IP Library › Granted Patent US 12,632,720
Granted Patent B2
US 12,632,720 · App. 17/075,797 · Granted May 19, 2026

Evaluating system generated historical transaction timeline images

Inventors: Yi-Hui Ma (Mechanicsburg, PA); Eugene Irving Kelton (Wake Forest, NC); Shuyan Lu (Cary, NC)
Assignee: International Business Machines Corporation
G06N3/08G06F16/2379
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,632,720
App. No.
17/075,797
Filed
Oct 21, 2020
Granted
May 19, 2026
Kind
B2
Art Unit
2146
USPC
706/25
Abstract

In an approach for evaluating system generated historical transaction timeline images for a computer vision deep learning process, a processor provides one or more chart evaluation factors for evaluating historical timeline images for a deep learning of patterns based on the historical timeline images. A processor generates a quantitative metric based on the one or more chart evaluation factors using a quantitative technique. A processor determines a score for an input timeline image based on the quantitative metric. A processor filters input space based on the score. A processor, in response to receiving a feedback, adjusts a chart setting based on the one or more chart evaluation factors.

Claims (35)

1 . A computer-implemented method, comprising:

providing, by one or more processors, one or more chart evaluation factors corresponding to a plurality of historical timeline images;

generating, by the one or more processors, a quantitative metric for the plurality of historical timeline images, based on the one or more chart evaluation factors, using one or more quantitative techniques, wherein the one or more quantitative techniques comprise at least one selected from the group consisting of analytical hierarchical process, fuzzy logic, and statistical distribution;

determining, by the one or more processors, a score for each of one or more input timeline images, based on the quantitative metric;

selecting, by the one or more processors, automatically an input timeline image from the one or more input timeline images for training a cognitive system in a deep event learning process, based on the score of the one or more input timeline images, wherein the cognitive system uses the selected input timeline image to detect patterns in the plurality of historical timeline images;

in response to receiving a feedback based on insights from a user, adjusting, by the one or more processors, a chart setting for evaluating the one or more input timeline images, wherein the chart setting comprises the one or more chart evaluation factors; and

applying, by the one or more processors, the adjusted chart setting for selecting further input timeline images, from the one or more input timeline images, for training the cognitive system in the deep event learning process,

wherein determining the score comprises determining a highest score among scores in the quantitative metric associated with the one or more chart evaluation factors.

2 . The computer-implemented method of claim 1 , wherein providing the one or more chart evaluation factors includes determining the one or more chart evaluation factors based on at least one of a chart type, scaling, color, marker shape, and data density.

3 . The computer-implemented method of claim 1 , wherein determining the score includes aggregating scores in the quantitative metric associated with the one or more chart evaluation factors.

4 . A computer program product, comprising:

one or more computer readable storage media; and

program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising:

program instructions to provide one or more chart evaluation factors corresponding to a plurality of historical timeline images;

program instructions to generate a quantitative metric for the plurality of historical timeline images, based on the one or more chart evaluation factors, using one or more quantitative techniques, wherein the one or more quantitative techniques comprise at least one selected from the group consisting of analytical hierarchical process, fuzzy logic, and statistical distribution;

program instructions to determine a score for each of one or more input timeline images, based on the quantitative metric;

program instructions to select automatically an input timeline image from the one or more input timeline images for training a cognitive system in a deep event learning process, based on the score of the one or more input timeline images, wherein the cognitive system uses the selected input timeline image to detect patterns in the plurality of historical timeline images;

program instructions to, in response to receiving a feedback, adjust a chart setting to evaluate the one or more input timeline images, wherein the chart setting comprises the one or more chart evaluation factors; and

program instructions to apply the adjusted chart setting to select further input timeline images, from the one or more input timeline images, for training the cognitive system in the deep event learning process,

wherein the program instructions to determine the score include program instructions to determine a highest score among scores in the quantitative metric associated with the one or more chart evaluation factors.

5 . The computer program product of claim 4 , wherein the program instructions to provide the one or more chart evaluation factors include program instructions to determine the one or more chart evaluation factors based on at least one of chart type, scaling, color, marker shape and data density.

6 . The computer program product of claim 4 , wherein the program instructions to determine the score include program instructions to aggregate scores in the quantitative metric associated with the one or more chart evaluation factors.

7 . A computer system, comprising:

one or more computer processors;

one or more computer readable storage media; and

program instructions stored on the one or more computer readable storage media for execution by at least one of the one or more computer processors, the program instructions comprising:

program instructions to provide one or more chart evaluation factors corresponding to a plurality of historical timeline images;

program instructions to generate a quantitative metric for the plurality of historical timeline images, based on the one or more chart evaluation factors, using one or more quantitative techniques, wherein the one or more quantitative techniques comprise at least one selected from the group consisting of analytical hierarchical process, fuzzy logic, and statistical distribution;

program instructions to determine a score for each of one or more input timeline images, based on the quantitative metric;

program instructions to select automatically an input timeline image from the one or more input timeline images for training a cognitive system in a deep event learning process, based on the score of the one or more input timeline images, wherein the cognitive system uses the selected input timeline image to detect patterns in the plurality of historical timeline images;

program instructions to, in response to receiving a feedback, adjust a chart setting to evaluate the one or more input timeline images, wherein the chart setting comprises the one or more chart evaluation factors; and

program instructions to apply the adjusted chart setting to select further input timeline images, from the one or more input timeline images, for training the cognitive system in the deep event learning process,

wherein the program instructions to determine the score include program instructions to determine a highest score among scores in the quantitative metric associated with the one or more chart evaluation factors.

