: https://www.banfacialrecognition.com

资料来源： https : //www.banfacialrecognition.com

结论 (In conclusion)

Part of the joy of being an event manager is the dynamics of the industry combined with the creative nature of designing experiences. The ultimate goal is to create stronger and more positive experiences for event attendees, boost business opportunities at conferences and exhibitions, and bring communities closer together. In a sense, the event industry’s ability to impact, unite, and benefit societies is truly unparalleled.

成为活动经理的部分乐趣在于行业动态以及设计体验的创造性。 最终目标是为活动参与者创造更强大，更积极的体验 ，增加会议和展览中的商机， 并使社区更加紧密地联系在一起 。 从某种意义上说，事件行业影响，团结和造福社会的能力是真正无与伦比的。

This is why it is easy for event organizers to be tempted to use new technology in order to do better. But as incredible as it may sound, and as intriguing as it is to incorporate the latest piece of technology to your event, the attendees in their most human form should always be in the center. And, as of today, AI cannot yet catch-up with our complexity. And maybe, it shouldn’t.

这就是为什么活动组织者很容易被诱惑使用新技术来做得更好。 但是，尽管听起来令人难以置信，并且将最新技术融入到您的活动中也很有趣，但最人性化的参加者应该始终处于中心位置。 而且，到目前为止， 人工智能还无法赶上我们的复杂性。 也许不应该。

翻译自: https://medium.com/swlh/personalization-or-generalization-is-big-brother-taking-over-the-event-industry-4746994f717d

泛化能力和泛化误差