Offer求比较: DS&ML@UCL vs MSCS@UCDavis

MSCS@UCDavis(2 years, CA北部，湾区不远）申请弱渣几乎无科研经历，没想到最后UC系被申爆，能录到招人巨少的'牛'校 UCD

地里ucd的信息似乎不多 没找到院系介绍？且CS似乎偏研究？。。。至于科研经历不多，也录取了, 是很意外


Pros：
- 在CA北部 离湾区较近
- 2年项目，时间长度非常适合刷题找工/转科研，跟"英国难留下"相比，美帝的优势不能再好了
- 真'牛'校,学费不贵(当然不能直接和英国一年的 还是伦敦，作比较)

Cons:
-名气: 跟UCL 比，名气差了点；(虽然UCL有部分水专业在给他招黑)
-实力：CSRANKING网站上，CS的各个方向，a. 即使是传统的CS， UCL>=UCD; b.更不要说UCL ML方向的实力了。(不知道 找工党 看这个网站对不对 ... 但考虑到两个项目可能都自带不少research的色彩）



DataSci&MachineLearning@UCL（1 year, London）
按照UCL所谓“一流九流“专业，MSc 纯ML 和 CSML如果是一流，那DSML是二 三流了，也是硬核项目；因为有著名的deepmind和盖茨比实验室，课程质量肯定是行业一流。

作为UCL ML-related三个项目里 最好录取的DSML，bar也不低，我看了下三个专业录取情况，的确是数本的多((DSML录非数学的稍微多一点)；不过大部分能录取的大佬最后真的不一定去....(反观我， 就很心动——体现在 直接有withdraw 人多滴USC的想法了，我怕我这个弱鸡当炮灰——可能也是因为海本+运气好，踩着线录取了，属于我的"彩票"档)

Pros
- 录了数学本多 项目整体确实是更偏学术，70+%理论和30-%工程,  对于走ML方向的，"learn from best"是个很好的选择：听前辈说，ML DL RL基本是从底层推一遍，非常系统的教学，辛苦学一年 出来直接找算法岗毛问题；而且，自己是CS本+ML算法所需求的CS技能可以课外补， 可以做的工作应该很多。
- ML AI泡沫虽然有，但冬天似乎过去，但未来最终还会是数据+AI的时代？

Cons: 不是很明显
- DSML 毕竟不是纯ML或者CSML， 选课稍微受限制，不能选太难的比如盖茨比(感觉我现在的数学能力，选了怕是要挂)
- DSML的学长似乎也是投国内的岗；据说想去Deepmind实习，得PhD，或者想留在London搬砖，可能不如ML和CSML的学生。(但换个角度横向对比，腐国本来就难留，作为UCL顶级项目之一，能有机会流，算是不错的机会了...)


-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

本人典型的找工向:
"最好是开发+DS/ML结合的，暂时没有指望能安心做学术研究，对于ML相关的技术，掌握+使用就好"

希望各位，如果 对我上述看法有指正的(e.g. 某个想法,不符合现实/业界规则?)，对UCD MSCS具体情况有了解的，对AI ML DS行业未来风向有建议的，等等，能给点建议 感激不尽 笔芯~


补充内容 (2019-4-18 18:40):
因为腐国的选课很清楚，附一个DSML CSML(可以互选) 链接


补充内容 (2019-4-18 18:41):
http://www.cs.ucl.ac.uk/prospect ... d_machine_learning/
http://www.cs.ucl.ac.uk/prospect ... d_machine_learning/

补充内容 (2019-4-18 18:42):
上面第一个是DSML的
CSML放这：http://www.cs.ucl.ac.uk/prospect ... d_machine_learning/

YuhaoNg 发表于 2019-4-19 06:14
Research experience for PhD and live happier?

so UCD indeed has more strengths in the respect  ...

