US 12,242,829 B2
Systems and methods for source code understanding using spatial representations
Ruibo Shi, London (GB); Sean Moran, London (GB); Lili Tao, Bristol (GB); Fanny Silavong, London (GB); and Rohan Saphal, Glasgow (GB)
Assigned to JPMORGAN CHASE BANK, N.A., New York, NY (US)
Filed by JPMORGAN CHASE BANK, N.A., New York, NY (US)
Filed on Nov. 14, 2022, as Appl. No. 18/055,182.
Claims priority of provisional application 63/279,562, filed on Nov. 15, 2021.
Prior Publication US 2023/0153085 A1, May 18, 2023
Int. Cl. G06F 9/44 (2018.01); G06F 8/41 (2018.01); G06N 20/00 (2019.01)
CPC G06F 8/41 (2013.01) [G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method for source code understanding using spatial representations, comprising:
receiving, by a source code understanding computer program, a source code snippet;
converting, by the source code understanding computer program, the source code snippet to a matrix representation using an encoding technique, wherein the matrix representation encodes structural and contextual information of the source code snippet;
encoding, by the source code understanding computer program, the matrix representation into a three-dimensional image representation using one hot encoding or trainable embeddings;
determining, by the source code understanding computer program, a discriminative task for the source code snippet using a deep learning network; and
providing, by the source code understanding computer program, the discriminative task for the source code snippet to a downstream system.