Systems and methods for parsing log files using classification and a plurality of neural networks

Patent metadata

Field	Value
Publication number	US-12174871-B2
Application number	US-202318370936-A
Country	US
Kind code	B2
Filing date	Sep 21, 2023
Priority date	Oct 17, 2018
Publication date	Dec 24, 2024
Grant date	Dec 24, 2024

Abstract

The present disclosure relates to systems and methods for parsing unstructured data with neural networks. In one implementation, a system for parsing unstructured data may include at least one processor and at least one non-transitory memory storing instructions that, when executed by the at least one processor, cause the system to: receive unstructured data; apply a classifier to the unstructured data to identify a type of the unstructured data; based on the identification, select a corresponding neural network; apply the selected neural network to the unstructured data to obtain structured data; and output the structured data.

First claim

What is claimed is:

1. A system for parsing unstructured data,
comprising:
at least one processor; and
at least one non-transitory memory storing instructions that, when executed by the at least one processor, cause the system to perform operations comprising:
receiving unstructured data;
performing at least one of:
searching the unstructured data for one or more keys associated with one or more types of the unstructured data and determining a frequency of the one or more keys;
calculating one or more distributions within the unstructured data using a classifier with one or more thresholds; or
extracting one or more feature vectors from the unstructured data and comparing the one or more extracted feature vectors with one or more representative feature vectors;

based on at least one of the frequency of the one or more keys, the calculated one or more distributions, or the comparison, selecting a neural network;
applying the selected neural network to the unstructured data to obtain structured data; and
outputting the structured data.

2. They system of claim 1 , wherein the selection of the neural network is based on one or more outputs from a plurality of neural networks.

3. The system of claim 2 , wherein the selected neural network is selected based on a loss function associated with the selected neural network being closest to a minimum threshold relative to the plurality of neural networks or based on an output associated with the selected neural network having a highest robustness measure relative to the plurality of neural networks.

4. The system of claim 1 , wherein the one or more keys comprise one or more predetermined characters.

5. The system of claim 1 , wherein:
the operations further comprise using one or more frequencies of the one or more keys within the unstructured data for identifying at least one or more candidate neural networks; and
the selected neural network is selected from among the one or more candidate neural networks.

6. The system of claim 1 , wherein the unstructured data comprises a category of unstructured data for which a neural network was not previously trained.

7. The system of claim 1 , wherein the classifier determines that the unstructured data comprises the category for which a neural network was not previously trained based on an output of the classifier exceeding one or more thresholds of one or more expected outputs associated with a plurality of neural networks.

8. A computer-implemented method for parsing unstructured data, comprising:
receiving unstructured data;
performing at least one of:
searching the unstructured data for one or more keys associated with one or more types of the unstructured data and determining a frequency of the one or more keys;
calculating one or more distributions within the unstructured data using a classifier with one or more thresholds; or
extracting one or more feature vectors from the unstructured data and comparing the one or more extracted feature vectors with one or more representative feature vectors;

based on at least one of the frequency of the one or more keys, the calculated one or more distributions, or the comparison, selecting a neural network;
applying the selected neural network to the unstructured data to obtain structured data; and
outputting the structured data.

9. The computer-implemented method of claim 8 , wherein the selection of the neural network is based on one or more outputs from a plurality of neural networks.

10. The computer-implemented method of claim 9 , wherein the selected neural network is selected based on a loss function associated with the selected neural network being closest to a minimum threshold relative to the plurality of neural networks or based on an output associated with the selected neural network having a highest robustness measure relative to the plurality of neural networks.

11. The computer-implemented method of claim 8 , wherein the one or more keys comprise one or more predetermined characters.

12. The computer-implemented method of claim 8 , further comprising using one or more frequencies of the one or more keys within the unstructured data for identifying at least one or more candidate neural networks, wherein the selected neural network is selected from among the one or more candidate neural networks.

13. The computer-implemented method of claim 8 , wherein the unstructured data comprises a category of unstructured data for which a neural network was not previously trained.

14. The computer-implemented method of claim 8 , wherein the classifier determines that the unstructured data comprises the category for which a neural network was not previously trained based on an output of the classifier exceeding one or more thresholds of one or more expected outputs associated with a plurality of neural networks.

15. A system comprising:
at least one processor; and
at least one non-transitory memory storing instructions that, when executed by the at least one processor, cause the system to perform operations comprising:
receiving unstructured data;
performing, to identify a type or category of the unstructured data, at least one of:
searching the unstructured data for one or more keys associated with one or more types of the unstructured data and determining a frequency of the one or more keys;
calculating one or more distributions within the unstructured data using a classifier with one or more thresholds; or
extracting one or more feature vectors from the unstructured data and comparing the one or more extracted feature vectors with one or more representative feature vectors;

based on at least one of the frequency of the one or more keys, the calculated one or more distributions, or the comparison, selecting a neural network;
applying the selected neural network to the unstructured data to obtain structured data; and
outputting the structured data.

16. The system of claim 15 , wherein the selection of the neural network is based on one or more outputs from a plurality of neural networks.

17. The system of claim 16 , wherein the selected neural network is selected based on a loss function associated with the selected neural network being closest to a minimum threshold relative to the plurality of neural networks or based on an output associated with the selected neural network having a highest robustness measure relative to the plurality of neural networks.

18. The system of claim 15 , wherein the one or more keys comprise one or more predetermined characters.

19. The system of claim 15 , wherein:
the operations further comprise using one or more frequencies of the one or more keys within the unstructured data for identifying at least one or more candidate neural networks; and
the selected neural network is selected from among the one or more candidate neural networks.

20. The system of claim 15 , wherein the selected neural network is trained using at least one of backpropagation of error or backpropagation through time.

Assignees

• Capital One Services Llc

Inventors

• Truong Anh
• Abdi Taghi Abad Fardin
• Walters Austin
• Goodsitt Jeremy
• Pham Vincent
• Key Kate

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/096

  Transfer learning · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title