Learn from master's works

Wu Enda, an expert at artificial intelligence, was once the chief scientist of the former Baidu as well as the founder of Google's brain. He answered questions about how to engage in artificial intelligence and work in machine learning filed in the American Q & A website Quora. As for how to learn the knowledge in the new areas, Wu Enda also put forward some suggestions which could be applied to learning in non- technical areas.

Wu Enda believes that a good way to think about new projects is to spend time doing research on the previous projects. The brain is amazing, on account of the fact that when you further study a lot of examples of a job, you can generalize and come up with new examples of such problems. That's why many artists learn to paint by copying the masters’. If you visit an art gallery, you sometimes see art students sitting on the floor copying art works displayed on the wall. Similarly, many researchers learn to invent new algorithms（算法）by reproducing the results of old research papers. "For me, it's because I've seen a lot of actual machine learning cases at different companies, and currently I'm able to find new opportunities for companies which need transitions in machine learning on a regular basis," he says.

So, if you want to know how to do interesting projects, you can read or reproduce your favorite previous projects, then you will begin to create your own ideas. In conclusion, Wu Enda also suggests that when you accomplish an interesting project, you should write a paper or blog article, or open source code online and share it with the community. Only this way can others learn from you in turn. In addition, you can get more feedback to speed up your study.

方法：复现大师作品来学习

人工智能专家吴恩达曾经是前百度首席科学家，他也是谷歌大脑创始人。他在美国问答网站Quora上回答了网友关于如何从事人工智能以及机器学习领域工作的问题。对于如何学习新领域的知识，吴恩达也提出了一些方法建议，这些方法也适用于非技术类领域的学习。

吴恩达认为，想新项目的一个好方法是花时间研究以前的老项目。大脑很神奇，当你学习一类工作中的很多例子之后，你就能概括并想出这类问题的新例子。这也是为什么很多艺术家通过复制大师作品来学习绘画——如果你参观美术馆，你有时会看到艺术生坐在地板上临摹墙上展示的艺术作品。同样，很多研究人员通过复现旧研究论文的结果来学习发明新的算法。他说：“对于我来说，正是因为在不同的公司看到了很多实际的机器学习案例，我现在才能定期为机器学习转型公司找到新的机会。”

所以，如果你想知道如何做有趣的项目，阅读或者复现你喜欢的以前的旧项目，你会开始产生你自己的想法。吴恩达还建议，最后，当你完成一个有趣的项目时，写一篇论文或者博客文章，或者在网上开源代码，与社区分享。这样，其他人可以反过来学习你。此外，你还可以获得更多的反馈，从而加速你的学习。

trans1