Unlock instant, AI-driven research and patent intelligence for your innovation.

Deep learning model for learning program embeddings

A program and program state technology, applied in the field of deep learning models for learning program embedding, can solve problems such as unstable semantic understanding of models, not completely effective, etc.

Active Publication Date: 2021-09-10
VISA INT SERVICE ASSOC
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These previous models were not stable in understanding semantics, and thus were not fully effective

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Deep learning model for learning program embeddings
  • Deep learning model for learning program embeddings
  • Deep learning model for learning program embeddings

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0022] Embodiments of the present disclosure include a deep learning model configured to learn dynamic program semantics. Semantics includes the meaning of a piece of text (e.g., the meaning of a sentence, the function of a computer program), rather than the syntax or content of a piece of text (e.g., the words in a sentence, the variables in a computer program). For example, the sentence "I have a black cat" and the sentence "I have a cat and it is black" have the same semantics despite having different syntax. The semantics of a program may be related to the function of the program. Program functionality can refer to the problem solved by the program, while program semantics refers to the way the problem is solved by the program.

[0023] Unlike static deep learning models that learn from program text, dynamic deep learning models can be models that learn from the execution of programs. Deep learning models can allow program embeddings to be learned through neural networks...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A system and method for using a deep learning model to learn program semantics is disclosed. The method includes receiving a plurality of execution traces of a program, each execution trace comprising a plurality of variable values. The plurality of variable values are encoded by a first recurrent neural network to generate a plurality of program states for each execution trace. A bi-directional recurrent neural network can then determine a reduced set of program states for each execution trace from the plurality of program states. The reduced set of program states are then encoded by a second recurrent neural network to generate a plurality of executions for the program. The method then includes pooling the plurality of executions to generate a program embedding and predicting semantics of the program using the program embedding.

Description

[0001] Related Application Cross Reference [0002] This application is a non-provisional application of and claims the benefit of U.S. Provisional Application No. 62 / 793,500, filed January 17, 2019, which is hereby incorporated by reference in its entirety. Background technique [0003] The growing trend of open source software and the rapid development of machine learning techniques have made the concept of "big code" concrete. This concept is reusing knowledge extracted from existing code repositories to, for example, simplify software development and improve product quality. Some early approaches in the field mainly processed source code into a piece of text and applied off-the-shelf models from the domain of natural language processing (Abram Hindle, Earl T Barr, Zhendong Su, Mark Gabel, and Premkumar Devanbu. On the naturalness of software, in Software Engineering (ICSE), 2012 34th International Conference, pp. 837-847, IEEE, 2012, Rahul Gupta, Soham Pal, Aditya Kanade,...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F11/36G06N3/08
CPCG06N3/08G06F11/3466G06F11/3612G06F8/36G06F8/436G06F8/4435G06N3/044G06N3/045G06F11/36G06N20/00
Inventor 王轲
Owner VISA INT SERVICE ASSOC