Machine Learning and Data Mining *  COMP9417

14 Comments

    1
ExtraSugar   Feb. 6, 2020, 1:20 p.m.

(19T3)「基本推荐」:Gelareh Mohammadi教(也是第一次教这门课),上课节奏偏快,如果没有机器学习基础的话听课会很容易抓不到重点,但整体讲的还算清晰,recording质量也不错;问题回复比较及时,PPT里会有些笔误需要自己看,(有些上课时提到的笔误,也会在课后更新PPT)。个人的Assignment不难,团队(五人)的Assignment需要分工写二十页左右的report。课程内容是机器学习的入门课,会涉及到Naive Bayes, k-NN, SVM, decision tree, k-means 等这些模型,介绍机器学习的基础。但考试非常难,更多的考察知识点之间的关联和对模型的整体把握(比如我们考了列举Autoencoder和PCA之间两个优点和两个缺点),需要对课程内容有整体认知而不是仅仅对着讲义复习,否则考试会非常痛苦。建议配合吴恩达的ML公开课看。除非ML基础真的很好,否则这门课需要大量精力去学习。


原文标题:UNSW MIT8543课程评价和选课推荐|落絮飞雁的个人网站

原文链接:https://www.luoxufeiyan.com/2019/12/24/unsw-mit8543-courses-review/

授权协议:创作共用 署名-非商业性使用 2.5 中国大陆

除注明外,本站文章均为原创;转载时请保留上述链接。


    2
Anonymous   Sept. 30, 2019, 12:01 p.m.

首先说一下,下面有个哥们,不知道是真厉害,还是报复心理,在很多课里刷 简单,随便90。大家还是要根据自己的实际情况做出判断。


9417 老师的讲课水平还是可以的,大部分该讲的东西都涉及到了,有的recording还有字幕,所以对某些同学来说,看recording会更好。大部分讲的都是西瓜书中的东西,结合着看效果更好。

平时作业也没有那么费劲,两个小quiz基本上理解一下,或者是相会讨论一下,答案很快就出来了。ass 属于小组作业,有很多题目可以选择,难度不等,可以选择一些有经验的同学一起组队,挑战一下高难度,或者是几个人一起做个一般的。作业不仅仅看重代码,更多的还有report,所以report写的好,也会得个高分。

吐槽一下考试,据说是一个tutor出的题目,很多东西都算半天才能出结果,难度比往年大了不少,而且卷面上还有错误。导致考完之后,很多同学都去argue了。最后还是提了一些分,如果还是这个tutor出题,想拿高分的同学需要多用点心复习。

    3
Anonymous   Sept. 8, 2019, 4:56 p.m.

I disagreed with some comments below, the course was not easy at all, unless you did machine learning before, you have to teach yourself to get a good result. And besides, the final exam was a really disorganized one. It is very misleading to say that the course is easy (even if I got 90+ in the end)! However, D.r. Mick Bain was really patient during office hours, if you have any questions, just go to the office consultation hours every week, I think this is the most efficient way to understand the algorithms. In the end, if you work hard enough, you will get a good result.

    4
Anonymous   Sept. 8, 2019, 11:40 a.m.

考试简单 作业简单 随随便便就90+了

    5
Anonymous   Sept. 6, 2019, 12:51 p.m.

很容易就挂了 考试时间严重不足 哎 

    6
Anonymous   Aug. 27, 2019, 12:42 p.m.

This course prefectly demonstrates what disorganized means. The erroneous and inconsistant final exam was a complete humiliation to student's hard work.

    7
Anonymous   Aug. 20, 2019, 11:08 a.m.

I wish someone had told me not to enroll. Now I'm paying 5k for a retake

    8
Anonymous   Aug. 10, 2019, 5:10 a.m.

terrible experience

    9
Anonymous   June 4, 2019, 12:55 p.m.

垃圾课,老师不负责!

    10
Anonymous   June 4, 2019, 12:55 p.m.

垃圾课,将是不负责!

    11
Anonymous   March 27, 2018, 12:14 a.m.

Definitely the worst course offered by CSE that I've ever experienced. 

说难听点简直就是狗屎, 课件内容一团糟, 每节课里面有100+页, 讲的完全不线性, 像是Machine Learning的百科全书的总结, 缺乏引导性. 

Lecturer(Michael Bain) 极度不负责, 课件从来不按时上传, 基本都是上课前5分钟才让你看到. Assignment release 一周之后题目文件才终于不是404, 论坛中的问题回复延迟基本是3-5天, tutorial materials到时间也不上传.

虽然说现在很多人盲目跟风ml, 但这门课开设出来真对得起300+人疯抢吗? 这5000多刀交的还不如自己去看cs229去看cousera.

也不要求跟os, ds这种课去比, 相比之下, 9318的Wei Wang虽然讲课也**真**烂, 但是能感觉到他有在努力讲课(笑. 

    12
Anonymous   Dec. 23, 2016, 2:22 p.m.

9417的老师每天上课都很累的样子,讲的也是有气无力的,这课一听就觉得老师是白天刚上完班,晚上过来上课,我知道的班上的很多同学都没在听这个老师讲课,都是自己上youtube按照课件的内容学。这课的作业和期中考试成绩出的也很慢,老师也多次为自己的低效率道歉,但是好是好在这门是学院里为数不多的不是double pass的课之一。这门课学的内容和Data Warehousing & Data Mining有大概3次课的内容重叠,但是两个老师讲的方向不十分一样。这课讲的内容是很好的,但是多数时间是要自学。



作者:陈迪威
链接:https://www.zhihu.com/question/24357695/answer/132740090 
来源:知乎
著作权归作者所有,转载请联系作者获得授权。


    13
FromFacebook   May 24, 2016, 10:55 p.m.

From Dan's Comp electives


- apparently really fun (see thread on csesoc) - if you are okay with some math?
- I disagree. A vacuum cleaner is less monotone than the lecturers, and the content doesn't make up for it.
- Waste of time. The exam is rote learning formulae, the assignments are boring, and the theory is only average.

    14
caixiangyu   Feb. 3, 2016, 2:43 p.m.

听说过一个人选,貌似就是做题,没有代码的东西,想选人工智能这个方向的可以选,但是貌似我们专业的人工智能方向比较水,因为讲不深,没太大用。


Your Comment

 or  Login first
(Send email to [email protected] to modify ur anonymous answer)