Machine learning-based analysis of computing device images included in requests to service computing devices

Patent metadata

Field	Value
Publication number	US-11941792-B2
Application number	US-202117226305-A
Country	US
Kind code	B2
Filing date	Apr 9, 2021
Priority date	Apr 9, 2021
Publication date	Mar 26, 2024
Grant date	Mar 26, 2024

Abstract

An apparatus comprises a processing device configured to receive a service request for a given computing device, the service request comprising a given image of a given computing device, to generate a given image embedding for the given image utilizing a machine learning model, and to determine similarity measures between the given image embedding and other image embeddings for other images of computing devices utilizing an angular similarity metric. The processing device is configured to identify whether the given image exhibits at least a threshold level of similarity to at least one other image based at least in part on the determined similarity measures, to classify the given image as potentially fraudulent responsive to identifying the given image as exhibiting at least the threshold level of similarity to at least one other image, and to modify processing of the service request responsive to classifying the given image as potentially fraudulent.

First claim

What is claimed is:

1. An apparatus comprising:
at least one processing device comprising a processor coupled to a memory;
the at least one processing device being configured to perform steps of:
receiving a service request for a given computing device, the service request comprising a given image of the given computing device, the given image comprising user identifying information for a user submitting the service request for the given computing device;
generating, utilizing a machine learning model, a given image embedding for the given image;
determining similarity measures between the given image embedding for the given image and one or more other image embeddings for one or more other images of computing devices utilizing an angular similarity metric;
identifying whether the given image exhibits at least a threshold level of similarity to at least one of the one or more other images based at least in part on the determined similarity measures;
classifying the given image as potentially fraudulent responsive to identifying the given image as exhibiting at least the threshold level of similarity to said at least one of the one or more other images; and
modifying processing of the service request responsive to classifying the given image as potentially fraudulent.

2. The apparatus of claim 1 wherein the given image further comprises device identifying information for the given computing device.

3. The apparatus of claim 2 wherein the device identifying information for the given computing device comprises at least a portion of at least one of a serial number and a product tag of the given computing device.

4. The apparatus of claim 2 wherein classifying the given image as potentially fraudulent comprises determining that at least one of the device identifying information and the user identifying information in the given image comprises a modification of at least one of device identifying information and user identifying information in said at least one of the one or more other images.

5. The apparatus of claim 1 wherein the machine learning model comprises a convolutional neural network, the convolutional neural network comprising one or more convolutional layers, one or more pooling layers, and a fully connected layer.

6. The apparatus of claim 5 wherein the given image embedding for the given image comprises a normalized representation of an output of the fully connected layer.

7. The apparatus of claim 6 wherein the output of the fully connected layer of the machine learning model is subject to dimensionality reduction prior to normalization to produce the given image embedding for the given image.

8. The apparatus of claim 1 wherein the machine learning model is trained utilizing a contrastive loss function that determines similarity between pairs of images in a training data set, the pairs of images in the training data set comprising two or more copies of at least one image having different image transforms applied thereto.

9. The apparatus of claim 1 wherein the angular similarity metric comprises a cosine similarity metric.

10. The apparatus of claim 1 wherein the at least one processing device is further configured to perform the step of generating a graph network of a plurality of images comprising the given image and the one or more other images, the graph network comprising nodes representing respective ones of the plurality of images and edges connecting pairs of images in the plurality of images, the edges being created based at least in part on the determined similarity measures.

11. The apparatus of claim 10 wherein determining the similarity measures between the given image embedding for the given image and the one or more other image embeddings for one or more other images of computing devices utilizing the angular similarity metric comprises generating a similarity matrix having entries comprising the similarity measures.

12. The apparatus of claim 11 wherein identifying the given image as exhibiting at least the threshold level of similarity to said at least one of the one or more other images based at least in part on the determined similarity measures comprises applying a thresholding filter to the similarity matrix to produce an adjacency matrix, the adjacency matrix having entries comprising values defining the edges of the graph network.

13. The apparatus of claim 10 wherein the at least one processing device is further configured to perform the step of detecting one or more clusters of images in the plurality of images by applying one or more community detection algorithms to the graph network.

14. The apparatus of claim 13 wherein the one or more community detection algorithms comprise a Louvain community detection algorithm.

15. A computer program product comprising a non-transitory processor-readable storage medium having stored therein program code of one or more software programs, wherein the program code when executed by at least one processing device causes the at least one processing device to perform steps of:
receiving a service request for a given computing device, the service request comprising a given image of the given computing device, the given image comprising user identifying information for a user submitting the service request for the given computing device;
generating, utilizing a machine learning model, a given image embedding for the given image;
determining similarity measures between the given image embedding for the given image and one or more other image embeddings for one or more other images of computing devices utilizing an angular similarity metric;
identifying whether the given image exhibits at least a threshold level of similarity to at least one of the one or more other images based at least in part on the determined similarity measures;
classifying the given image as potentially fraudulent responsive to identifying the given image as exhibiting at least the threshold level of similarity to said at least one of the one or more other images; and
modifying processing of the service request responsive to classifying the given image as potentially fraudulent.

16. The computer program product of claim 15 wherein the given image further comprises device identifying information for the given computing device.

17. The computer program product of claim 16 wherein classifying the given image as potentially fraudulent comprises determining that at least one of the device identifying information and the user identifying information in the given image comprises a modification of at least one of device identifying information and user identifying information in said at least one of the one or more other images.

18. A method comprising steps of:
receiving a service request for a given computing device, the service request comprising a given image of the given computing device, the given image comprising user identifying information for a user submitting the service request for the given computing device;
generating, utilizing a machine learning model, a given image embedding for the given image;
determining similarity measures between the given image embedding for the given image and one or more other image embeddings for one or more other images of computing devices utilizing an angular similarity metric;
identifying whether the given image exhibits at least a threshold level of similarity to at least one of the one or more other images based at least in part on the determined similarity measures;
classifying the given image as potentially fraudulent responsive to identifying the give

Assignees

• Dell Products Lp

Inventors

• Fauber Benjamin
• Saeugling Lee

Classifications

• G06N3/0895

  Weakly supervised learning, e.g. semi-supervised or self-supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06T7/0002 Primary

  Inspection of images, e.g. flaw detection · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title