which control theory textbook ? (for reference) - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>EE - 电子工程

which control theory textbook ? (for reference)

which control theory textbook ? (for reference)# EE - 电子工程

m*w2010-05-04 07:05

1 楼

【以下文字转载自 Stock 讨论区】
发信人: mountainview (山景城), 信区: Stock
标题: BGI deal could land windfall for army of PhDs
发信站: BBS 未名空间站 (Fri May 15 12:26:59 2009)
英文不太好，这land windfall啥意思？谁给翻译一下啊？
BGI deal could land windfall for army of PhDs
By Claire Milhench and Steve Slater
LONDON, May 15 (Reuters) - The biggest ever fund management takeover could
land windfalls for Barclays Global Investors' (BGI) small army of PhDs,
renowned for designing its complex products to beat the stock market.
Once known as a s

P*y2010-05-04 07:05

2 楼

诚聘通勤或者住家保姆，主要工作：照顾2月小女孩，一顿晚饭及简单家务，家在
fairfax，邮编22181，要求性格温和，有爱心，待遇从优，有健康检查者优先。有意
者请将个人情况和要求发邮件至p*******[email protected]

r*t2010-05-04 07:05

3 楼

【以下文字转载自 ChineseMed 讨论区】
发信人: rabitrabit (兔子), 信区: ChineseMed
标题: 白血球数低
发信站: BBS 未名空间站 (Wed Jan 6 15:21:41 2010, 美东)
低于正常值。其他指标正常。
是什么原因？应该怎么改进？

s*y2010-05-04 07:05

4 楼

☆─────────────────────────────────────☆
sasa (上来透口气) 于 (Wed Nov 18 17:54:21 2009, 美东) 提到:
我有很多音乐CD，想把他们弄到硬盘里面，请问有什么好的建议（用什么软件）。我现
在也没有想清楚是做成ISO，还是就烧成mp3或者什么，只是觉得CD做成mp3，是不是可
惜了？有什么好点的办法？
☆─────────────────────────────────────☆
snowsunny (困了就睡吧) 于 (Wed Nov 18 18:09:18 2009, 美东) 提到:
卡卡和李察克同学比较懂行，我就不说话了

☆─────────────────────────────────────☆
sasa (上来透口气) 于 (Wed Nov 18 18:18:23 2009, 美东) 提到:
你也说两句嘛。你现在主要还是听CD？
☆─────────────────────────────────────☆
LeChuck (No Itch No!) 于 (Wed

a*a2010-05-04 07:05

5 楼

吭哧吭哧来回搬花盆推累了。。555

s*x2010-05-04 07:05

6 楼

【以下文字转载自 Fashion 讨论区】
发信人: liubros (liubros), 信区: Fashion
标题: 【墨镜墨镜墨镜】摄影版的过来参加活动
发信站: BBS 未名空间站 (Fri Sep 13 12:30:03 2013, 美东)
两个墨镜。