8 . The computer system of claim 7 , wherein the program instructions to provide the one or more chart evaluation factors include program instructions to determine the one or more chart evaluation factors based on at least one of a chart type, scaling, color, marker shape, and data density.

9 . The computer system of claim 7 , wherein the program instructions to determine the score include program instructions to aggregate scores in the quantitative metric associated with the one or more chart evaluation factors.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Oct 21, 2020
From: MA, YI-HUI; KELTON, EUGENE IRVING; LU, SHUYAN
To: INTERNATIONAL BUSINESS MACHINES CORPORATION
Reel/Frame 054122/0496 →
Continuity (1)
Related Publication 20220121923A1 · Apr 21, 2022
References Cited (39)
US 6742124B1 · Kilpatrick et al. · 2004 [cited by applicant]
US 6996551B2 · Hellerstein et al. · 2006 [cited by applicant]
US 9235752B2 · Eaton et al. · 2016 [cited by applicant]
US 10375098B2 · Oliner et al. · 2019 [cited by applicant]
US 10642898B1 · Pratt · 2020 [cited by examiner]
US 10732902B1 · Lazier · 2020 [cited by examiner]
US 11551081B2 · Krishnan · 2023 [cited by examiner]
US 20030153299A1 · Perfit et al. · 2003 [cited by applicant]
US 20050256813A1 · Bahbouh · 2005 [cited by examiner]
US 20110044492A1 · Cobb et al. · 2011 [cited by applicant]
US 20110071969A1 · Doctor · 2011 [cited by examiner]
US 20140081713A1 · Gifford · 2014 [cited by examiner]
US 20140234814A1 · Krosky · 2014 [cited by examiner]
US 20150339689A1 · Chen · 2015 [cited by examiner]
US 20160259896A1 · Yang · 2016 [cited by applicant]
US 20170140384A1 · Zoldi et al. · 2017 [cited by applicant]
US 20170330075A1 · Tuysuzoglu et al. · 2017 [cited by applicant]
US 20180114169A1 · Wiig · 2018 [cited by examiner]
US 20180276541A1 · Studnitzer · 2018 [cited by examiner]
US 20180336506A1 · Malhotra · 2018 [cited by examiner]
US 20180365065A1 · Guttmann · 2018 [cited by examiner]
US 20190034828A1 · Cagadas · 2019 [cited by examiner]
US 20190037081A1 · Rao · 2019 [cited by examiner]
US 20190259033A1 · Reddy · 2019 [cited by examiner]
US 20200097851A1 · Alvarez · 2020 [cited by examiner]
US 20200250472A1 · Abhyankar · 2020 [cited by examiner]
US 20220067752A1 · Fang · 2022 [cited by examiner]
“Basic classification: Classify images of clothing”, TensorFlow, Printed on Oct. 19, 2020, 14 pages, < https://www.tensorflow.org/tutorials/keras/classification>. [cited by applicant]
Adhianto et al. “HPCToolkit: Tools for performance analysis of optimized parallel programs.” Concurrency and Computation: Practice and Experience 22.6, 2010, pp. 685-701. [cited by applicant]
Aguilar et al., “Visual MPI Performance Analysis using Event Flow Graphs” Procedia Computer Science, vol. 51, 2015, pp. 1353-1362. [cited by applicant]
Andrearczyk, Vincent, “Deep learning for texture and dynamic texture analysis”, Dissertation for the degree of Doctor of Philosophy. Dublin City University, Sep. 2017, 189 pages. [cited by applicant]
Chen et al., “Machine learning techniques for anti-money laundering (AML) solutions in suspicious transaction detection: a review”, Knowledge and Information Systems, Feb. 10, 2018, 41 pages, <https://doi.org/10.1007/s1… [cited by applicant]
Kelton et al., “Classifying Behavior Through System-Generated Timelines and Deep Learning”, U.S. Appl. No. 16/867,355, filed May 5, 2020, 32 pages. [cited by applicant]
Ma et al., “Cognitive Methodology for Sequence of Events Patterns in Fraud Detection”, U.S. Appl. No. 15/971,943, filed May 4, 2018, 33 pages. [cited by applicant]
Sim et al., “Is Deep Learning for Image Recognition Applicable to Stock Market Prediction?”, Hindawi, Complexity, vol. 2019, Article ID 4324878, 11 pages, <https://doi.org/10.1155/2019/4324878>. [cited by applicant]
Soderlind Simon, “Application Performance Evaluation using Deep Learning” MS thesis. Universitat Politècnica de Catalunya, 2017, 69 pages. [cited by applicant]
Suzumura et al., “Towards Federated Graph Learning for Collaborative Financial Crimes Detection”, Cornell University arXiv, preprint arXiv:1909.12946, Oct. 2, 2019, 10 pages. [cited by applicant]
Wang et al. “Deep learning for smart manufacturing: Methods and applications”, Journal of Manufacturing Systems, 48, 2018, pp. 144-156. [cited by applicant]
Wolf et al. “Efficient Pattern Search in Large Traces Through Successive Refinement” European Conference on Parallel Processing. Springer, Berlin, Heidelberg, 2004, 8 pages. [cited by applicant]