Forums

empirical mode decomposition..please help

Started by amidala November 10, 2008
hi everyone, million thanks in advance...i would like to know about EMD
which is part one of hilbert huang transform. the procedure of EMD is :

1)For a signal X(t), let m1 be the mean of its upper and lower envelopes
as determined from a cubic-spline interpolation of local maxima and minima.
The locality is determined by an arbitrary parameter; the calculation time
and the effectiveness of the EMD depends greatly on such a parameter. 
2)The first component h1 is computed:
  h1=X(t)-m1 
3)In the second sifting process, h1 is treated as the data, and m11 is the
mean of h1’s upper and lower envelopes:
h11=h1-m11 
This sifting procedure is repeated k times, until h1k is an IMF, that is:
h1(k-1)-m1k=h1k 
Then it is designated as c1=h1k, the first IMF component from the data,
which contains the shortest period component of the signal. We separate it
from the rest of the data: X(t)-c1 = r1 The procedure is repeated on rj:
r1-c2 = r2,....,rn-1 - cn = rn.

the problem is, which iteration result should i apply the hilbert
transform(2nd part of HHT)?
please help me, i`m stuck. thanks again



On 10 Nov, 15:00, "amidala" <raseedaham...@yahoo.com> wrote:
> hi everyone, million thanks in advance...i would like to know about EMD > which is part one of hilbert huang transform.
I did look briefly into the Hilbert-Huang transform a few years ago. My interest faded when I learned that it was patented. One might breach patent laws by discussing it.
> the procedure of EMD is : > > 1)For a signal X(t), let m1 be the mean of its upper and lower envelopes > as determined from a cubic-spline interpolation of local maxima and minima. > The locality is determined by an arbitrary parameter; the calculation time > and the effectiveness of the EMD depends greatly on such a parameter.
Which accounts for the word "Empirical" in "Empirical Mode Decomposition".
> 2)The first component h1 is computed: > &#2013266080; h1=X(t)-m1 > 3)In the second sifting process, h1 is treated as the data, and m11 is the > mean of h1&#2013266066;s upper and lower envelopes: > h11=h1-m11 > This sifting procedure is repeated k times, until h1k is an IMF, that is: > h1(k-1)-m1k=h1k > Then it is designated as c1=h1k, the first IMF component from the data, > which contains the shortest period component of the signal. We separate it > from the rest of the data: X(t)-c1 = r1 The procedure is repeated on rj: > r1-c2 = r2,....,rn-1 - cn = rn. > > the problem is, which iteration result should i apply the hilbert > transform(2nd part of HHT)? > please help me, i`m stuck. thanks again
That's the second reason the term "Empirical" is used for the algorithm: You as analyst is responsible for choosing the parameter(s) that best fits the data. Or your purpose. Or some other arbitrarily chosen parameter. The word "empirical" doesn't sound wuite as bad as "subjective" or "guesswork" (and maybe it isn't), but that's the kind of context where it belongs. Rune
Rune Allnor wrote:
> On 10 Nov, 15:00, "amidala" <raseedaham...@yahoo.com> wrote: >> hi everyone, million thanks in advance...i would like to know about EMD >> which is part one of hilbert huang transform. > > I did look briefly into the Hilbert-Huang transform a few > years ago. My interest faded when I learned that it was > patented. One might breach patent laws by discussing it.
