A Recommender Framework for BP Neural Networks with an Attention Mechanism

PROJECT TITLE :

A BP Neural Network Based Recommender Framework with Attention Mechanism

ABSTRACT:

Due to the nonlinear representation learning capability of DNNs, there have recently been some attempts made to implement deep neural networks (DNNs) into recommender systems. The goal of these attempts is to generate more accurate predictions using the systems. However, the high computational and storage costs that result from their use are unavoidable. Worse still, given the limited number of ratings that can be input into DNNs, it is possible for these networks to be susceptible to the overfitting problem. In order to address these concerns, we have developed a novel recommendation framework that we refer to as BPAM++. This framework is based on a Back Propagation (BP) neural network and includes an attention mechanism. In particular, the BP neural network is used to learn the intricate relationship that exists between the target user and his or her neighbors, as well as the intricate relationship that exists between the target item and its neighbors. When compared to DNNs, shallow neural networks, also known as BP neural networks, have the ability to not only reduce the costs associated with computation and storage, but also alleviate the overfitting issues that can arise in DNNs as a result of a relatively low number of ratings. Additionally, an attention mechanism is designed to capture the global impact of the nearest users of the target user on their respective nearest target user sets. This is done by focusing on the user's nearest target user sets. The effectiveness of the proposed model has been validated through an exhaustive set of experiments carried out on eight benchmark datasets.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here