m*p2010-05-04 07:05

7 楼

如果用K fold CV 的话，会得到K个model，然后再实际应用中应该选哪个model呢？用
accuracy or AUC 最高的那个？

C*S2010-05-04 07:05

8 楼

【以下文字转载自 Military 讨论区】
发信人: xuedi2006 (southman), 信区: Military
标题: 又一中国科学家被FBI调查
发信站: BBS 未名空间站 (Tue Sep 8 09:43:11 2015, 美东)
http://www.dispatch.com/content/stories/local/2015/09/08/fbi-in
FBI investigates China ties of Ohio State professor who resigned,
disappeared
Professor Rongxing Li was a star at Ohio State University, attracting
international attention as he helped NASA rovers explore Mars in the past
decade.
Then, early last year, Li quit his post as OSU’s premier mapping expert and
disappeared. No news release was issued to explain his departure, and most
information about his 18-year tenure at Ohio State was removed from the
university’s website.
Now, federal search warrants filed in U.S. District Court in Columbus reveal
that the FBI was investigating Li, trying to determine whether he shared
defense secrets with the Chinese.
Li, 56, a U.S. citizen who grew up in China, had been director of the OSU
mapping and geographic information system laboratory. Also known as Ron Li,
he held an endowed chair in the OSU Department of Civil, Environmental and
Geodetic Engineering and was known worldwide for his work in mapping.
His renown came, in part, when NASA selected him to help with its 2003 and
2009 Mars exploration missions. His troubles, too, began with NASA projects.
OSU, the FBI and NASA declined to comment about the case. But search
warrants unsealed in August lay out why the government took a hard look at
Li.
In January 2014, Li submitted a $36.9 million proposal to NASA for imaging
work for a 2020 Mars mission. As part of that proposal, Li had access to
Department of Defense technical information that he was prohibited from
sharing with the Chinese, according to search warrants.
Li had claimed in the proposal that he had no relationships with Chinese
scientists. But OSU researchers knew he had spent 2012 on sabbatical at
Tongji University in Shanghai, so Ohio State began an internal investigation
to determine why Li had not notified NASA of his China connections.
Investigators determined that Li had numerous, ongoing connections with
Tongji, including being listed as a professor and as the director of a
center for spatial information. The investigators also found online evidence
that he had collaborated with Chinese-government programs to develop
advanced technologies, serving as chief scientist for one project.
On Feb. 15, 2014, Li notified Ohio State and NASA that he was withdrawing
from the Mars 2020 project. He also told Ohio State that he was in China
caring for his sick parents. A few days later, he emailed his resignation to
the university. According to a search warrant, he said, “With this email I
resign from my position at the Ohio State University.”
Jeff Grabmeier, a spokesman for university research, confirmed Li’s
unexpected resignation but said he couldn’t discuss it further and could
not say why Li left.
According to the search warrant, Ohio State then called the FBI because of
the “unusual circumstances of Li’s departure and the restricted and
sensitive nature of some of his research.” The university told the FBI that
Li had access to International Traffic in Arms Regulations information with
NASA and with Raytheon, a defense contractor.
As part of the FBI investigation, Homeland Security agents stopped and
searched Li’s wife, Jue Tian, 56, in San Francisco before she boarded a
plane for China on March 1, 2014. Agents seized Tian’s computer, a
cellphone and several thumb drives. The thumb drives contained restricted
defense information, the warrant says.
Investigators also searched Li’s home in Upper Arlington. According to
Franklin County property records, Li and his wife purchased the five-bedroom
house on Lane Road in 1997.
Assistant U.S. Attorney Deborah Solove said no charges have been filed
against Li or his wife. She would not comment further.
Charles Toth, an OSU researcher who worked in the same department as Li,
said Li ran “his own show” at the university and didn’t work with many
other professors. He said he couldn’t say more because of “the sensitivity
” of the case. Other professors contacted by The Dispatch didn’t return
phone calls.
In 2014, Li was one of five scientists named as fellows for the American
Society for Photogrammetry and Remote Sensing. His name, however, is no
longer on the society’s online list of 2014 fellows.
Society President Stephen D. DeGloria said he knew Li had left Ohio State
and was living in China but had no details of the FBI investigation. He said
Li had been nominated as a fellow before he quit at Ohio State.
The Upper Arlington house has been for sale since June.
Tian, who returned from China in mid-2014, did not respond to a request for
an interview.

D*a2010-05-04 07:05

9 楼

I am writing a document which have description of block diagrams. I need to
refer to a control theory book for that. Please let me know which control
theory book is classic, so I can refer it in my document. Thank you!

P*y2010-05-04 07:05

10 楼

继续找

g*j2010-05-04 07:05

11 楼

Should be fine if it is only slightly below normal and you have no other
symptoms. If you are concerned, discuss with you doctor and have it checked
in 4 weeks. I bet it will be normal.

z*72010-05-04 07:05

12 楼

二月份会有最冷的温度.

s*x2010-05-04 07:05

13 楼

D*r2010-05-04 07:05

14 楼

这个是比较主观的
一般就直接用accuracy

【在 m*p 的大作中提到】

: 如果用K fold CV 的话，会得到K个model，然后再实际应用中应该选哪个model呢？用
: accuracy or AUC 最高的那个？

l*g2010-05-04 07:05

15 楼

Automatic Control Systems, Benjamin Kuo

to
control

【在 D*******a 的大作中提到】

: I am writing a document which have description of block diagrams. I need to
: refer to a control theory book for that. Please let me know which control
: theory book is classic, so I can refer it in my document. Thank you!

r*t2010-05-04 07:05

16 楼

一个月前检查是3000，最近刚检查变成3300. 好一些但还是低。主要是什么因素造成
？应该怎么改善？

checked

【在 g*****j 的大作中提到】

: Should be fine if it is only slightly below normal and you have no other
: symptoms. If you are concerned, discuss with you doctor and have it checked
: in 4 weeks. I bet it will be normal.

