发新话题
打印

智能设计

乐乎设计,乐乎生活~

TOP

直觉有助于人类战胜算法

发表于2016年09月20日 11时57分 星期二

根据剑桥大学的一项研究,对“直觉”的灵敏度是金融交易成功的一个强大预测指标。对18名对冲基金交易员的研究发现,具有较大“内感”——感知自己的身体状态的能力——的人赚到更多钱,在纷乱的金融市场生存也更久。研究人员表示,通过结合身体和大脑,最优秀的人类交易员可以胜过计算机算法。研究人员说,这些发现削弱了“有效市场”理论,后者认为人类参与者不可能持续跑赢大市。论文作者之一的John Coates说,“如果我们认识到,身体和大脑作为一个单一的功能单位发挥作用,他们形成一个抛物面反射镜,收集无法被显意识分辨的信号,那么我们也将认识到,我们在快速模式识别方面是怎样的构造精美。人类确实可以抗衡机器
乐乎设计,乐乎生活~

TOP

http://www.ftchinese.com/story/001069426
研究:“直觉”有助人类交易员战胜算法

更新于2016年9月20日 06:17 英国《金融时报》 克莱夫•库克森

根据剑桥大学(Cambridge university)牵头开展的研究,对“直觉”的灵敏度是金融交易成功的一个强大预测指标。

对18名对冲基金交易员的研究发现,具有较大“内感”(interoception)——感知自己的身体状态的能力——的人赚到更多钱,在纷乱的金融市场生存也更久。研究结果发表在《科学报告》(Scientific Reports)刊物上。

“我们的研究结果似乎表明,来自身体的信号,即金融业人士所称的直觉,有助于在市场上操盘成功,”研究报告的作者们总结称。他们表示,通过结合身体和大脑,最优秀的人类交易员可以胜过计算机算法。

为了评估内感,研究人员测量了受试者在不把脉、也不接触身体任何部位的前提下察觉自己心率的能力。这18名男性志愿者都是高频交易员,在欧元区危机尾声阶段在一家名称不详的伦敦基金买卖期货合约。他们只在短时间内持有自己的交易头寸,这种操作需要他们在吸收大量信息和快速识别价格模式的基础上作出瞬间决策。

“金融市场的这个角落尤为无情,淘汰很快,”资深作者、曾在华尔街当过交易主管的约翰•科茨(John Coates)表示。“赚不到钱的交易员撑不了多久。”

在研究对象组中,根据各人的内感能力,可以预测出其产生的利润和作为一名金融交易员的“工龄”。“我们的研究结果似乎表明,他们成功地读取了真正的、宝贵的生理交易信号,即使他们不知道自己在这么做,”科茨说。

研究人员说,这些发现削弱了“有效市场”理论,后者认为人类参与者不可能持续跑赢大市。“一个交易员的生存和成功似乎在很大程度上与他们的生理存在关联,”研究团队的另一名成员、剑桥大学的马克•格内尔(Mark Gurnell)表示。

科茨补充说,这项研究也给有关计算机算法是否能够胜过人类交易员的辩论带来启示。“如果我们认识到,身体和大脑作为一个单一的功能单位发挥作用,他们形成一个抛物面反射镜,收集无法被显意识分辨的信号,那么我们也将认识到,我们在快速模式识别方面是怎样的构造精美。人类确实可以抗衡机器。”

译者/和风
乐乎设计,乐乎生活~

TOP

http://www.cam.ac.uk/research/ne ... l-financial-traders
Gut feelings’ help make more successful financial traders

Financial traders are better at reading their ‘gut feelings’ than the general population – and the better they are at this ability, the more successful they are as traders, according to new research led by the University of Cambridge.


‘Gut feelings’ – known technically as interoceptive sensations – are sensations that carry information to the brain from many tissues of the body, including the heart and lungs, as well as the gut. They can report anything from body temperature to breathlessness, racing heart, fullness from the gut, bladder and bowel, and they underpin states such as hunger, thirst, pain, and anxiety.

We are often not conscious – or at least barely aware – of this information, but it provides valuable inputs in risky decision making. High-risk choices are accompanied by rapid and subtle physiological changes that feed back to the brain, affecting our decisions, and steering us away from gambles that are likely to lead to loss and towards those that are likely to lead to profit. This can enable people to make important decisions even before they are able to articulate the reasons for their choices.

Traders and investors in the financial markets frequently talk of the importance of gut feelings for selecting profitable trades. To find out the extent to which this belief is correct, researchers from the Universities of Cambridge and Sussex in the UK and Queensland University of Technology in Australia compared the interoceptive abilities of financial traders against those of non-trader control subjects. Their results are published today in the journal Scientific Reports.

