US 10,755,184 C2 (12,745th)
Predictive machine learning models
Brian Holligan, San Francisco, CA (US); and Andy Mahdavi, San Francisco, CA (US)
Filed by States Title, Inc., San Francisco, CA (US)
Assigned to STATES TITLE, LLC, San Francisco, CA (US)
Reexamination Request No. 90/019,210, May 17, 2023.
Reexamination Certificate for Patent 10,755,184, issued Aug. 25, 2020, Appl. No. 16/716,289, Dec. 16, 2019.
Reexamination Certificate C1 10,755,184, issued Jul. 1, 2022.
Application 90/019,210 is a continuation of application No. 16/505,259, filed on Jul. 8, 2019, granted, now 10,510,009.
Ex Parte Reexamination Certificate issued on Oct. 21, 2024.
Int. Cl. G06N 5/04 (2023.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01); G06Q 40/03 (2023.01)
CPC G06N 5/04 (2013.01) [G06N 7/01 (2023.01); G06N 20/00 (2019.01); G06Q 40/03 (2023.01)]
OG exemplary drawing
AS A RESULT OF REEXAMINATION, IT HAS BEEN DETERMINED THAT:
Claims 1, 5, 7, 11, 13, 17, 20-23, 29 and 34 are determined to be patentable as amended.
Claims 2-4, 6, 8-10, 12, 14-16, 18-19, 24-28, 30-33 and 35-40, dependent on an amended claim, are determined to be patentable.
New claim 41 is added and determined to be patentable.
1. A method comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
using a machine learning model to generate a prediction from the obtained plurality of data points, [ comprising:
identifying, by the machine learning model, an indication of a potentially open mortgage based on a record of a subordinate mortgage among the plurality of data points without an accompanying primary mortgage;]
the prediction indicating a likelihood that the real property will satisfy a particular parameter , wherein [ and ] the particular parameter being predicted is a likelihood that one or more potentially open mortgages attached to the specified parcel of real property are actually open, and
[ wherein generating the prediction comprises:
for each potentially open mortgage attached to the specified parcel of real property, generating a respective probability value that the mortgage is actually open; and
comparing each respective probability value to a corresponding threshold value to determine a likelihood that the mortgage is actually open:]
wherein the machine learning model is trained using a training set comprising a collection of data points associated with [ and labels for ] a set of real property parcels distinct from the specified parcel of real property, wherein each real property parcel of the training set includes information about each mortgage attached to the parcel; and
based on the prediction, classifying the specified parcel of real property according to a determination of whether the predicted value of the parameter satisfies a [ the corresponding ] threshold value.
5. The method of claim 4, wherein the data points include an identification of potentially open [ first ] mortgages both directly identified from parcel data or indirectly identified from the parcel data.
7. A system comprising:
one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
using a machine learning model to generate a prediction from the obtained plurality of data points, comprising:
identifying, by the machine learning model, an indication of a potentially open [ first ] mortgage [ that (i) is not included in the obtained plurality of data points or otherwise known to the machine learning model, and (ii) may or may not exist, ] based on a record of a subordinate mortgage among the plurality of data points without an accompanying primary mortgage;
the prediction indicating a likelihood that the real property will satisfy a particular parameter and the particular parameter being predicted is a likelihood that one or more potentially open [ first ] mortgages attached to the specified parcel of real property are actually open,
wherein generating the prediction comprises:
for each potentially open [ first ] mortgage attached to the specified parcel of real property, generating a respective probability value that the mortgage is actually open; and
comparing each respective probability value to a corresponding threshold value to determine a likelihood that the mortgage is actually open;
wherein the machine learning model is trained using a training set comprising a collection of data points and labels for a set of real property parcels distinct from the specified parcel of real property, wherein each real property parcel of the training set includes information about each mortgage attached to the parcel; and
based on the prediction, classifying the specified parcel of real property according to a determination of whether the predicted value of the parameter satisfies the corresponding threshold value.
11. The system of claim 10, wherein the data points include an identification of potentially open [ first ] mortgages both directly identified from parcel data or indirectly identified from the parcel data.
