wh
发帖数: 141625 | 1 你现在读读说不定会喜欢哎，小时候不懂社会政治历史。我们一个老教授说，年纪大了
以后来读king lear，满腹辛酸苍凉，人情就是这么悲凉。喜欢政治历史的人很推荐
coriolanus，说看莎剧历史就跟现在的政治一样。 |
|
N**D
发帖数: 10322 | 2 bioinformatics: search KDD, check its proceedings. basically it is using/
desigining statistics/machine learing/data mining in biology problems, like
analyzing gene expression data, protein secondary structure prediction, etc. |
|
|
d******e
发帖数: 7844 | 4 不要来做这个了。
Machine Learning是CS圈子里20世纪最大的骗局之一 |
|
w***g
发帖数: 5958 | 5 我觉得目前machine learning还是很make sense的。machine learning是噱头，其实就
是data fitting。fit已知数据来预测未知数据，没觉得有什么不对的。不管怎么样，在
各行各业都已经有实际应用。
90年代以前的artificial intelligence更像骗局。 |
|
d******e
发帖数: 7844 | 6 data fitting是统计的人做的。
现在的Machine Learning就是玩玩统计的人玩剩下的东西
，在 |
|
|
N**D
发帖数: 10322 | 8 R is written by stat people, and is the worst software ever |
|
h***i
发帖数: 3844 | 9 做ssimulaton和一些基本的事情足够了，做统计research主要是用纸。 |
|
|
c***a
发帖数: 655 | 11 按你这说法Google也是全世界最大的骗子。
等等。。20世纪？现在21世纪了吧。。。 |
|
|
d******e
发帖数: 7844 | 13 Google可不是单靠Machine Learning起家的。
虽然说IR和ML有一定的联系，但是基于ML的IR十有八九都不是靠谱的东西。 |
|
l*******r
发帖数: 511 | 14 IR跟ML是两个方向吧。虽说有点联系？
能否说下ML的job怎么样，一般找些什么样的工作？
比较菜的问题，希望大侠回答一下 |
|
|
b*********d
发帖数: 139 | 16 ir早有了
google用ml把ir提高了很多，那不叫不靠谱 |
|
b*********d
发帖数: 139 | 17 ml是理论，ir是application
不矛盾 |
|
|
b*******g
发帖数: 131 | 19 其实就是统计啊
我觉得是伪科学
因为你即便数据相关，也不能证明有什么逻辑上的因果性
然后那些人还解释得头头是道就更可笑了
，在 |
|
b*******g
发帖数: 131 | 20 现在还是上课必须 呵呵
我去修统计课，project都是让用R做
结果我们几个CS去的人用matlab写
工作量大不少 |
|
s****i
发帖数: 216 | 21 为啥是骗局, 我觉得用到好数学, 就不是骗局, 换句话说, 把machinelearng 当做是数
学应用, 心里就平衡了.... |
|
R********n
发帖数: 519 | 22 nod，如果model正确且实用，数学很深是最好；
有个还凑或的model，加点数学分析也算不错
model不靠谱，数学没多少，这样的就算了
ML里面前两种的topic还是有很多的 |
|
e***e
发帖数: 168 | 23 if you want to find a faculty job, data mining/ machine learning is good.
For industry, I think statistics degree is better than CS degree on this
topic. |
|
e********r
发帖数: 26 | 24 very very hot at present. almost in all CS major conferences you can find
some ML paper or ML applications paper. almost all major CS companies
recruite ML people. So, very good in job market at present. But not sure in
5 years later |
|
l*******r
发帖数: 511 | 25 我问了些人，说这个东西research lab招的比较多，一般的engineer职位不多。当然
resea
rch是很hot的。不过现在的research topic都是几年换一拨，人全部都过去。不对的地
方大
家帮忙说说，谢谢。
in |
|
j****a
发帖数: 1277 | 26 我个人觉得还是很有意思的 很多很实际的应用可以研究 faculty也可以 09 stanford
有个fresh assistant prof 不就是搞ml的么 |
|
d******e
发帖数: 7844 | 27 找工作太难了，呵呵。ML这种虚招子，在经济不景气的时候应该是最先遭到遗弃
in |
|
w***g
发帖数: 5958 | 28 上面讲的都有理.我再加一点.
我觉得CV不是正统的CS研究领域. 这从CVPR和ICCV都是IEEE的会议可以看出来. 从研究
方向看, 早期的CV跟图像处理更沾边, 而后者则是传统的EE的研究方向. machine lear
ning应用到CV也就是近十年的事情, 那时CV作为EE的一个方向已经确立了.从目前的形势
看, 信号处理的方法在CV能用的也都用了,剩下不少难以解决的问题貌似都跟AI有关, 所
以现在CV反倒成了AI研究的一个前沿领域. 然而作为CS的一个方向, CV又过于年轻, 没
什么成功的应用. 所以没什么牛人在ACM给CV出头. 哪天CV出了一个killer app, 保管会
出SIGVISION.其实在美国大多数事情最终都归结在钱上.别看CV现在搞得人多,其实大部
分人都在耗钱,你们搞CV的应该觉得悲哀.