The researchers recruited 18 male traders from a hedge fund engaged in high frequency trading, which involves buying and selling futures contracts for only a short period of time – seconds or minutes, a few hours at the most. This form of trading requires an ability to assimilate large amounts of information flowing through news feeds, to rapidly recognize price patterns, and to make large and risky decisions with split-second timing. This niche of the financial markets is particularly unforgiving: while successful traders may earn in excess of £10 million per year, unprofitable ones do not survive for long.

The study took place during a particularly volatile period – the Eurozone crisis – so the performance of each trader reflected his ability to make money during periods of extreme uncertainty. The researchers measured individual differences in each trader’s capacity to detect subtle changes in the physiological state of their bodies by means of two established heartbeat detection tasks. These tasks test how accurately a person, when at rest, can count their heartbeats. Each trader was given a score which, essentially, measured the percentage of right answers, and these scores were compared against data from 48 students at the University of Sussex.

The researchers found that traders performed significantly better at the heart rate detection tasks compared to the controls: the mean score for traders was 78.2, compared to 66.9 for the controls. Even within the group of traders, those who were better at the heart rate detection tasks also performed better at trading, generating greater profits.

Strikingly, an individual’s interoceptive ability could be used to predict whether they would survive in the financial markets. The researchers plotted heartbeat detection scores against years of experience in the financial markets and found that a trader’s heartbeat counting score predicted the number of years he had survived as a trader.

“Traders in the financial world often speak of the importance of gut feelings for choosing profitable trades – they select from a range of possible trades the one that just ‘feels right’,” says Dr John Coates, a former research fellow in neuroscience and finance at the University of Cambridge, who also used to run a trading desk on Wall Street. “Our findings suggest they’re right – they manage to read real and valuable physiological trading signals, even if they are unaware they are doing so.”

Although the results are consistent with recent studies showing that heartbeat detection skills predict more effective risk taking, the researchers caution that there may be other interpretations. For example, one study has found that heartbeat detection ability increases during stress, so it could be argued that heartbeat detection skills correlated with years of survival merely because experienced traders, taking larger risks, are subjected to greater stresses. The authors of the current study think this unlikely – in trading, as in many other professions, experienced and successful individuals, being more in control, are commonly less stressed than beginners.

The findings also appear to contradict the influential ‘Efficient Markets Hypothesis’ of economic theory, which argues that the market is random, meaning that no trait or skill of an investor or trader – not their IQ, education, nor training – can improve their performance, any more than these traits and skills could improve their performance at flipping coins.

“A large part of a trader’s success and survival seems to be linked to their physiology. Such a finding has profound implications for how we understand financial markets,” adds Dr Mark Gurnell from the Wellcome Trust-Medical Research Council Institute of Metabolic Science at the University of Cambridge.

“In economics and finance most models analyse conscious reasoning and are based on psychology,” Dr Coates continues. “We’re looking instead at risk takers’ physiology – how good are they at sensing signals from their viscera? We should refocus on the body, or more exactly the interaction between body and brain. Medics find this obvious; economists don't.”

The research was largely funded by the Economic and Social Research Council, the European Research Council and the Dr Mortimer and Theresa Sackler Foundation. Additional support was provided by the National Institute for Health Research Cambridge Biomedical Research Centre.

Reference
Kandasamy, N, Garfinkel, SN, Page, L et al. Interoceptive Ability Predicts Survival on a London Trading Floor. Scientific Reports; 19 Sept 2016; DOI: 10.1038/srep32986
乐乎设计,乐乎生活~

TOP

李彦宏认为人工智能不能只靠一家公司的推动,所以百度已经向社会开放了其深度学习开源平台PaddlePaddle。据悉,这也是继Google、Facebook、IBM后又一家将人工智能技术开源的公司,目的就是让年轻人利用之后不断学习,改进和创新。
乐乎设计,乐乎生活~

TOP

http://www.paddlepaddle.org/
百度的深度学习开源平台。
乐乎设计,乐乎生活~

TOP

近日,百度在开源社区Github及百度大脑平台开放全新深度学习开源平台PaddlePaddle,供全球程序开发者下载使用。继Google、Facebook及IBM之后,百度成为另一个人工智能技术开源的科技巨头,也是国内首个开源深度学习平台的科技公司。这意味着百度将通过PaddlePaddle开源平台将整个社会推进人工智能领域,促进人类发展及社会进步。

在不久前的2016年百度世界大会上,百度首席科学家吴恩达首次宣布对外开放百度深度学习平台,以推动人工智能技术的快速普及。此前,百度已于2013年自主研发了深度学习平台Paddle,但仅供内部工程师研发使用。此次百度开放PaddlePaddle深度学习平台,将其自身在搜索、图像识别、语音识别、自然语言处理、用户画像及情感分析等人工智能领域的优势整合升级,为程序开发者提供了一个功能更全、效果更好的深度学习框架。

百度开放人工智能平台能带来哪些改变?