s*d2010-05-04 07:05

17 楼

未必，
今年反常的暖和
我觉得冬天就这么过去了

【在 z******7 的大作中提到】

: 二月份会有最冷的温度.

e*s2010-05-04 07:05

18 楼

皮肤挺好。

s*w2010-05-04 07:05

19 楼

In my understanding, k-fold cross-validation is used to estimate the general
ization error. So you do not really just pick the best one from one fold.
If you have a certain set of paraemters for your model, and an evaluation cr
iteria for your model, then k-fold c-v is able to tell you which set of para
meters got what evaluation.
since k-fold came into being from random parition into k-folds, it is better
to run multiple times of k-fold c-v and got a mean and std of your evalutat
ion results from all runs, where each run has a single number

【在 m*p 的大作中提到】

: 如果用K fold CV 的话，会得到K个model，然后再实际应用中应该选哪个model呢？用
: accuracy or AUC 最高的那个？

n*s2010-05-04 07:05

20 楼

always got people asking why my blood cells are low, why blood pressure are
low, why blood sugar are low. most of them are normal. normal range of any
parameters are man-made, something like mean and SD. so don't worry. you
should be OK.

c*p2010-05-04 07:05

21 楼

我们夏天了，nnd

p*e2010-05-04 07:05

22 楼

不是每个fold选一个model/parameter，
而是选一个model在所有fold上平均最好,
最后在把这个model在所有数据上训练一遍。

【在 m*p 的大作中提到】

: 如果用K fold CV 的话，会得到K个model，然后再实际应用中应该选哪个model呢？用
: accuracy or AUC 最高的那个？

g*a2010-05-04 07:05

23 楼

下周降温

【在 c***p 的大作中提到】

: 我们夏天了，nnd

m*p2010-05-04 07:05

24 楼

thanks guys.
one way is to average K model outputs. that's easy to understand.
what do you mean by "而是选一个model在所有fold上平均最好"?
could you please explain it using linear SVM as an example? thanks!

【在 p*********e 的大作中提到】

: 不是每个fold选一个model/parameter，
: 而是选一个model在所有fold上平均最好,
: 最后在把这个model在所有数据上训练一遍。

y*82010-05-04 07:05

25 楼

这个难说,我觉着还会有几回.TX气候最捉摸不定,记忆中有一年三月里还下了回冰雹,把
我的黄瓜苗都打坏了.

【在 a****a 的大作中提到】

: 吭哧吭哧来回搬花盆推累了。。555

p*e2010-05-04 07:05

26 楼

假设对linear svm的参数C，要从C1到Ck中找到最好的一个值，
先把数据分成N份，每次拿出1份测试，剩下的N-1份训练，一共有N中不同的拿法。
对于每一种拿法，训练对应不同参数C的linear svm，并在剩下的1份上测试，得到
accuracy。所以对于每个参数值，都得到N种拿法对应的N个accuracy，最后对应平均
accuracy最高的那个参数值作为选定的C，再到所有N份数据上训练。

【在 m*p 的大作中提到】

: thanks guys.
: one way is to average K model outputs. that's easy to understand.
: what do you mean by "而是选一个model在所有fold上平均最好"?
: could you please explain it using linear SVM as an example? thanks!

y*82010-05-04 07:05

27 楼

小杯子不是也在TX啊?好像是SAN ANATONIO?

【在 c***p 的大作中提到】

: 我们夏天了，nnd

m*p2010-05-04 07:05

28 楼

ic. that's the grid search in libsvm...
how about for ensemble classifiers like adaboost? there is no simple
parameter like C.
Thanks!

【在 p*********e 的大作中提到】

: 假设对linear svm的参数C，要从C1到Ck中找到最好的一个值，
: 先把数据分成N份，每次拿出1份测试，剩下的N-1份训练，一共有N中不同的拿法。
: 对于每一种拿法，训练对应不同参数C的linear svm，并在剩下的1份上测试，得到
: accuracy。所以对于每个参数值，都得到N种拿法对应的N个accuracy，最后对应平均
: accuracy最高的那个参数值作为选定的C，再到所有N份数据上训练。

a*a2010-05-04 07:05

29 楼

这是什么时候的事？

【在 y*****8 的大作中提到】

: 小杯子不是也在TX啊?好像是SAN ANATONIO?