Data processing: artificial intelligence – Neural network – Learning task
Reexamination Certificate
2001-06-25
2004-07-13
Starks, Jr., Wilbert L. (Department: 2121)
Data processing: artificial intelligence
Neural network
Learning task
C706S015000, C706S016000, C704S205000
Reexamination Certificate
active
06763339
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to signal processing, and more particularly, to extracting signals from noisy data.
BACKGROUND OF THE INVENTION
Much research effort has been expended on sensors and instrumentation such as those built with the latest Micro-Electro-Mechanical (MEMs) technology. However, intelligent signal processing, such as advanced signal and pattern recognition algorithms, has not been rigorously pursued. It is important to note that with intelligent processing, information from sensors may be made easier to interpret, transmit, conceal, and store, than if it was a large amount of raw data. Sensors or instrumentation deployed in real-world settings usually produce signals corrupted by various types of noise. As a result, noise removal is a prerequisite for accurate data interpretation analysis and effective storage/transmission. Noisy data would make data compression much harder and thus affect the issues of storage and transmission. Typical techniques such as the Fast Fourier Transform (FFT) have demonstrated limited capability when the noise amplitude is large and/or strongly overlapping of the signal's frequency spectrum. More advanced techniques include wavelets. Wavelets split a signal into overlapping subbands. Wavelets show promise for their ability to handle simultaneous localization of frequency and position, and thus offer more flexibility than the FFT because truncating certain transform coefficients has more a local effect than a global effect as in the case of FFT processing; but even there, if the signal-to-noise ratio is extremely large, even wavelet technology will fail.
SUMMARY OF THE INVENTION
Aspects of the present invention include a method comprising: receiving a signal corrupted with noise; decomposing the signal using a wavelet transform; re-synthesizing the decomposed signal; and inputting the re-synthesized signal into a neutral network to filter out the noise from the signal and recover an uncorrupted signal.
Aspects of the present invention further include a system comprising: a wavelet transformer capable of decomposing a signal and re-synthesizing the signal to eliminate the high frequency noise and most of the low frequency interference; and a neural network operatively coupled to said wave transformer and capable of filtering out noise from the signal and outputting an uncorrupted signal.
REFERENCES:
patent: 5388182 (1995-02-01), Benedetto et al.
patent: 5561431 (1996-10-01), Peele et al.
patent: 5598481 (1997-01-01), Nishikawa et al.
patent: 5619998 (1997-04-01), Abdel-Malek et al.
patent: 5673332 (1997-09-01), Nishikawa et al.
patent: 5785653 (1998-07-01), Kiyuna et al.
patent: 6182018 (2001-01-01), Tran et al.
patent: 6208951 (2001-03-01), Kumar et al.
patent: 6421467 (2002-07-01), Mitra
patent: 6650779 (2003-11-01), Vachtesvanos et al.
Fu Chi Yung
Petrich Loren I.
Lee Ann M.
Scott Eddie E.
Starks, Jr. Wilbert L.
Thompson Alan H.
LandOfFree
Biologically-based signal processing system applied to noise... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Biologically-based signal processing system applied to noise..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Biologically-based signal processing system applied to noise... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3229400