Server-based method of using a trained deep learning model and ground truth labels
A server-implemented method for determining, at a server, using a deep learning model, a cluster of device identifiers associated with computing devices having characteristics that are related to a match set of computing devices based on locations data streams and ground truth labels.
1. A server-implemented method for determining, at a server, using a trained deep learning model, a cluster of device identifiers associated with computing devices having truth labels based on locations data streams obtained from the computing devices, the method comprising:
obtaining, at the server, a first dataset of a first set of entities, each entity of the first data set being a user associated with a client computing device, wherein the first dataset comprises entity identifiers, locations, cookies, and hashed email addresses of the entities;
obtaining, at the server, a second dataset of a second set of entities, each entity associated with a client computing device in a geographical area, wherein the second dataset comprises behavioral attributes of the second set of computing devices, mobile entity identifiers, hashed email addresses of the entities, and the locations data streams, the locations data streams comprising characteristics, connection characteristics, and user agent strings, the user agent strings comprising a plurality of tokens comprising a request from the second set of computing devices;
generating, at the server, a plurality of match sets of entities, each match set comprising a first client computing device associated with an entity of the first dataset and a second client computing device associated with an entity of the second dataset;
determining, at the server, for each of the plurality of match sets, a ground truth label, each ground truth label determined using at least high confidence entities of the first dataset and the second dataset;
providing, at the server, to the trained deep learning model, the ground truth labels;
using the trained deep learning model to generate the cluster of device identifiers, the cluster of device identifiers generated based on a level of similarity between behavioral attributes of the match set and behavioral attributes of the second set of computing devices, each device identifier of the cluster being associated with one of the computing devices associated with the entities of the second dataset; and
obtaining, at the server, from the trained deep learning model, the cluster, the cluster comprising a plurality of device identifiers and a behavioral attribute for each device identifier.
2. The server-implemented method of claim 1 , wherein the trained deep learning model is trained using the ground truth labels and at least one custom feature specific to the client.
3. The server-implemented method of claim 1 , further comprising: determining, using the trained deep learning model and a one-class classification method on the second dataset, the cluster of the device identifiers associated with the computing devices having the characteristics that are similar to the match set from the first dataset, from the second dataset.
4. The server-implemented method of claim 1 , wherein at least one characteristic of the first computing device is mutually exclusive from at least one characteristic of the second computing device.
5. The server-implemented method of claim 4 , further comprising: merging a first behavioral attribute and a second behavioral attribute of the matched set using the ground truth labels, wherein the first behavioral attribute and the second behavioral attribute are associated with two mutually exclusive classes of behavior.
6. The server-implemented method of claim 1 , further comprising: determining, using the trained deep learning model and a classification method, the cluster of the device identifiers having multiple overlapping characteristics of behavior to the confident computing devices from the first dataset, from the second dataset, wherein the device identifiers having the multiple overlapping characteristics of behavior are obtained when a plurality of first behavioral characteristics of the matched set overlap in comparison with a plurality of second behavioral characteristics of the second set.
7. The server-implemented method of claim 1 , further comprising: scoring the matched set of computing devices against a behavioral attribute by: generating a user scoring model based on a function of the behavioral attributes of the matched set of computing entities; and assigning, using the user scoring model, a score for each of the matched set of computing entities against the behavioral attribute.
8. The server-implemented method of claim 1 , further comprising: obtaining weights of a plurality of behavioral attributes from the client; configuring the trained deep learning model based on the weights to obtain a reconfigured deep learning model; and generating a cluster for a matched set of computing devices using the re-configured deep learning model.
9. The server-implemented method of claim 1 , wherein the trained machine learning model employs a classification method, the classification method depending on a level of learning between behavioral attributes of the matched set and behavioral attributes of the second set.
10. A system for determining, at a server, using a trained deep learning model, a cluster of device identifiers associated with computing devices having characteristics based on locations data streams obtained from the computing devices, the system comprising:
a processor; and
a memory that stores a set of instructions, which when executed by the processor, causes it to perform:
obtaining, at the server, a first dataset of a first set of computing entities, each entity of the first data set being a user associated with a client computing device, wherein the first dataset comprises entity identifiers, locations, cookies, and hashed email addresses of the entities;
obtaining, at the server, a second dataset of a second set of entities, each entity associated with a client computing device in a geographical area, wherein the second dataset comprises behavioral attributes of the second set of computing devices, mobile entity identifiers, hashed email addresses of the entities, and the locations data streams, the location data streams comprising characteristics, connection characteristics, and user agent strings, the user agent strings comprising a plurality of tokens comprising a request from the second set of computing devices;
generating, at the server, a plurality of match sets of entities, each match set comprising a first client computing device associated with an entity of the first dataset and a second client computing device associated with an entity of the second dataset;
determining, at the server, for each of the plurality of match sets, a ground truth label, each ground truth label determined using at least high confidence entities of the first dataset and the second dataset;
providing, at the server, to the trained deep learning model, the ground truth labels; using the trained deep learning model to generate the cluster of device identifiers, the cluster of device identifiers generated based on a level of similarity between behavioral attributes of the match set and behavioral attributes of the second set of computing devices, each device identifier of the cluster being associated with one of the computing devices associated with the entities of the second dataset; and
obtaining, at the server, from the trained deep learning model, the cluster, the cluster comprising a plurality of device identifiers and a behavioral attribute for each device identifier.
11. The system of claim 10 , wherein the processor further performs determining, using the trained deep learning model and a one-class classification method on the second dataset, the cluster of the device identifiers associated with the computing devices of the computing devices having the characteristics that are similar to the confident computing devices from the first dataset, from the second dataset.
12. The system of claim 10 , wherein at least one characteristic of the first computing device is mutually exclusive from at least one characteristic of the second computing device.
13. The system of claim 12 , wherein the processor further performs merging a first behavioral attribute and a second behavioral attribute of the matched set using the ground truth labels, wherein the first behavioral attribute and the second behavioral attribute are associated with two mutually exclusive classes of behavior.
14. The system of claim 10 , wherein the processor further performs determining, using the trained deep learning model and a classification method, the cluster of the device identifiers having multiple overlapping characteristics of behavior to the confident computing devices from the first dataset, from the second dataset, wherein the device identifiers having the multiple overlapping characteristics of behavior are obtained when a plurality of first behavioral characteristics of the matched set overlap in comparison with a plurality of second behavioral characteristics of the second set.
15. The system of claim 10 , wherein the processor further performs scoring a matched set of computing devices against a behavioral attribute by: generating a user scoring model based on a function of behavioral attributes of the matched set of computing devices; and assigning, using the user scoring model, a score for each of the matched set of computing devices against the behavioral attribute.
16. The system of claim 10 , wherein the processor further performs: obtaining weights of a plurality of behavioral attributes from the client; configuring the trained deep learning model based on the weights to obtain a reconfigured deep learning model; and generating a cluster for a matched set of computing devices using the re-configured deep learning model.
17. The system of claim 10 , wherein the trained machine learning model employs a classification method, the classification method depending on a level of learning between behavioral attributes of a matched set of computing devices and behavioral attributes of the second set of computing devices.