AFAIK there's no restriction anywhere on _discussing_ a patented technology -- there certainly isn't one in the US. The whole point of a patent is to induce the inventor (or thief, depending on who you're talking to) of a technology to put it into the public domain, in return for which the inventor (I'll stop being provocative) gets a monopoly on the technology for some period. That way (in theory) the invention gets put into the public domain so that after the patent has lapsed (and the inventor has presumably made his bundle of money) anyone can go and make the stuff. Even though patent law in most places seems to have morphed into protectionism for big corporations and a stifling, rather than an encouragement, to innovation, I would be surprised that it could be used as a means of abrogating one's freedom of speech. (Non-disclosure agreements are different -- in that case you're signing a contract where you promise not to blab in return for "hush money" in cash or barter.) Now, go and try to _sell_ me a product using patented technology; then some feathers may fly. See http://groups.google.com/group/sci.electronics.design/browse_frm/thread/d138368694e28a9c/60116b2336f4e4ce?tvc=1&q=What+happens+when+you+get+caught+using+somebody%27s+patent#60116b2336f4e4ce; pay particular attention to option "H". Note that even if you _do_ sell me patented technology, no government agency (in the US at least) is going to go after you. It's the responsibility of the patent owner to go after you (and possibly me) to make you (and possibly me) pay for his intellectual property, and it must be done through civil courts. I suspect the law is pretty much the same in any "Western" country.
>> the procedure of EMD is : >> >> 1)For a signal X(t), let m1 be the mean of its upper and lower envelopes >> as determined from a cubic-spline interpolation of local maxima and minima. >> The locality is determined by an arbitrary parameter; the calculation time >> and the effectiveness of the EMD depends greatly on such a parameter. > > Which accounts for the word "Empirical" in "Empirical > Mode Decomposition". > >> 2)The first component h1 is computed: >> h1=X(t)-m1 >> 3)In the second sifting process, h1 is treated as the data, and m11 is the >> mean of h1&#2013266066;s upper and lower envelopes: >> h11=h1-m11 >> This sifting procedure is repeated k times, until h1k is an IMF, that is: >> h1(k-1)-m1k=h1k >> Then it is designated as c1=h1k, the first IMF component from the data, >> which contains the shortest period component of the signal. We separate it >> from the rest of the data: X(t)-c1 = r1 The procedure is repeated on rj: >> r1-c2 = r2,....,rn-1 - cn = rn. >> >> the problem is, which iteration result should i apply the hilbert >> transform(2nd part of HHT)? >> please help me, i`m stuck. thanks again > > That's the second reason the term "Empirical" is used > for the algorithm: You as analyst is responsible for > choosing the parameter(s) that best fits the data. > Or your purpose. Or some other arbitrarily chosen parameter. > > The word "empirical" doesn't sound wuite as bad as > "subjective" or "guesswork" (and maybe it isn't), but > that's the kind of context where it belongs. > > Rune
I like "wild-ass guess". Unless I'm talking to a client; then I say "empirical" or "heuristic". -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Do you need to implement control loops in software? "Applied Control Theory for Embedded Systems" gives you just what it says. See details at http://www.wescottdesign.com/actfes/actfes.html