IEEE和ACM方向不同没法比档次.EE的journal和CS的会议也不能说哪个更难中.这些组织
你交点钱两边都加入了不就行了? 你我作为年轻人更应该注重于提高自己研究的档次. |
|
r********3
发帖数: 2998 | 29 有道理。
CV搞成这样，也有自甘堕落的原因。本来就是一个很狭窄应用的领域，CVPR这些一年录
个200，300篇paper。大牛们不自律，没有起到很好的监督作用。也是导致这个领域太
水太烂的原因之一。
当然最主要的原因还是没有杀手级应用，和现实脱节，不能为社会大众带来利益。
lear
形势
所
管会 |
|
d*****u
发帖数: 17243 | 30 如果做分析数据的工作（machine learing, data mining）微积分还是必须的 |
|
j******y
发帖数: 2578 | 31 有兴趣的可以站内联系我（可以附上简历，如果可以的话），我帮忙推荐。谢谢！
Data Scientist
Engineering | CA, United States
PERFORMANCE PROFILE:
As a Data Scientist, you will use your math wizardry to develop radically
new methods for data access, manipulation, and modeling that will enable us
to provide fair, transparent credit alternatives to underbanked Americans.
Successful candidates will be crazy smart, creative thinkers who are capable
of devising innovative solutions to old financial problems. They will
interface directly bet... 阅读全帖 |
|
y**b
发帖数: 10166 | 32 一直没怎么在乎，最近处理照片较多，忍不住试了试最简单的办法如下，
谁有更好的办法麻烦告一声。
1. 如果用win 7， 有个自带的calibrate color功能，操作很简单，除了色彩，
亮度和对比也可以调整，这可能是最懒最简单的方式了，但试无妨。
2. 用Samsung的MagicTune工具，下载见：http://www.samsung.com/us/consumer/lear
ningresources/monitor/magetune/pop_intro.html
这个工具在x64下有问题，还好只需用到里面的LCDGamma0101.exe程序，步骤如下：
1. install.
2. run LCDGamma0101.exe to tune up.
3. save as MagicTune.
4. multiUser -> MagicTune -> apply.
5. msconfig -> keep GammaTray active. |
|
S***d
发帖数: 1802 | 33 Like you probably know, there are no official OpenOffice.org packages in the
Ubuntu repositories
for the latest version (3.2) and also no packages in the OpenOffice.org
Scribblers PPA (for now at
least).
We wrote about installing OpenOffice 3.2 in Ubuntu from it's official
website but that's not the
same OpenOffice.org as the one in the Ubuntu repositories because Ubuntu
uses Go-OO, an
optimized version of OpenOffice. Well, here is how to install this optimized
version called Go-OO.
You can lear |
|
X****r
发帖数: 3557 | 34 No, it is not exactly copied twice for each push. Some push are more costly
than others.
More hints:
1. Add tracking statements between the pushes, so you know which push causes
what.
2. Print 'this' in the constructor and both 'this' and the address of the
object being copied from in the copy constructor, so you know which object
is copied to where.
After you have done these,
3. Try adding 'circles.reserve(3);' before the first push_back, see if that
changes anything, and think about why.
Lear |
|
l***y
发帖数: 190 | 35 Thank you guys very much for the suggestions. It seems that I must lear WTL
and ATL. Is there any good textbooks for beginers? |
|
m******0
发帖数: 1306 | 36 What I talked is I never use C++, C# before, I only leared C when I was
student. If I can take C# job (noboby teach me in new company) why he can't
transfer from C to C++?
By the way, I forgot to say that I have used 2 years PowerBuilder6.5, Foxpro 2.6/VFP6 7 years. Database I used Orcal, SQL in pasted 9 years. |
|
s*******e
发帖数: 664 | 37 ☆─────────────────────────────────────☆
ilvch (From here to eternity) 于 (Wed Aug 5 00:22:05 2009, 美东) 提到:
在看Petzold
第一个例子是用MessageBox
下面就是经典的 WinMain 和 WinProc 的例子.
可是我把 MessageBox 加到第二个例子的WinMain() 最前面,
跑的时候 MessageBox 关掉后就什么都没有了
这是为什么?
