Configure PyCharm and get started with Tensorflow and Tensorboard – 配置 PyCharm IDE,开始整 Tensorflow

Stackoverflow 上的回答 – The answers in Stackoverflow

刚回答了一个问题,分享给和我一样的 Tensorflow 文盲,快速入手 Tensorflow

Just answered a question of PyCharm and Tensorflow dev. Share it with newbie of Tensorflower like me.

https://stackoverflow.com/a/47049834/931908

 

 

为什么学习 AI ?  — Why AI today?

前些天,前室友提到他们单位(铁道部机车轮组损坏程度智能侦测)花钱请外面的高手提供高级图像识别算法,以更加精确的识别列车车牌序号。

A couple of days ago, my ex-roommate mentioned they spent some budget hiring OCR experts to provide more accurate image recognition algorithm to their company for better train license plate recognition (Computer detection of train wheel damage / defects). 

我跟他说,人工智能领域有图像识别的最新成果——谷歌开源的 Tensorflow ,可以做到免费轻松实现他需要的功能。当前, AI 领域的图片识别精确率,甚至超过人眼;为什么还要使用传统的 OCR 呢?

I told him, there are a lot of research outputs of image recognition in AI profession — Google has open-sourced Tensorflow which could easily solve their problem in a very high accuracy, even better than human eyes. Why bother traditional OCR technicians?

况且他有大量的一手图片资源,有了这些数据,就可以不用求人的。使用 AI 进行铁道系统,他还可能为公司做出重大贡献(提高体制内产品竞争力)。

No mention that he has a lot of images of train license plates all over China, with which could be used to train AI neural network. He could even potentially make a great positive influence to get the most advanced AI technology anticipated into China train system.

这是最近遇到的一个可以应用 AI 的业务。AI 理论,就像其他数学基础理论一样,如果没有全社会来热情参与,结合各行各业的实际需求,实现理论与实际的结合,产生更大的生产力——理论,可能几十年之后,仍然只是停留在理论层面(AI 几十年前就有,只是因为算力限制和业务实施限制,一会热一会衰,多年来似乎更多是雷声大雨点小)。

This is just a business need  which could be fulfilled by AI I encountered those days. As like other mathematical theories, theory of Artificial Intelligence could possibly stay in “useless” theory papers without engaging of the widest landscape of all career professions in coming years — just as like what it was decades ago. That’s why even software engineers like me need to kick started with it today. 

相信有一天,我们公司软件产品的设计、开发、测试、用户习惯反馈(预测)、 DevOps 能或多或少应用一些 AI。只是那时再学 AI 基础知识,恐怕为时晚矣。

What if one day, AI sheds its light over software product application ranging from design phase to development staging, test staging, user experience prediction and / or DevOps in my own company? It would be too late for me to learn AI fundamental knowledge that day.