变异数分析_人工智能系统中分析变异的祸害

变异数分析

迈向AI生产力的高原 (Towards AI’s Plateau of Productivity)

There is a famous paradox within the classical interpretation of probability theory called the Bertrand’s paradox. Bertrand formulated a very simple problem as follows.

在概率论的经典解释中有一个著名的悖论，即贝特朗悖论 。 贝特朗提出了一个非常简单的问题，如下。

Inscribe an equilateral triangle in a circle and then determine the probability of randomly choosing a chord such that its size is greater than the side of the triangle.

将等边三角形刻成圆形，然后确定随机选择和弦的可能性，使其和弦长大于三角形的边。

Wikipedia) Wikipedia ) CC BY-SA 3.0 CC BY-SA 3.0

He gave three valid arguments to solve this problem but each yielding a different result. Check the Wikipedia article for details on the three arguments, if interested. The resolution to this paradox arises from the fact that each of the valid arguments seems to be addressing the same problem but essentially they are solving different mathematical questions.

他提出了三个 有效的论点来解决此问题，但每个论点都有不同的结果。 如果有兴趣，请查阅Wikipedia文章以获取有关这三个参数的详细信息。 解决这一悖论的原因是，每个有效论点似乎都在解决相同的问题，但实际上它们正在解决不同的数学问题。

I will revisit this paradox in the context of this article, but before that need to highlight some issues in the way we are building AI systems currently.

我将在本文的上下文中再次探讨这个悖论，但是在此之前，需要强调一些我们目前构建AI系统的方式中的问题。

In the ICT industry, engineers are increasingly moving towards building AI systems to add value to customers by solving existing problems and making processes more efficient. With the seemingly successful application of deep learning, experts are opining, with conviction, that the AI winter has finally come to an end.

在ICT行业中，工程师越来越倾向于构建AI系统，以解决现有问题并提高流程效率，从而为客户增加价值。 随着深度学习的成功应用，专家们坚信AI的冬天终于结束了。

But, there are at least three major issues (as also reported by various experts) which we are dealing with while building the AI systems