Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Addressing Two Problems in Deep Knowledge Tracing via Prediction-Consistent Regularization, Yeung+, 2018, L@S #421

Open
AkihikoWatanabe opened this issue Oct 29, 2021 · 3 comments

Comments

@AkihikoWatanabe
Copy link
Owner

https://arxiv.org/pdf/1806.02180.pdf

@AkihikoWatanabe
Copy link
Owner Author

AkihikoWatanabe commented Oct 29, 2021

Deep Knowledge Tracing (DKT)では、下記の問題がある:

  • 該当スキルに正解/不正解 したのにmasteryが 下がる/上がる (Inputをreconstructしない)
  • いきなり習熟度が伸びたり、下がったりする(時間軸に対してmastery levelがconsistentではない)
    上記問題に対処するようなモデルDKT+を提案。

DKT+では、DKTのloss functionに対して3つのregularization termを追加することで上記問題に対処している。
DKT+はDKTの性能を落とすことなく、上記2問題を緩和できたとのこと。

image

@AkihikoWatanabe
Copy link
Owner Author

AkihikoWatanabe commented Oct 29, 2021

@AkihikoWatanabe
Copy link
Owner Author

image

DKT+とDKTのheatmapを比較すると、問題点は確かに緩和されているかもしれないが、
依然としてinputはreconstructionされていないし、習熟度も乱高下しているように見える。
根本的な解決にはなっていないのでは。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant