Wide and Deep Recommendation Algorithm
A Wide and Deep Recommendation Algorithm is a joint training algorithm with a wide network (a linear estimator) and a deep neural network (which the latent representations for user and item).
- Context:
- It can be used by a Wide and Deep-based Recommender System.
- …
- Counter-Example(s):
- DeepFM.
- …
- See: Recommender System.
References
2018
- https://humboldt-wi.github.io/blog/research/information_systems_1718/08recommendation/
- QUOTE: ...
Wide and Deep Model
The wide and deep learning has two individual components. The wide network is a linear estimator or a single layer feed-forward network. By assigning weights to each features and adding them with a bias term, it models the matrix factorization method. The deep neural network learns better representations of the latent vectors and introduces non-linearities in the latent representations for user and item. By jointly training the wide and deep network, the weights are optimized by back propagating the gradients from the output to each network simultaneously. ...
- QUOTE: ...
2016
- (Cheng et al., 2016) ⇒ Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, Rohan Anil, Zakaria Haque, Lichan Hong, Vihan Jain, Xiaobing Liu, and Hemal Shah. (2016). “Wide & Deep Learning for Recommender Systems.” In: Proceedings of the 1st Workshop on Deep Learning for Recommender Systems. ISBN:978-1-4503-4795-2 doi:10.1145/2988450.2988454
- QUOTE: ... Generalized linear models with nonlinear feature transformations are widely used for large-scale regression and classification problems with sparse inputs. Memorization of feature interactions through a wide set of cross-product feature transformations are effective and interpretable, while generalization requires more feature engineering effort. With less feature engineering, deep neural networks can generalize better to unseen feature combinations through low-dimensional dense embeddings learned for the sparse features. However, deep neural networks with embeddings can over-generalize and recommend less relevant items when the user-item interactions are sparse and high-rank. In this paper, we present Wide & Deep learning --- jointly trained wide linear models and deep neural networks --- to combine the benefits of memorization and generalization for recommender systems. ...