Method for analyzing activity in a signal

• US 5,617,513 A
• Filed: 01/04/1995
• Issued: 04/01/1997
• Est. Priority Date: 03/06/1992
• Status: Expired due to Term

• Current Assignee: Noiseout, Inc.
• Original Assignee: Bonnie S. Schnitta
• Inventors: Schnitta, Bonnie S.
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/368,686
• Time in Patent Office: 818 Days
• Field of Search: 395/2, 395/2.1, 395/2.3, 395/27, 395/25, 395/2.45, 395/2.55, 395/20-24, 395/50-51, 395/61-64, 395/76, 395/26, 381/51-53, 381/42, 364/572, 364/152, 364/487, 382/155-159, 382/209
• US Class Current: 706/14
• CPC Class Codes: G01S 7/292 Extracting wanted echo-signals
• CPC Class Codes: G01S 7/417 involving the use of neural networks
• CPC Class Codes: G01V 1/01 Measuring or predicting earthquakes
• CPC Class Codes: G01V 1/288 Event detection in seismic signals, e.g. microseismics G01V1/36 takes precedence
• CPC Class Codes: G06F 2218/04 Denoising
• CPC Class Codes: H03H 21/0012 Digital adaptive filters
• CPC Class Codes: H03H 2222/04 using neural networks
• CPC Class Codes: Y10S 706/925 Business
• CPC Class Codes: Y10S 706/929 Geological, e.g. seismology

Abstract

A method for analyzing a signal. The signal is passed through a noise filter to generate a filtered signal. The filtered signal is passed through an adaptive pattern recognition unit. At the adaptive pattern recognition unit, a pattern that is a function of the filtered signal is predicted. A mathematical deviation in the filtered signal is detected by comparing the filtered signal with the predicted pattern. The detected mathematical deviation is flagged. The filtered signal is passed through at least two pattern recognition weighing units. At each pattern recognition weighing unit, the filtered signal at about the detected mathematical deviation is matched with a collection of stored patterns. A weight is assigned to each stored pattern. Each stored pattern and the assigned weight of the stored pattern are input to an expert system ranking unit. At the expert system ranking unit, one of the stored patterns is selected by applying a predefined set of generalized rules. Additionally, the method may determine if activity in a signal is individual event activity. A moment of chaos in the signal may also be determined.

Claims

1. A method for determining if activity in a signal is market activity comprising the steps of:

receiving an unfiltered signal at a noise filter unit;

passing the unfiltered signal through a non-adaptive noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a first pattern that is a function of the filtered signal;

determining if there is a first deviation in the filtered signal by comparing the filtered signal with the predicted first pattern;

passing the unfiltered signal through the adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a second pattern that is a function of the unfiltered signal;

determining if there is a second deviation in the unfiltered signal by comparing the unfiltered signal with the predicted second pattern; and

if there exists a second deviation in the unfiltered signal and no first deviation in the filtered signal, outputting an indication of market activity.

2. The method of claim 1 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

3. The method of claim 1 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

4. The method of claim 1 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

5. The method of claim 1 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

6. A method for determining if activity in a signal is individual event activity comprising the steps of:

receiving an unfiltered signal at a noise filter unit;

passing the unfiltered signal through a non-adaptive noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a first pattern that is a function of the filtered signal;

determining if there is a first deviation in the filtered signal by comparing the filtered signal with the predicted first pattern;

passing the unfiltered signal through the adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a second pattern that is a function of the unfiltered signal;

determining if there is a second deviation in the unfiltered signal by comparing the unfiltered signal with the predicted second pattern; and

if it is determined that there is a first deviation in the filtered signal and no second deviation in the unfiltered signal, outputting an indication of individual event activity.

7. The method of claim 6 wherein the unfiltered signal represents stock prices for a plurality of stocks and the individual event activity represents event activity for one stock.

8. A method for determining if activity in a signal is market activity comprising the steps of:

receiving an unfiltered signal at a noise filter unit;

passing the unfiltered signal through an adaptive noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a first pattern that is a function of the filtered signal;

determining if there is a first deviation in the filtered signal by comparing the filtered signal with the predicted first pattern;

passing the unfiltered signal through the adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a second pattern that is a function of the unfiltered signal;

determining if there is a second deviation in the unfiltered signal by comparing the unfiltered signal with the predicted second pattern; and

if there exists a second deviation in the unfiltered signal and no first deviation in the filtered signal, outputting an indication of market activity.

9. The method of claim 8 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

10. The method of claim 8 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

11. The method of claim 8 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

12. The method of claim 8 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

13. A method for determining if activity in a signal is individual event activity comprising the steps of:

receiving an unfiltered signal at a noise filter unit;

passing the unfiltered signal through an adaptive noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a first pattern that is a function of the filtered signal;

determining if there is a first deviation in the filtered signal by comparing the filtered signal with the predicted first pattern;

passing the unfiltered signal through the adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a second pattern that is a function of the unfiltered signal;

determining if there is a second deviation in the unfiltered signal by comparing the unfiltered signal with the predicted second pattern; and

if it is determined that there is a first deviation in the filtered signal and no second deviation in the unfiltered signal, outputting an indication of individual event activity.

14. The method of claim 13 wherein the unfiltered signal represents stock prices for a plurality of stocks and the individual event activity represents event activity for one stock.

15. The method of claim 13 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

16. The method of claim 13 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through a self adaptive and variant error ("SAVE") filter.

17. The method of claim 13 wherein the step of passing the unfiltered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

18. The method of claim 13 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the unfiltered signal through an adaptive TRAPS noise filter.

