Cousera

To understand machine learning,  I have discovered the cousera website, which accumulates the lectures from some well-known universities.

During registration, I discover the way that put out the form is very user-friendly, really put user in mind.

I discovered there are several things that worth mentioned in the lecture given by Andrew Ng.
First, he has minimal assumption on what students know - e.g., it assumes student does not know calculus and linear algebra. Second, the contents of the lectures are more on application of machine learning, and for maths, it use example to concretize our understanding; for advance level math that do not hinder our understanding by not knowing their working details, it omits them, and only provides the general idea for us. Thirdly, it uses electronic pen to draw picture, and this makes him easier to communicate the idea to us. Forth, it provides some review questions in the midst of the talk, to let us review the materials immediately.

The video player is also very good - it provides function to increase/decrease speed of video, where such functions are necessary for lecture video so we can quickly get through the contents that are easy or is already known by us. It also retains all the slide materials that are used in the video, so that students could download.

All in all, it puts students in the shoes, let students understand the materials in a shorter time, and it also uses interesting example to increase the interest of student.
Nothing is perfect. One thing that worth improving is to provide summary note for lecture contents so that people can revise from those notes. Lecture slides provides many details, but it is not useful for efficient review for knowledge that we have learnt. Furthermore, the playback of video sometimes make the screen stagnant, while the audio keeps going.  Sometimes, it just buffer till certain stage, and it won't be playing anymore, need to reload the screen to make it playing.


最近为了了解machine learning，发现这个大学教科网站，聚集各所著名大学的讲课。
这个网站做得相当不错，相当为用户考量。
register填form的时候，发现用最少的功夫，很容易就将他填好了。我觉得Andrew Ng讲课有几点值得学习的。第一点，对学生做minimal assumption，基本都是assume学生不懂calculus 和linear algebra, 这个我觉得还是挺好的，如果是比较难，或者比较简单的，会有一些 optional 的章节，然后让学生optional的章节会说些什么，让学生决定要不要看；第二点讲课内容偏实际，着重在实际运用，数学方面用example来concretize 我们的观念，一些比较深，但是不太需要知道的观念，就进行省略；第三点，就是用电子笔花图，来着重我们的了解；第四点，lecture中会有一些习题，让我们能够及时复习思考。

Video Player 的制作也相当不错。能加快，变慢，lecture video 最需要这样的功能了。易懂的内容，就给他赶快的过去。同时也能share 和 provide comment。那些在教科过程中用电子笔画过的slide，也保留下来让学生下载。

整体来说，很为学生着想 - 让学生用最少的时间，来了解，并懂得运用。用各种生动的例子，来增加学生的兴趣。

可以进步的地方是，可以有个summary note，让学生容易复习所讲重点。有时候slide比较着重的是讲的过程，没那么着重在重点，summary note 重点在去其糟粕，取其精华。