☆─────────────────────────────────────☆
NeverLearn (Libertarian Ace) 于 (Fri Aug 7 13:03:45 2009, 美东) 提到:
Are you forced to write classic Windows apps, or just self-learning? If
it's latter then switch to .Net and WPF instead. You spend less time
lear |
|
k******t
发帖数: 1498 | 38 just started to lear/use it a few weeks back. |
|
g****t
发帖数: 31659 | 39 来自主题: Programming版 - 忙起来了 怎么是乱码啊？我再发一遍个人浅见吧。
atom什么的都是浏览器吧？这种之前本版有人说不少坑。
真正靠谱的notepad++，sublime。
但是sublime不更新了，作者出问题了。
vim／emacs起码没这方面的问题。能折腾的大能多的是。
散户跟着划一下水，写点让自己方便的copy／paste以及
文件备份，debug小工具就够了。这些经验是多种语言
通用的。
人都是会变的。你今天的需求也许和下个月就不一样。
你今天写machine learing 矩阵计算的代码，下个月弄界面，
弄slides了。不修改tool，不科学啊。
所以折腾这些东西，真不是为了折腾而折腾。
总比换一个任务就学一个新的IDE好吧？ |
|
d**********9
发帖数: 11 | 40 Do you mind share with us how to align accounting skills with SQL/VBA/
Advance Access to get $100K job? I have accouting background and recently
have been learing database and VB, but not sure how to transfer those skills
into $$$ and how to combine it to my accouting background.
If you can shed some light, I really appreciate it. |
|
a*******s
发帖数: 3 | 41 I think probably I am the only one here majored in speech and language
pathology.
Basically, master is in the clinical track. You take courses, you need to
have clinical hours, and you need to pass ETS Praxis exam to have the
qualification. Your oral English should be perfect lest your clients
misunderstand you. The courses are not that hard, but may include some
anatomy and physiology, which may be hard since you don't have any
background.
And, to tell you the truth, the phonetics you have lear |
|
b******k
发帖数: 1874 | 42 I am writing to ask for help.
I am using Miltenyi anti-NKp46-biotin-bead- MS column kit. The first step is
to get splenocyte from spleen by gentlMACS without enzyme treatment. A very
simple protocol, just put spleen into the PEB buffer and let gentleMACS run
m_spleen_01 program once, and filter, spin, aspirate, resuspend the pellet.
The problem comes from the resuspending process. The small tight pellet will
produce(or contain) big clump that is almost impossible to break when 10 ml
PEB buffer i... 阅读全帖 |
|
K****n
发帖数: 5970 | 43 已经把他们的 data scientist team 给拆了。true story。
因为不写 code 没有 impact。
这儿不懂行的太多了。不想写code，非想着什么养老。别瞎琢磨了，还是养养鸡养养鸭
吧。
要我老说，还是白天当房产经纪，晚上倒倒秋裤最靠谱。Ma 什么 chine Lear 什么
ning，推了半天公式还不如搬砖发发汗。 |
|
Z**X
发帖数: 366 | 44 know very well about general documents such as ASCE-7, IBC.
know as many design codes as possible. besides common steel (AISC) and
conrete (ACI-318) design, better to lear more codes cush as wood (NDS),
masonry (MSJC), precast (PCI), if you can.
know basic CAD drawing...
if you know bridge design, try bridge companies also... |
|
g****y
发帖数: 199 | 45 ☆─────────────────────────────────────☆
cilyfg (cily) 于 (Fri Sep 28 18:51:22 2007) 提到:
好多人在问 MATLAB 的问题，这里有一本好书，推荐给大家，入门级的。
http://bbs.itepub.net/viewthread.php?tid=154303&fromuid=255718
copy and paste to your internet browser's address bar, you need register to
download the book.
There are many other computer related books. very good website
☆─────────────────────────────────────☆
jzxu (自然) 于 (Fri Sep 28 18:52:41 2007) 提到:
你弄上来，我给你发双黄包子
to
☆─────────────────────────────────────☆
l |
|
f*********d
发帖数: 46 | 46 Thank you for your sharing.
Education itself is kind of teaching and learing process. You share your
experience ,at the same time, you should also learn somthing, I believe.
If more and more friends want to share general topics about education, this
area will benefit more people.
how
Since
well. |
|
f*********d
发帖数: 46 | 47 Thank you for your sharing.
Education itself is kind of teaching and learing process. You share your
experience ,at the same time, you should also learn somthing, I believe.
If more and more friends want to share general topics about education, this
area will benefit more people.
how
Since
well. |
|
i*****t
发帖数: 24265 | 48 deep neural network怎么样？还有deep belief？或者有其他更好的？
多谢 |
|
s***d
发帖数: 15421 | 49 据说上周末在winpro 有行家试验了一下，qcom被完爆。
和GPU 比计算能力， LOL 贻笑大方了。
刚刚在朋友威信上看到打脸全过程。 |
|
|