19. A method for determining a moment of chaos in a signal comprising the steps of:

passing the signal through a noise filter to generate a filtered data signal;

passing the filtered data signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a pattern that is a function of the filtered data signal;

detecting a mathematical deviation in the filtered data signal by comparing the filtered data signal with the predicted pattern; and

flagging the detected mathematical deviation as the moment of chaos.

20. The method of claim 19 wherein the noise filter is a non-adaptive noise filter.

21. The method of claim 19 wherein the noise filter is an adaptive TRAPS noise filter.

22. The method of claim 19 wherein the step of passing the filtered data signal through the adaptive pattern recognition unit further comprises passing the filtered data signal through a self adaptive and variant error ("SAVE") filter.

23. The method of claim 19 wherein the step of passing the filtered data signal through the adaptive pattern recognition unit further comprises passing the filtered data signal through an adaptive TRAPS noise filter.

24. A method for analyzing a signal comprising the steps of:

providing an expert system ranking unit;

passing the signal through a noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a pattern that is a function of the filtered signal;

detecting a mathematical deviation in the filtered signal by comparing the filtered signal with the predicted pattern;

flagging the detected mathematical deviation;

passing the filtered signal through at least two pattern recognition weighing units, the at least two pattern recognition weighing units comprising a first pattern recognition weighing unit being an entopic differential weighing unit and a second pattern recognition weighing unit being an adaptive pattern recognition weighing unit;

at each pattern recognition weighing unit, matching the filtered signal at about the detected mathematical deviation with a collection of stored patterns and assigning a weight to each stored pattern;

inputting each stored pattern and the assigned weight of the stored pattern to the expert system ranking unit; and

at the expert system ranking unit, selecting one of the stored patterns by applying a predefined set of generalized rules.

25. The method of claim 24 wherein the noise filter is a non-adaptive noise filter.

26. The method of claim 24 wherein the noise filter is an adaptive TRAPS noise filter.

27. The method of claim 24 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the filtered signal through a self adaptive and variant error ("SAVE") filter.

28. The method of claim 24 wherein the step of passing the filtered signal through the adaptive pattern recognition unit further comprises passing the filtered signal through an adaptive TRAPS noise filter.

29. A method for analyzing a signal comprising the steps of:

receiving an unfiltered signal at a noise filter unit;

passing the unfiltered signal through a non-adaptive noise filter to generate a first filtered data signal;

passing the unfiltered signal through an adaptive TRAPS filter to generate a second filtered data signal;

passing the unfiltered signal, the first filtered data signal and the second filtered data signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a first pattern that is a function of the first filtered data signal;

at the adaptive pattern recognition unit, predicting a second pattern that is a function of the second filtered data signal;

at the adaptive pattern recognition unit, predicting a third pattern that is a function of the unfiltered signal;

detecting a first mathematical deviation in the first filtered data signal by comparing the first filtered data signal with the first predicted pattern;

detecting a second mathematical deviation in the second filtered data signal by comparing the second filtered data signal with the second predicted pattern;

detecting a third mathematical deviation in the unfiltered signal by comparing the unfiltered signal with the third predicted pattern;

flagging the first mathematical deviation, the second mathematical derivation and the third mathematical deviation;

passing the first filtered data signal, the second filtered data signal and the unfiltered signal through at least two pattern recognition weighing units;

at each pattern recognition weighing unit, matching the first filtered data signal at about the detected first mathematical deviation with a collection of stored patterns and assigning a weight to each stored pattern;

at each pattern recognition weighing unit, matching the second filtered data signal at about the detected second mathematical deviation with the collection of stored patterns and assigning a weight to each stored pattern;

at each pattern recognition weighing unit, matching the unfiltered signal at about the detected third mathematical deviation with the collection of stored patterns and assigning a weight to each stored pattern;

inputting each stored pattern and the assigned weights of the stored pattern to an expert system ranking unit; and

at the expert system ranking unit, selecting one of the stored patterns by applying a predefined set of generalized rules.

30. The method of claim 29 where the at least two pattern recognition weighing units comprise a first pattern recognition weighing unit being an entopic differential weighing unit and a second pattern recognition weighing unit being an adaptive pattern recognition weighing unit.

31. The method of claim 29 further comprising the step of, at the expert system ranking unit, ordering the stored patterns by applying the predefined set of generalized rules.

32. The method of claim 29 wherein the step of matching the first filtered data signal at about the detected first mathematical deviation with the collection of stored patterns further comprises the step of matching the first filtered data signal at about the detected first mathematical deviation with the collection of stored patterns using self similarity matching.

33. The method of claim 29 wherein the step of matching the second filtered data signal at about the detected second mathematical deviation with a collection of stored patterns further comprises the step of matching the second filtered data signal at about the detected second mathematical deviation with the collection of stored patterns using self similarity matching.

34. The method of claim 29 wherein the step of matching the unfiltered data signal at about the detected third mathematical deviation with a collection of stored patterns further comprises the step of matching the unfiltered data signal at about the detected third mathematical deviation with the collection of stored patterns using self similarity matching.

35. A method for analyzing a signal representing stock prices comprising the steps of:

receiving, at a noise filter unit, a signal representative of the price of stocks over time;

passing the signal through a noise filter to generate a filtered signal;

passing the filtered signal through an adaptive pattern recognition unit;

at the adaptive pattern recognition unit, predicting a pattern that is a function of the filtered signal;

detecting a mathematical deviation in the filtered signal by comparing the filtered signal with the predicted pattern; and

flagging the detected mathematical deviation;

passing the filtered signals through at a pattern recognition weighing unit;

at the pattern recognition weighing unit, matching the filtered signal at about the detected mathematical deviation with a collection of stored patterns; and

at the pattern recognition unit, determining, for a particular stock, whether the stock is in an uptrend or in a downtrend.