Deep learning has received leapfrog progress in the realm of Machine learning such as image processing, speech recognition, the natural language processing, and recommendation systems. The traditional recommendation algorithm has existed the matter of cold start and data sparsity. To alleviate such problems, we propose a deep autoencoder algorithm for collaborative filtering recommendation AE-CF algorithm, which incorporates autoencoder and collaborative filtering recommended algorithms. The proposed AE-CF algorithm learn deep latent factors from users feature data and ratings. We evaluate the proposed AE-CF algorithm by applying the MovieLens dataset, a public dataset for movie recommendations. The experimental results demonstrate that AE-CF algorithm can effectively reduce the recommendation error and thus improve the recommendation quality.
Paper
Full text
Towards a Deep Learning Autoencoder algorithm for Collaborative Filtering Recommendation
Semantic Scholar · Computer Science · 2019
Abstract
Deep learning has received leapfrog progress in the realm of Machine learning such as image processing, speech recognition, the natural language processing, and recommendation systems. The traditional recommendation algorithm has existed the matter of cold start and data sparsity. To alleviate such problems, we propose a deep autoencoder algorithm for collaborative filtering recommendation AE-CF algorithm, which incorporates autoencoder and collaborative filtering recommended algorithms. The proposed AE-CF algorithm learn deep latent factors from users feature data and ratings. We evaluate the proposed AE-CF algorithm by applying the MovieLens dataset, a public dataset for movie recommendations. The experimental results demonstrate that AE-CF algorithm can effectively reduce the recommendation error and thus improve the recommendation quality.