Automatic tuning of value-series analysis tasks based on visual feedback

• US 9,355,479 B2
• Filed: 11/26/2012
• Issued: 05/31/2016
• Est. Priority Date: 11/15/2012
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Pompey, Pascal, Verscheure, Olivier, Wurst, Michael
• Primary Examiner(s): Faulk, Devona
• Assistant Examiner(s): Sun, Hai Tao
• Application Number:
  US13/685,301
• Publication Number:
  US 20140146078A1
• Time in Patent Office: 1,282 Days
• Field of Search: 345/629
• US Class Current: 1/1
• CPC Class Codes: G06F 18/2178 based on feedback of a supervisor
• CPC Class Codes: G06F 2218/08 Feature extraction
• CPC Class Codes: G06F 3/01 Input arrangements or combined input and output arrangements for interaction between user and computer G06F3/16 takes precedence
• CPC Class Codes: G06T 11/00 2D [Two Dimensional] image generation
• CPC Class Codes: G06T 11/206 Drawing of charts or graphs

Abstract

A method for selecting an analysis procedure for a value series, including displaying a value series on a computer display monitor, receiving one or more sequences of user provided annotations, where the annotations overlay at least a sub-interval of the value series on the computer display monitor, using the sequences of user provided annotations to select an optimal value series analysis method from a set of value series analysis methods, where selecting an optimal value series analysis method includes determining parameter values for the optimal value series analysis method, and presenting the selected optimal value series analysis method and parameters, and the optimal reconstruction of the annotation sequences to the user.

Claims

1. A non-transitory program storage device readable by a computer, tangibly embodying a program of instructions executed by the computer to perform the method steps for selecting an analysis procedure for a time series, the method comprising the steps of:

displaying a time series on a computer display monitor;

receiving one or more sequences of user provided annotations, wherein said annotations overlay at least a sub-interval of said time series on said computer display monitor;

using said sequences of user provided annotations to select an optimal time series analysis method from a set of time series analysis methods, wherein selecting an optimal time series analysis method includes determining parameter values for said optimal time series analysis method;

segmenting said time series into segments based on a sequence of user provided annotations; and

presenting said selected optimal time series analysis method and parameters, and said optimal reconstruction of said annotation sequences to the user,

wherein selecting an optimal time series analysis method that optimally reconstructs said sequences of user provided annotations comprises, for segments that are not annotated by the user, identifying a similar annotated segment, and determining an optimal filter that minimizes an error between time series points in the unannotated segment and the annotated segment, wherein a similar annotated segment is identified based on an error distribution between the unannotated segment before and after being filtered by said optimal filter.

2. The computer readable program storage device of claim 1, the method further comprising, after presenting said selected optimal time series analysis method to the user, receiving additional sequences of user provided annotations from the user, and selecting another optimal time series analysis method from a set of time series analysis methods that optimally reconstructs said sequences of user provided annotations.

3. The computer readable program storage device of claim 1, the method further comprising displaying a plurality of time series on said computer display monitor, wherein at least one sequence of user provided annotations connects points in different time series.

4. The computer readable program storage device of claim 1, wherein selecting an optimal time series analysis method that optimally reconstructs said sequences of user provided annotations comprises, for segments that are annotated by the user, determining an optimal filter that minimizes an error between time series points in the segment and the user annotations.

5. The computer readable program storage device of claim 1, wherein a similar annotated segment is identified based on a variance and frequency distribution of the unannotated segment and the annotated segment.

6. The computer readable program storage device of claim 1, the method further comprising clustering all segments for which no similar annotated segment is identified into one or more clusters based on general characteristics of each segment, and presenting a representative of each cluster to the user for annotation.

7. The computer readable program storage device of claim 1, the method further comprising extracting features that maximally correlate different sequences of user provided annotations.

8. The computer readable program storage device of claim 7, wherein said features include one or more of a lag value and a window width.

9. The computer readable program storage device of claim 1, wherein said optimal time series analysis method optimally reconstructs the user provided annotation sequences.