From programs to interpretable deep models and back

Eran Yahav

doi:https://doi.org/10.1007/978-3-319-96145-3_2

From programs to interpretable deep models and back

Eran Yahav

Computer Science

Technion - Israel Institute of Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

We demonstrate how deep learning over programs is used to provide (preliminary) augmented programmer intelligence. In the first part, we show how to tackle tasks like code completion, code summarization, and captioning. We describe a general path-based representation of source code that can be used across programming languages and learning tasks, and discuss how this representation enables different learning algorithms. In the second part, we describe techniques for extracting interpretable representations from deep models, shedding light on what has actually been learned in various tasks.

Original language	English
Title of host publication	Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings
Editors	Georg Weissenbacher, Hana Chockler
Pages	27-37
Number of pages	11
DOIs	https://doi.org/10.1007/978-3-319-96145-3_2
State	Published - 2018
Event	30th International Conference on Computer Aided Verification, CAV 2018 Held as Part of the Federated Logic Conference, FloC 2018 - Oxford, United Kingdom Duration: 14 Jul 2018 → 17 Jul 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10981 LNCS

Conference

Conference	30th International Conference on Computer Aided Verification, CAV 2018 Held as Part of the Federated Logic Conference, FloC 2018
Country/Territory	United Kingdom
City	Oxford
Period	14/07/18 → 17/07/18

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

https://doi.org/10.1007/978-3-319-96145-3_2

Cite this

Yahav, E. (2018). From programs to interpretable deep models and back. In G. Weissenbacher, & H. Chockler (Eds.), Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings (pp. 27-37). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10981 LNCS). https://doi.org/10.1007/978-3-319-96145-3_2

Yahav, Eran. / From programs to interpretable deep models and back. Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings. editor / Georg Weissenbacher ; Hana Chockler. 2018. pp. 27-37 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{a6b8f028ba0a4ef4b0664d2da89ee282,

title = "From programs to interpretable deep models and back",

abstract = "We demonstrate how deep learning over programs is used to provide (preliminary) augmented programmer intelligence. In the first part, we show how to tackle tasks like code completion, code summarization, and captioning. We describe a general path-based representation of source code that can be used across programming languages and learning tasks, and discuss how this representation enables different learning algorithms. In the second part, we describe techniques for extracting interpretable representations from deep models, shedding light on what has actually been learned in various tasks.",

author = "Eran Yahav",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2018.; 30th International Conference on Computer Aided Verification, CAV 2018 Held as Part of the Federated Logic Conference, FloC 2018 ; Conference date: 14-07-2018 Through 17-07-2018",

year = "2018",

doi = "https://doi.org/10.1007/978-3-319-96145-3_2",

language = "الإنجليزيّة",

isbn = "9783319961446",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

pages = "27--37",

editor = "Georg Weissenbacher and Hana Chockler",

booktitle = "Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings",

}

Yahav, E 2018, From programs to interpretable deep models and back. in G Weissenbacher & H Chockler (eds), Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10981 LNCS, pp. 27-37, 30th International Conference on Computer Aided Verification, CAV 2018 Held as Part of the Federated Logic Conference, FloC 2018, Oxford, United Kingdom, 14/07/18. https://doi.org/10.1007/978-3-319-96145-3_2

From programs to interpretable deep models and back. / Yahav, Eran.
Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings. ed. / Georg Weissenbacher; Hana Chockler. 2018. p. 27-37 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10981 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - From programs to interpretable deep models and back

AU - Yahav, Eran

PY - 2018

Y1 - 2018

N2 - We demonstrate how deep learning over programs is used to provide (preliminary) augmented programmer intelligence. In the first part, we show how to tackle tasks like code completion, code summarization, and captioning. We describe a general path-based representation of source code that can be used across programming languages and learning tasks, and discuss how this representation enables different learning algorithms. In the second part, we describe techniques for extracting interpretable representations from deep models, shedding light on what has actually been learned in various tasks.

AB - We demonstrate how deep learning over programs is used to provide (preliminary) augmented programmer intelligence. In the first part, we show how to tackle tasks like code completion, code summarization, and captioning. We describe a general path-based representation of source code that can be used across programming languages and learning tasks, and discuss how this representation enables different learning algorithms. In the second part, we describe techniques for extracting interpretable representations from deep models, shedding light on what has actually been learned in various tasks.

UR - http://www.scopus.com/inward/record.url?scp=85051120513&partnerID=8YFLogxK

U2 - https://doi.org/10.1007/978-3-319-96145-3_2

DO - https://doi.org/10.1007/978-3-319-96145-3_2

M3 - منشور من مؤتمر

SN - 9783319961446

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 27

EP - 37

BT - Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings

A2 - Weissenbacher, Georg

A2 - Chockler, Hana

T2 - 30th International Conference on Computer Aided Verification, CAV 2018 Held as Part of the Federated Logic Conference, FloC 2018

Y2 - 14 July 2018 through 17 July 2018

ER -

Yahav E. From programs to interpretable deep models and back. In Weissenbacher G, Chockler H, editors, Computer Aided Verification - 30th International Conference, CAV 2018, Held as Part of the Federated Logic Conference, FloC 2018, Proceedings. 2018. p. 27-37. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: https://doi.org/10.1007/978-3-319-96145-3_2

From programs to interpretable deep models and back

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this