Open AccessNovel Approach of Neural Collaborative Filter by Pairwise Objective Function with Matrix Factorization
Author(s) -
Ram Sethuraman,
Akshay Havalgi
Publication year - 2018
Publication title -
international journal of engineering and technology
Language(s) - English
Resource type - Journals
ISSN - 2227-524X
DOI - 10.14419/ijet.v7i3.12.17840
Subject(s) - collaborative filtering , computer science , pairwise comparison , matrix decomposition , gradient descent , artificial neural network , artificial intelligence , recommender system , non negative matrix factorization , gaussian , activation function , matrix (chemical analysis) , machine learning , algorithm , eigenvalues and eigenvectors , physics , materials science , composite material , quantum mechanics
The concept of deep learning is used in the various fields like text, speech and vision. The proposed work deep neural network is used for recommender system. In this work pair wise objective function is used for emphasis of non-linearity and latent features. The GMF (Gaussian matrix factorization) and MLP techniques are used in this work. The proposed framework is named as NCF which is basically neural network based collaborative filtering. The NCF gives the latent features by reducing the non-linearity and generalizing the matrix. In the proposed work combination of pair-wise and point wise objective function is used and tune by using the concept of cross entropy with Adam optimization. This optimization approach optimizes the gradient descent function. The work is done on 1K and 1M movies lens dataset and it is compared with deep matrix factorization (DMF).