SETIlive.org - by the SETI Institute

I registered, read the tutorial, examined and reported back on a number of waterfall images at seti live, yesterday. I saw several vertical or nearly vertical lines in a few of the images. These are obviously artificial radio emissions. Most slanted slightly from the vertical, indicating doppler shift due to relative motion between the receiver and the source. A few appeared to be entirely vertical. One image contained a number of parallel vertical lines, which seem to indicate simultaneous emissions on a number of discrete frequencies. All of the lines I saw yesterday were fairly distinct. I wonder if an automated search program could have missed any of these. Michael

Going back to SETI Live and doing some more work yesterday, I found that I was seeing some much subtler lines; things that a computer might reasonably have missed. Very possibly part of learning to work with this system. ( For some reason, human-like visual interpretation has been one of the more difficult things to teach computers to do.) I wonder, of course, if I am seeing valid, weak traces of radio emissions, or am being misled by the well-known human propensity to impose order on a truly random collection of bits and pieces. In many instances there is no certain answer to this question. Signal or noise can be a judgement call here. Michael

Michael Watson,
I like your logical way of thinking. I'll be honest here, if SETIlive can get people who think logically like yourself, they will have one very powerful team of citizen scientists. It will be way more powerful than any computer! Why.... Because human beings have intuition!

John.

Signed up, but their tutorial seems to be having issues loading some graphics, will try their site back a little later today. Perhaps it is just getting too many hits right now.

Thanks for the link and information!
____________

I know we are not alone, I just hope to see proof in my lifetime...

So if by means of observation you think what you are seeing are "valid, weak traces of radio emissions", how would such weak traces show up or possibly be detected by means of the Seti@home client being run by a user?

Is it really supposed to be all noise or do you ever get a clean signal?

My guess is that scientists are trying to be able to obtain waterfall plots of a signal as it is being detected. I assume you are trying to detect a pulse of interest and from then trying to determine whether such a pulse is carrying something of interest.

Does it possibly show up in the numbers being reported, or may it possibly be visible in the graphics as well?

Anyway, what is the point of having the graphics showing up in the window which comes with a running Seti@home task containing the corresponding graphics for such a running task?

Is it ever meant to be showing anything of possible significance? The way this graphics is showing up on a users monitor may be as a result of the graphics skin which has been selected in the preferences of the account for such a user.

This looks to be a worthwhile project. Our brain is superior to a computer for pattern recognition, especially in low-SNR data. There's definitely a lot of junk in there. I've done 2100 classifications and called almost 1300 signals. Most are easily identifiable as satellite signals or terrestrial RFI, but I can certainly see how a computer would go haywire trying to sort through all this.
____________

The brain is of course a pattern recognizer; but I doubt it could beat a computer at detecting signals embedded in noise. Listen to an old modem using QAM modulation and you will be hard pressed to tell the zero's from the one's at say 56 Kbps., yet, a modem does it quite well with lots of processing as long as the noise and bandwidth are within reason. If you can estimate the form of the signal correctly you can do even better by using advanced techniques--See "The Wiener Filter"

The assumption is that the spectrum of noise is known--cosmic noise spectra of varying types are well known. Auto correlation will bring out a non random signal--Thats a lot of computing since Fourier Transforms require a fair amount of computing as well

Check out Wiener's book on signal processing.

You're comparing auditory recognition with visual recognition. They're very different, and use different parts of the brain. I'm much better with visual than with audio, as I believe most people are. And of course it's also a matter of training and experience.
____________

I am talking visual; as an autocorrelation function would give you a graph. You can take a picture of your voice. If you have ever used an oscilloscope you will know what I mean.

I suspect that a trnasmission from another world may not contain audio and if it did we may not recognize it as audio. Visually we could determine that there was intelligence contained in the signal as opposed to thermal noise.

by looking at a dolphin's clicks we could possibly relate them to certain situations better than if we could actually hear them which we can't entirely since our frequency of hearing cuts off around 20,000 cps.

That's a valid point, some things are just beyond our physiological capability to discern. Computers definitely play an important part filling in those gaps. And of course our brain has capabilities that surpass even our best computers when it comes to fast and efficient signal recognition. I'm also quite familiar with how a lot of noise in data can make a computer system go insane and produce total garbage for results. And this I believe is the entire point of the SETILive. I'm finding some nice signals, barely above background, behind some much more powerful RFI. Would the computer have spotted them? Dunno, but I did.

And at about 30W powered from sugar, it's quite an efficient supercomputer also!