DNNRec: A novel deep learning based hybrid recommender system

• Published in: Expert systems with applications Vol. 144; p. 113054
• Main Authors: R, Kiran, Kumar, Pradeep, Bhasker, Bharat
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 15-04-2020; Elsevier BV
• Subjects: Accuracy; Artificial neural networks; Cold; Cold start problem; Cold starts; Collaboration; Cyclical learning rates; Datasets; Deep learning; Deep neural network; Embedded systems; Embeddings; Error analysis; Filtration; Hybrid systems; Machine learning; Neural networks; Recommender systems; Root-mean-square errors; Side information; Time measurement; Deep learning; Embeddings; Deep neural network; Cyclical learning rates; Side information; Cold start problem; Recommender systems
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2019.113054

•A novel hybrid deep learning based recommender system ‘DNNRec’ is proposed.•DNNRec leverages embeddings, combines side information and a very deep network.•DNNRec addresses cold start case and learns of non-linear latent factors.•Proposed solution is benchmarked against existing methods on accuracy and run time.•DNNRec outperforms state-of-the-art methods overall and in cold start case.
We propose a novel deep learning hybrid recommender system to address the gaps in Collaborative Filtering systems and achieve the state-of-the-art predictive accuracy using deep learning. While collaborative filtering systems are popular with many state-of-the-art achievements in recommender systems, they suffer from the cold start problem, when there is no history about the users and items. Further, the latent factors learned by these methods are linear in nature. To address these gaps, we describe a novel hybrid recommender system using deep learning. The solution uses embeddings for representing users and items to learn non-linear latent factors. The solution alleviates the cold start problem by integrating side information about users and items into a very deep neural network. The proposed solution uses a decreasing learning rate in conjunction with increasing weight decay, the values cyclically varied across epochs to further improve accuracy. The proposed solution is benchmarked against existing methods on both predictive accuracy and running time. Predictive Accuracy is measured by Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and R-squared. Running time is measured by the mean and standard deviation across seven runs. Comprehensive experiments are conducted on several datasets such as the MovieLens 100 K, FilmTrust, Book-Crossing and MovieLens 1 M. The results show that the proposed technique outperforms existing methods in both non-cold start and cold start cases. The proposed solution framework is generic from the outperformance on four different datasets and can be leveraged for other ratings prediction datasets in recommender systems.