吸血鬼教授vs狼人工程师
“Vampires need blood from human donors. Professors need publications, which they extract from their grad students.” – Peter Tu
A few weeks ago I attended two conferences: the European Conference on Computer Vision (ECCV) and the IEEE Conference on Advanced Video and Signal-Based Surveillance (AVSS). At both events I was invited to participate on the vision and industry discussion panels, where we considered how computer vision research relates to the business community. During these debates it became clear to me that academic researchers or the Professors and Industrial Researchers (IRs) view themselves as distinctly different breeds. I came to the conclusion that there is a good analogy between Computer Vision researchers and the supernatural.
Lets start with the Professors, whose analog would be vampires. Vampires tend to be solitary creatures skulking around in their musty castles. Professors also tend to keep to themselves, but instead of castles they are found lurking in universities. Vampires need blood from human donors. Professors need publications, which they extract from their grad students. While vampires are immortal, Professors have tenure. Vampire and academic societies are both very hierarchical. The position of a given Professor can be directly inferred by observing where he or she sits during the conference dinners. Many enthralled mortals hope to become vampires, but few ever do. Likewise many grad students hope to become Professors, however at any given time one can count on one’s fingers the number of open computer vision Professorships. Many grad students end up drifting away from the field all together. Some end up on Wall Street solving differential equations for their financier and money-lending masters. However others can make the conversion to industry, where they become the research equivalent of werewolves.
Werewolves have larger physiques when compared to the diminutive Vampires. In the research world muscle is equated to gear and big working systems, which IRs spend much of their time collecting and building. Werewolves live in packs, for IRs such communities are known as corporate research labs. Like Vampires, Professors tend to have aristocratic tastes and interests such as how to make proper measures on Riemannian manifolds, in contrast werewolves and IRs are a little earthier – IRs will talk at length about such issues as runtime and robustness. Werewolves must periodically make a metamorphosis into hideous monsters. IRs are equally cursed. Instead of the full moon they have quarterly reviews. During such distressing times they must put down their slide rules and compasses to become power-point spewing corporate stooges. For senior IRs the transformation can be so complete that they can be seen dawning both jacket and tie!
There has always been animosity between vampires and werewolves. Professors and IRs also share a certain amount of contempt for each other. Professors feel that they are the only one’s fit to consider fundamental problems and that once they have figured things out, the problems can then be sent to IRs who should be relegated to working out the fine details. While IRs assert that the professors simply play around with problems, come up with solutions that don’t really work, subsequently claim the problem to be no longer interesting and then move on the next topic de jour. Like all prejudices, both views are neither charitable nor accurate.
Wherever there are vampires and werewolves there too are vampire and werewolf hunters. Researchers also have their nemeses and they are known as Masters of Business Administration (MBAs). Instead of stakes and silver bullets, MBAs come armed with spreadsheets, schedules and market surveys. Their zealous attacks can come from any angle including questions regarding relevance to society, projections of customer return on investment and concerns regarding the timeliness of results.
In the movies, the young starlets always fall in love with the supernatural heartthrob – well I guess all analogies have their breaking points… As I mentioned at the beginning of this piece I recently attended a convening of computer vision researchers. At all such events one must flex a little muscle – see the video below…
cybercser翻译了这篇文章，如下
几周前我参加了两个会议：欧洲计算机视觉大会和IEEE视频监控（AVSS）大会。两场大会上我都受邀参加了视觉与产业讨论组，这个讨论组关注的是计算机视觉研究和商业领域之间的关系。讨论中，我越来越清晰地感觉到学术研究者（或者说是教授和工程师们）视自身为异类。我得出一条结论：计算机视觉研究人员和超自然的东西刚好能对号入座。
就从教授开始吧，他们就像是吸血鬼，吸血鬼喜欢独自隐居在发霉的城堡里。教授也喜欢一个人呆着，只不过不是呆在城堡，而是在大学校园里“潜伏”着。吸血鬼得吸食人血。教授则要发论文，而这些论文就是从他们的研究生身上榨取来的。吸血鬼有不死之身，教授则是终身聘用。吸血鬼和学术界都是等级森严。某教授的身份等级从他们开会时坐的位置就看得出来。很多着了魔的人类想变成吸血鬼，但没多少人能搞定。与此类似，许多研究生希望评上教授，但晋升计算机视觉教授职称的几率几乎掰着指头就能算出来。很多研究生最终还是转行干别的去了。有些则去了华尔街替他们的投资人和“放高利贷”主子解微分方程去了。尽管如此，另外一部分人还是干了老本行，变成了“狼人”。
与羸弱的吸血鬼相比，狼人可谓身强力壮。在研发领域，设备和牛叉的系统就是“肌肉”。工程师们为了搜集和构架这些设备、系统花了不少功夫。狼人成群而居，工程师们则都混迹于“企业研发实验室”中。教授像吸血鬼，口味和兴趣都带着贵族气息，喜欢诸如“黎曼流形计算方法”的课题。于此形成对比的是，狼人和工程师们要“土鳖”一点——工程师们一直挂在嘴边的话题就是运行时和健壮性。狼人会周期性地变身为狰狞的怪兽。工程师们也同样难逃此咒。不过他们不是每个月圆时分，而是每次季度评审时发作。在季度评审的沮丧时光中，工程师得收起计算尺和圆规，变身为“幻灯片吐槽大虾”。高级工程师变身更为彻底，你甚至能看到他们一大早就穿着西服，打着领带!
吸血鬼与狼人之间一直火药味呛人。教授和工程师也是互相鄙视。教授觉得他们才是考虑基础性问题的大牛，而且一旦他们解决了基本问题，剩下的工作就可以下放给工程师，让他们去细化了。工程师们则一口咬定，说教授们只会绕圈子，净琢磨些没用的解决方案，最后才放出话来，说该问题实在无趣，要忙着搞下一个问题去了。偏见就是偏见，两种观点都不厚道，也不准确。
哪里有吸血鬼和狼人，哪里就有范海辛。工程师们的死对头就是传说中的工商管理硕士（MBA）。MBA用不着装备银杵和银弹，他们使的家伙是数据表、进度规划和市场调查。他们能从各个方向不懈攻击，拿“社会效益”、“客户投资回报率”、“有关产品时间线的担忧”等问题轰炸你。
电影里的年轻女星总是和超自然的帅哥坠入爱河——好吧，我估摸着打比方总有不靠谱的地方……。就像我在文章开头说的那样，我最近出席了一个计算机视觉研究者大会。在这种盛会上，你得拿出点颜色给同行们瞧瞧 —— 请看下面的视频……