Relational Stacked Denoising Autoencoder for Tag Recommendation

By Hao Wang, Xingjian Shi, Dit-Yan Yeung. 2015

145 citations recorded by the source corpus.

Tag recommendation has become one of the most important ways of organizing and indexing online resources like articles, movies, and music. Since tagging information is usually very sparse, effective learning of the content representation for these resources is crucial to accurate tag recommendation. Recently, models proposed for tag recommendation, such as collaborative topic regression and its variants, have demonstrated promising accuracy. However, a limitation of these models is that, by using topic models like latent Dirichlet allocation as the key component, the learned representation may not be compact and effective enough. Moreover, since relational data exist as an auxiliary data source in many applications, it is desirable to incorporate such data into tag recommendation models. In this paper, we start with a deep learning model called stacked denoising autoencoder (SDAE) in an attempt to learn more effective content representation. We propose a probabilistic formulation for SDAE and then extend it to a relational SDAE (RSDAE) model. RSDAE jointly performs deep representation learning and relational learning in a principled way under a probabilistic framework. Experiments conducted on three real datasets show that both learning more effective representation and learning from relational data are beneficial steps to take to advance the state of the art.

Subfields: Generative Models, Recommender Systems

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Generative Models

1. Dit-Yan Yeung
2. Hao Wang
3. Xingjian Shi
4. Yi Tay
5. Shuai Zhang
6. M. J. Jesús
7. Francisco Herrera
8. Aixin Sun

In Recommender Systems

1. Dit-Yan Yeung
2. Hao Wang
3. Xingjian Shi
4. Naiyan Wang
5. Yi Tay
6. Shuai Zhang
7. Aixin Sun
8. Ali Yurekli

Citations in the corpus

• Collaborative Deep Learning for Recommender Systems (2014, background): “The Bayesian nature of CDL also provides potential performance boost if other side information is incorporated as in [37].”
• A practical tutorial on autoencoders for nonlinear feature fusion: Taxonomy, models, software and guidelines (2018, background): “The foundation of the approach in [150] is an AE variation named RSDAE ( Relational Stacked Denoising Au-toencoder ).”
• A Hybrid Collaborative Filtering Model with Deep Structure for Recommender Systems (2017, methodology): “The first two datasets, MovieLens100K and MovieLens-1M, are commonly used for evaluating the performance of recommender systems (Wang, Shi, and Yeung 2015; Li, Kawale, and Fu 2015).”
• Deep Learning Based Recommender System (2017, methodology): “eters. In each iteration, it updatesU andV •rst, and then updatesW andb by •xingU andV. „e authors also introduced a sampling-based algorithm [161] to avoid the local optimum. Before CDL, Wang et al. [158] proposed a similar model, relational stacked denoising autoencoders (RSDAE), for tag recommend”
• Ieee Transactions on Knowledge and Data Engineering 1 towards Bayesian Deep Learning: a Framework and Some Existing Methods (unknown): “A natural way is to represent the perception component as a PGM and seamlessly connect it to the task-specific PGM, as done in [17], [64], [67].”
• Towards Bayesian Deep Learning: A Survey (2016, methodology): “As mentioned in [59], RSDAE is formulated as a novel probabilistic model which can seamlessly integrate layered”
• Relational Deep Learning: A Deep Latent Variable Model for Link Prediction (2017, methodology): “Although DeepWalk (Perozzi, Al-Rfou, and Skiena 2014) and (Wang, Shi, and Yeung 2015) are relevant to both relational data and deep learning, they are used for learning the low-dimensional representation for each node in the network instead of performing link prediction.”
• A review on deep learning for recommender systems: challenges and remedies (2018, methodology): “Wang et al. (2015b) utilize Bayesian SDAE whereas Li et al. (2015) utilize marginalized DAE to integrate auxiliary information into their recommender systems.”
• Collaborative Recurrent Autoencoder: Recommend while Learning to Fill in the Blanks (2016, methodology): “This way the feedforward computation of DRAE would be seamlessly chained together, leading to more efficient learning algorithms than the layer-wise algorithms in [23, 22] (see Section 6 of the supplementary materials for more details).”
• A Survey on Bayesian Deep Learning (2016, methodology): “Learning and Inference: [118] provides an EM-style algorithm for MAP estimation.”