作为国内人工智能的领头羊,百度的人工智能领域的发展已经将其他对手远远抛在身后,早在2014年开始,百度就逐渐加大了在人工智能领域的投入,目前,人工智能在百度的产品及技术中都扮演着极其重要的角色,如百度外卖、智能数据中心、杀毒、机器翻译、黄反鉴别、OCR、CTR、人脸识别和自动驾驶等多个领域。以百度外卖为例,PaddlePaddle的大数据技术可以为外卖预估出餐时间,合理规划取餐送餐的时间和路径,提高外卖送餐的效率;在反黄鉴别领域,PaddlePaddle可以通过图像识别技术从图片和视频中提取更多关联性特征,将隐晦的色情内容鉴别出来,避免错误判断。此次百度开放深度学习平台PaddlePaddle,将以更低的门槛赋能程序开发者,给予创业者技术支持,让他们通过百度的人工智能技术造福行业,推动社会进步。

“在PaddlePaddle的帮助下,深度学习模型的设计如同编写伪代码一样容易,设计师只需关注模型和高层结构,而无需担心任何琐碎的低层问题。未来,程序员可以快速应用深度学习模型来解决医疗、金融等实际问题,让人工智能发挥出最大作用。”百度深度学习平台PaddlePaddle的研发负责人徐伟如是表示。

正如Facebook的视觉工具开源平台DeepMask+SharpMask和MultiPathNet,利用像素水平分析图像和物体,除了普通人可以使用,也可以帮助视障人群评估照片,描述照片内容。当视觉障碍用户在Facebook信息推送中遇到图片时,将可以听到更加丰富具象的图片描述。这一开源平台的开放将进一步帮助视障人群阅读图片,提升盲人的日常生活水平。

由此可见,开源平台的开放不仅可以推动互联网行业的裂变,也可以改善特殊人群的生活水平。此次,百度摒弃了闭门造车的做法,而把自己定位为服务大众、促进社会发展的基础设施,利用自身技术优势构建商业新生态。未来,将会有越来越多的创业公司使用PaddlePaddle的标准与框架,这些创业公司将与百度的人工智能技术相对接,从而推动国内人工智能产业及整个社会的进步与发展。

百度的技术责任

PaddlePaddle并不是百度首次开源项目。此前,百度曾开放“百度超级搜霸”、“百度通用通信组件”及“Kity Minder”等19个开源平台,而在2015年时,百度牵头微软亚洲研究院、华盛顿大学等知名公司和高校,开放了全球最大分布式深度机器学习开源平台,世界各地的开发者们通过这一开源平台可以免费获取优质易用的机器学习代码,让更多个人和机构享受到人工智能带来的技术便利。

“能力越大,责任越大”,对于百度来讲,如何利用技术创新推动互联网产业,甚至于包括传统产业的前行是其应有的社会责任,PaddlePaddle的开放对于百度来说也并非第一次,百度曾在之前将技术平台开放给盲人创业者,用技术为盲人群体带来真正意义上的生活便利。开放PaddlePaddle深度学习平台,也是百度一贯发挥自身技术优势,促进社会发展的社会责任,也将为商业新生态带来更多的可能性,以良性循环的方式推动社会的进步。
乐乎设计,乐乎生活~

TOP

https://www.wired.com/2008/06/pb-theory/
The End of Theory: The Data Deluge Makes the Scientific Method Obsolete
https://www.edge.org/3rd_culture/anderson08/anderson08_index.html


https://www.edge.org/
挺有意思的网站,寻找知识边界。
乐乎设计,乐乎生活~

TOP

乐乎设计,乐乎生活~

TOP

http://www.swemorph.com
形态学分析法,有空可以尝试
乐乎设计,乐乎生活~

TOP

除了深度神经网络已经实现的特性以外,大脑还有哪些特性是值得机器学习领域借鉴的?
https://www.zhihu.com/question/51807859
乐乎设计,乐乎生活~

TOP

https://www.codecademy.com
编程入门网站
乐乎设计,乐乎生活~

TOP

乐乎设计,乐乎生活~

TOP

http://www.epagogix.com/
预测电影未来的公司
乐乎设计,乐乎生活~

TOP

https://www.musicxray.com/
音乐推荐系统,值得看看。
乐乎设计,乐乎生活~

TOP

发新话题