Hierarchical Representation

Part of the SpringerBriefs in Computer Science book series (BRIEFSCOMPUTER)


This chapter introduces a hierarchical interaction representation (HIR) model, which treats the interaction among different entities and contexts as representation. This model generates the interaction representation of two entities via tensor multiplication, which is performed iteratively to construct a hierarchical structure among all entities and contexts. Moreover, the model employs several hidden layers to reveal the underlying properties of this interaction representation and enhance the model performance further. After generating the final representation, the prediction can be calculated using a variety of machine learning methods according to different application tasks (e.g., linear regression for regression tasks, pair-wise ranking method for ranking tasks, and logistic regression for classification tasks).


Interaction Representation Tensor Multiplication Joint Representation Bayesian Personalized Ranking (BPR) Click-through Rate Prediction 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Liu, Q., Wu, S., Wang, L.: Collaborative prediction for multi-entity interaction with hierarchical representation. In: CIKM, pp. 613–622 (2015)Google Scholar
  2. 2.
    Mnih, A., Salakhutdinov, R.: Probabilistic matrix factorization. In: Proceedings on Neural Information Processing systems (2007)Google Scholar
  3. 3.
    Rendle, S.: Factorization machines with libfm. ACM Trans. Intell. Syst. Technol. (TIST) 3(3), 57 (2012)Google Scholar
  4. 4.
    Bengio, Y., Courville, A., Vincent, P.: Representation learning: a review and new perspectives. IEEE Trans. Pattern Anal. Mach. Intell. 35(8), 1798–1828 (2013)CrossRefGoogle Scholar

Copyright information

© The Author(s) 2017

Authors and Affiliations

  1. 1.National Laboratory of Pattern Recognition, Institute of AutomationChinese Academy of SciencesBeijingChina

Personalised recommendations