Upgrade to Pro — share decks privately, control downloads, hide ads and more …

[卒論中間発表] Encoder-Decoder DKTと中間表現の分析

Qiushi Pan
October 09, 2019

[卒論中間発表] Encoder-Decoder DKTと中間表現の分析

Qiushi Pan

October 09, 2019
Tweet

More Decks by Qiushi Pan

Other Decks in Research

Transcript

  1. 提案⼿法: Encoder-Decoder DKT (EDDKT) (1) X = src x ,

    x , … , x { 1 2 T−1} X = trg x { T } 9
  2. Encoder-Decoder DKT (EDDKT) (2) (⽣成モデル応⽤案) X = src x ,

    x , … , x { 1 2 T−1} X = trg x , x { T−1 T } X = trg x , x { T T+1} X = trg x , y δ(q ) { T T+1 T+1 } 10
  3. 実験結果 (3) EDDKT extend forward 1; EDDKT extend forward 19;

    len(x ) = src 19, len(x ) = trg 1 len(x ) = src 1, len(x ) = trg 19 13
  4. WIP: 指摘した遷移振動問題(wavy transition problem)[3]を定量的評価する実装 w = 1 M T −1

    ∑ i=1 n ( i ) y −y ∑ i=1 n ∑ t=1 T −1 i ∥ t+1 i t i ∥ 1 w = 2 2 M T −1 ∑ i=1 n ( i ) y −y ∑ i=1 n ∑ t=1 T −1 i ∥ t+1 i t i ∥ 2 2 14
  5. 参考⽂献(1) [1] Corbett, A. T. and Anderson, J. R.: Knowledge

    tracing: Modeling the acquisition of procedural knowledge, User Modeling and User-adapted Interaction, Vol. 4, No. 4, pp. 253–278 (1994). [2] Piech, C., Bassen, J., Huang, J., Ganguli, S., Sahami, M., Guibas, L. J. and Sohl- Dickstein, J.: Deep knowledge tracing, Advances in Neural Information Processing Systems, pp. 505–513 (2015). [3] Yeung, C.-K. and Yeung, D.-Y.: Addressing two problems in deep knowledge tracing via prediction-consistent regularization, arXiv preprint arXiv:1806.02180 (2018). 16
  6. 参考⽂献(2) [4] Zhang, J., Shi, X., King, I. and Yeung,

    D.-Y.: Dynamic Key-Value Memory Networks for Knowledge Tracing, Proceedings of the 26th International Conference on World Wide Web, WWW ʼ17, Republic and Canton of Geneva, Switzerland, International World Wide Web Conferences Steering Committee, pp. 765–774 (online), DOI: 10.1145/3038912.3052580 (2017). [5] Yeung, C.: Deep-IRT: Make Deep Learning Based Knowledge Tracing Explainable Using Item Response Theory, CoRR, Vol. abs/1904.11738 (online), available from ⟨ http://arxiv.org/abs/1904.11738⟩ (2019). 17
  7. 遷移振動問題(wavy transition problem) Yeungらによる定量評価の定式化 w = 1 M T −1

    ∑ i=1 n ( i ) y −y ∑ i=1 n ∑ t=1 T −1 i ∥ t+1 i t i ∥ 1 w = 2 2 M T −1 ∑ i=1 n ( i ) y −y ∑ i=1 n ∑ t=1 T −1 i ∥ t+1 i t i ∥ 2 2 19