Learning Program Embeddings to Propagate Feedback on Student Code

Chris Piech, Jonathan Huang, Andy Nguyen, Mike Phulsuksombati, Mehran Sahami, Leonidas Guibas
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:1093-1102, 2015.

Abstract

Providing feedback, both assessing final work and giving hints to stuck students, is difficult for open-ended assignments in massive online classes which can range from thousands to millions of students. We introduce a neural network method to encode programs as a linear mapping from an embedded precondition space to an embedded postcondition space and propose an algorithm for feedback at scale using these linear maps as features. We apply our algorithm to assessments from the Code.org Hour of Code and Stanford University’s CS1 course, where we propagate human comments on student assignments to orders of magnitude more submissions.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-piech15, title = {Learning Program Embeddings to Propagate Feedback on Student Code}, author = {Piech, Chris and Huang, Jonathan and Nguyen, Andy and Phulsuksombati, Mike and Sahami, Mehran and Guibas, Leonidas}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {1093--1102}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/piech15.pdf}, url = {https://proceedings.mlr.press/v37/piech15.html}, abstract = {Providing feedback, both assessing final work and giving hints to stuck students, is difficult for open-ended assignments in massive online classes which can range from thousands to millions of students. We introduce a neural network method to encode programs as a linear mapping from an embedded precondition space to an embedded postcondition space and propose an algorithm for feedback at scale using these linear maps as features. We apply our algorithm to assessments from the Code.org Hour of Code and Stanford University’s CS1 course, where we propagate human comments on student assignments to orders of magnitude more submissions.} }
Endnote
%0 Conference Paper %T Learning Program Embeddings to Propagate Feedback on Student Code %A Chris Piech %A Jonathan Huang %A Andy Nguyen %A Mike Phulsuksombati %A Mehran Sahami %A Leonidas Guibas %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-piech15 %I PMLR %P 1093--1102 %U https://proceedings.mlr.press/v37/piech15.html %V 37 %X Providing feedback, both assessing final work and giving hints to stuck students, is difficult for open-ended assignments in massive online classes which can range from thousands to millions of students. We introduce a neural network method to encode programs as a linear mapping from an embedded precondition space to an embedded postcondition space and propose an algorithm for feedback at scale using these linear maps as features. We apply our algorithm to assessments from the Code.org Hour of Code and Stanford University’s CS1 course, where we propagate human comments on student assignments to orders of magnitude more submissions.
RIS
TY - CPAPER TI - Learning Program Embeddings to Propagate Feedback on Student Code AU - Chris Piech AU - Jonathan Huang AU - Andy Nguyen AU - Mike Phulsuksombati AU - Mehran Sahami AU - Leonidas Guibas BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-piech15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 1093 EP - 1102 L1 - http://proceedings.mlr.press/v37/piech15.pdf UR - https://proceedings.mlr.press/v37/piech15.html AB - Providing feedback, both assessing final work and giving hints to stuck students, is difficult for open-ended assignments in massive online classes which can range from thousands to millions of students. We introduce a neural network method to encode programs as a linear mapping from an embedded precondition space to an embedded postcondition space and propose an algorithm for feedback at scale using these linear maps as features. We apply our algorithm to assessments from the Code.org Hour of Code and Stanford University’s CS1 course, where we propagate human comments on student assignments to orders of magnitude more submissions. ER -
APA
Piech, C., Huang, J., Nguyen, A., Phulsuksombati, M., Sahami, M. & Guibas, L.. (2015). Learning Program Embeddings to Propagate Feedback on Student Code. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:1093-1102 Available from https://proceedings.mlr.press/v37/piech15.html.

Related Material