Tim Wescott wrote:

> Rune Allnor wrote: >
>> The word "empirical" doesn't sound wuite as bad as >> "subjective" or "guesswork" (and maybe it isn't), but >> that's the kind of context where it belongs.
> I like "wild-ass guess". Unless I'm talking to a client; then I say > "empirical" or "heuristic".
"Heuristic" sounds like "hebrew", so it is like something that you could expect from a jew... VLV
Thank You for your reply. I really appreciate it.

I would like to know further regarding patented Hilbet Huang Transform as
u have mentioned before. 

"Rune:
I did look briefly into the Hilbert-Huang transform a few
years ago. My interest faded when I learned that it was
patented. One might breach patent laws by discussing it."

May I know, do you mean that I cannot use Hilbert Huang Transform in my
study? Or is it mean that I cannot discuss it openly once it has been
patented (i.e forum, conference,etc)? I try to search about the issue, but
I can't found any. 

Currently, I have found some latest publication (2008) about signal
processing that used Hilbert Huang Transform. 

I`m looking forward to hear from you and Thank You in advanced for your
coorperation.

Thanks and Regards. 

amidala wrote:
> Thank You for your reply. I really appreciate it. > > I would like to know further regarding patented Hilbet Huang Transform as > u have mentioned before. > > "Rune: > I did look briefly into the Hilbert-Huang transform a few > years ago. My interest faded when I learned that it was > patented. One might breach patent laws by discussing it." > > May I know, do you mean that I cannot use Hilbert Huang Transform in my > study? Or is it mean that I cannot discuss it openly once it has been > patented (i.e forum, conference,etc)? I try to search about the issue, but > I can't found any. > > Currently, I have found some latest publication (2008) about signal > processing that used Hilbert Huang Transform. > > I`m looking forward to hear from you and Thank You in advanced for your > coorperation.
I don't know the law where you are, but I never heard of patents anywhere that prevented discussion or, in practice, experimentation. What a US patent gives the owner is the right to sue foe any damages to his income that an infringer causes. US patents are available for a nominal fee (that used to be $0.25) to anyone who asks, and is invalid if the disclosure doesn't provide enough information so that "a practitioner versed in the art" can reproduce the invention. Rest easy! Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
On 11 Nov, 02:38, "amidala" <raseedaham...@yahoo.com> wrote:
> Thank You for your reply. I really appreciate it. > > I would like to know further regarding patented Hilbet Huang Transform as > u have mentioned before. > > "Rune: > I did look briefly into the Hilbert-Huang transform a few > years ago. My interest faded when I learned that it was > patented. One might breach patent laws by discussing it." > > May I know, do you mean that I cannot use Hilbert Huang Transform in my > study? Or is it mean that I cannot discuss it openly once it has been > patented (i.e forum, conference,etc)? I try to search about the issue, but > I can't found any.
I don't know what restrictions apply regarding patents. Which is why the simple solution, in order to avoid potentially serious trouble, is to not discuss them. I've heard of companies who grant employees who invent stuff that they (the company) patent the right to use the invention "for personal use." Some of the comments in this thread indicate that everyone have this right anyway. But again, I'm not a lawyer so I don't know the details. So *maybe* it's OK if you play with patented technology on your experiments, and *maybe* it's OK to mention what you did on, say, a conference. Now, what happens if somebody, based on such a presentation, asks for your collaboration or help in some project? At this point it's no longer a matter of "personal use" or "experimentation", so with a little bit of bad luck, all hell might break loose. Even if everything else is harmless, merely letting people know that you look into patented technology might cause suspicions among patent holders along the lines of "Do these people use my patent? They don't pay for it..." Which will make no one any good. So the only advice I can give is: 1) Find out if the techniques you use are patented. 2) Find out what restrictions wrt patents apply in your location and line of work. 3) Decide whether you are better served by leaving patented technology alone. 4) Whatever you decide to do, keep quiet about it. Rune
On Nov 10, 12:27&#2013266080;pm, Vladimir Vassilevsky <antispam_bo...@hotmail.com>
wrote:
> Tim Wescott wrote: > > Rune Allnor wrote: > > >> The word "empirical" doesn't sound wuite as bad as > >> "subjective" or "guesswork" (and maybe it isn't), but > >> that's the kind of context where it belongs. > > I like "wild-ass guess". &#2013266080;Unless I'm talking to a client; then I say > > "empirical" or "heuristic". > > "Heuristic" sounds like "hebrew", so it is like something that you could > expect from a jew... > > VLV
wow, I used to think you were a smart guy... guess I ws wrong. Mark
On Nov 10, 5:38 pm, "amidala" <raseedaham...@yahoo.com> wrote:
> > I would like to know further regarding patented Hilbet Huang Transform as > u have mentioned before. > ...
> ... > Currently, I have found some latest publication (2008) about signal > processing that used Hilbert Huang Transform. > ... > Thanks and Regards.
The original paper (1998) on the hht is available at: http://www.keck.ucsf.edu/~schenk/Huang_etal98.pdf At 96 pages it has room for a number of examples. Dale B. Dalrymple http://dbdimages.com