13. One or more non-transitory computer-readable storage media encoded with instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
using a machine learning model to generate a prediction from the obtained plurality of data points, comprising:
identifying, by the machine learning model, an indication of a potentially open [ first ] mortgage [ that (i) is not included in the obtained plurality of data points or otherwise known to the machine learning model, and (ii) may or may not exist, ] based on a record of a subordinate mortgage among the plurality of data points without an accompanying primary mortgage;
the prediction indicating a likelihood that the real property will satisfy a particular parameter and the particular parameter being predicted is a likelihood that one or more potentially open [ first ] mortgages attached to the specified parcel of real property are actually open,
wherein generating the prediction comprises:
for each potentially open [ first ] mortgage attached to the specified parcel of real property, generating a respective probability value that the mortgage is actually open; and
comparing each respective probability value to a corresponding threshold value to determine a likelihood that the mortgage is actually open;
wherein the machine learning model is trained using a training set comprising a collection of data points and labels for a set of real property parcels distinct from the specified parcel of real property, wherein each real property parcel of the training set includes information about each mortgage attached to the parcel; and
based on the prediction, classifying the specified parcel of real property according to a determination of whether the predicted value of the parameter satisfies the corresponding threshold value.
17. The one or more non-transitory computer-readable storage media of claim 16, wherein the data points include an identification of potentially open [ first ] mortgages both directly identified from parcel data or indirectly identified from the parcel data.
20. The method of claim 19, wherein:
the particular parameter being predicted corresponds to a score among a plurality of respective scores; and
each respective score of the plurality of respective scores is generated for a potentially open [ first ] mortgage.
21. The method of claim 19, further comprising:
generating a respective score for each of the potentially open [ first ] mortgages attached to the specified parcel of real property; and
comparing each respective score to a corresponding threshold value to determine a likelihood that the mortgage is actually open.
22. The method of claim 21, further comprising:
determining that a respective score for a potentially open [ first ] mortgage exceeds the threshold value for indicating whether the mortgage is actually open; and
predicting the particular parameter based on the determination that the respective score for the potentially open [ first ] mortgage exceeds the threshold value for indicating whether the mortgage is actually open.
23. The method of claim 22, further comprising:
predicting that the potentially open [ first ] mortgage is actually open based on the respective score for the potentially open [ first ] mortgage exceeding the threshold value for indicating whether the mortgage is actually open.
29. The method of claim 28, wherein determining a second threshold value comprises:
determining a second threshold value that requires a lower likelihood that a potentially open mortgage [ the first mortgage ] is actually open.
34. The method of claim 1, further comprising:
learning, by the machine-learning model, a plurality of factors that increase or decrease a likelihood of a potentially open [ first ] mortgage actually being open; and
generating, using the machine-learning model, a prediction that a given mortgage identified for the parcel of real property is still open based on the plurality of factors learned by the machine-learning model.
[ 41. A method comprising:
obtaining, from one or more sources, a plurality of data points associated with a specified parcel of real property;
using a machine learning model to generate a prediction from the obtained plurality of data points, comprising:
identifying, by the machine learning model, an indication of a potentially open mortgage based on a record of a subordinate mortgage among the plurality of data points without an accompanying primary mortgage;
the prediction indicating a likelihood that the real property will satisfy a particular parameter and the particular parameter being predicted is a likelihood that one or more potentially open mortgages attached to the specified parcel of real property are actually open,
wherein generating the prediction comprises:
for each potentially open mortgage attached to the specified parcel of real property, generating a respective probability value that the mortgage is actually open; and
comparing each respective probability value to a corresponding threshold value to determine a likelihood that the mortgage is actually open;
wherein the machine learning model is trained using a training set comprising a collection of data points and labels for a set of real property parcels distinct from the specified parcel of real property, wherein each real property parcel of the training set includes information about each mortgage attached to the parcel;
based on the prediction, classifying the specified parcel of real property according to a determination of whether the predicted value of the parameter satisfies the corresponding threshold value; and
dynamically determining the threshold value based on actual performance of the machine learning model.]