UCD has 2 years while UCL only has 1 year. For freshman, it means you need to apply for PhD the moment the master starts. If so, why not apply for it now. If you have 2 years program, then you can learn and do research for about 1.5 years then make the application. Considering the competition for ML/AI PhD is quite strong, it is only safe after you have at least 3 publications as master applicant. And if you choose to work for 1 year then attend UCL, i am afraid it maybe diffcult to find a suitable job in half year based on your current background. Even you get the job, how much time left before you start UCL? I know the diffcult thing is the UCD has general much lower rank than UCL. But i think brand name is not everything. What matters is what you can get here and whether it is useful for your career in the future. The US has much better job markets and higher pay and you can work here at least 3 years after graduate. While, for UK, it is 6 months. And UCL maybe not much stronger than UCD though do stronger in ML. I know Deepmind is in London. But UCD is close to silicon vally. And silicon vally has much more than deepmind.

感谢学弟分享，不在意distinction平时划水的话，ic还是可以艰难挤出时间的

Davis, if you want to get a good job, enough research experience for PhD and live happier. UCL, if you really love UK and London and ready to go home after graduate.

vote for UCL

学校实力/科研能力UCL>>UCD（ML方面确实厉害，感觉至少相当于US Top20，可能Top10）。找工作方面要分开来看，如果想找国内工作UCL无疑（名气大），但国内工作环境相对恶劣，再加上AI泡沫算法工程师未来的方向也不确定；想找国外的工作UCD肯定是个更好的选择，毕竟在加州，再加上传统的cs系，努力刷题感觉能找到还不错的工作，而UCL毕竟在腐国找工不容易。综上，楼主一心想在国外工作的话建议去UCD，不然建议UCL。

BirkhoffG 发表于 2019-4-18 22:22
学校实力/科研能力UCL>>UCD（ML方面确实厉害，感觉至少相当于US Top20，可能Top10）。找工作方面要分开来看 ...

好的 谢谢老哥的advice。

第一 我UCL可以申请defer，来弥补自己再知识上和经历的不足，这样准备充分后 来London读完 就可以直接投当地的了，应该比不defer就去读仓促准备好很多, UCL ML这几个load都大

第二 我在想如果以后想去北美就业吧，可以找个短期项目做二硕，家里还是可以负担的起的。

huanpassword 发表于 2019-4-19 04:48
Davis, if you want to get a good job, enough research experience for PhD and live happier. UCL, if y ...

Research experience for PhD and live happier?

so UCD indeed has more strengths in the respect of research than professionals?

But, why happier?

huanpassword 发表于 2019-4-19 08:39
UCD has 2 years while UCL only has 1 year. For freshman, it means you need to apply for PhD the mo ...

Really appreciate your specific reply !

For the UCL machine learning stuff, what if I don't wanna apply for a PhD?
Or say, in terms of professionals, can I do the SDE with the ML techniques(my MSc DSML is meant for more applications of ML than research work,  not as the CSML or pure ML)?

YuhaoNg 发表于 2019-4-19 19:59
Really appreciate your specific reply !

For the UCL machine learning stuff, what if I don't wan ...

You are talking about another advantage of choosing US program. The UK program normally only specific to one area. So, if it is data, then it is data. You can only learn quite few about normal CS courses. While, CS program normally aims to prepare you for SDE rather than ML engineer. Even you may lost interest in ML finally, it is also a very good option to learn SDE. Are you sure UCL have courses like OS/Network/Database/Algorithm/SDE/Compiler/website?

huanpassword 发表于 2019-4-19 22:48
You are talking about another advantage of choosing US program. The UK program normally only speci ...

Yep, exactly no CS course(That's why the department gives more offers to the Math-majored undergrads). Though I am in a CS major,  it is not a strong one and I believe that anyway, I will review all the CS basis knowledge.

How about studying the SDE-reuqired professional techniques on my own, like taking the MOOC courses, (well, I found most of them are for the undergrad XD...)? Or by reading classic textbooks?  The "Self-study" does matter for ANY CS-related areas.