# Neural Collaborative Filtering

**Authors:** Xiangnan He, Lizi Liao, Hanwang Zhang  
**arXiv:** [1708.05031](https://arxiv.org/abs/1708.05031)  
**Date:** 2017-08-16  
**Summary page:** https://www.summarizepaper.com/en/paper/1708.05031v1/

## Key points

- Authors Xiangnan He, Lizi Liao, and Hanwang Zhang explore recommender systems using deep neural networks
- They introduce a novel framework called Neural network-based Collaborative Filtering (NCF)
- NCF aims to address the gap in utilizing deep learning for collaborative filtering in recommendation systems
- The framework offers a generic structure capable of expressing and generalizing matrix factorization
- To enhance NCF modeling with non-linearities, they suggest leveraging a multi-layer perceptron to learn the user-item interaction function
- Extensive experiments on real-world datasets show significant improvements in recommendation performance compared to traditional methods
- The study highlights the potential of deep neural networks in revolutionizing recommender systems through innovative frameworks like NCF that focus on learning complex user-item interactions using neural architectures

## Summary

In their paper titled "Neural Collaborative Filtering," authors Xiangnan He, Lizi Liao, and Hanwang Zhang delve into the realm of recommender systems by exploring the application of deep neural networks. The authors aim to address the gap in utilizing deep learning for collaborative filtering in recommendation systems. They propose a novel framework called Neural network-based Collaborative Filtering (NCF) that offers a generic structure capable of expressing and generalizing matrix factorization. To enhance NCF modeling with non-linearities, they suggest leveraging a multi-layer perceptron to learn the user-item interaction function. Through extensive experiments on real-world datasets, they demonstrate significant improvements in recommendation performance compared to traditional methods. The study showcases the potential of deep neural networks in revolutionizing recommender systems through innovative frameworks like NCF that focus on learning complex user-item interactions using neural architectures.