Implement DeepFM model in Keras

Introduction

Wide and deep architect has been proven as one of deep learning applications combining memorization and generalization in areas such as search and recommendation. Google released its wide&deep learning in 2016.

• wide part: helps to memorize the past behavior for specific choice
• deep part: embed into low dimension, help to discover new user, product combinations

Later, on top of wide & deep learning, deepfm was developed combining DNN model and Factorization machines, to further address the interactions among the features.

wide & deep model

DeepFM model

Comparison

wide&deep learning is logistic regression + deep neural network. In wide part of wide & deep learning, it is a logistic regression, which requires a lot of manual feature engineering efforts to generate the large-scale feature set for wide part.

While the deepfm model instead is factorization machines + deep neural network, as known as neural factorization machines.

Implementation of Model Based Recommendation System in R

Overview

The most straight forward recommendation system are either user based CF (collaborative filtering) or item based CF, which are categorized as memory based methods. User-Based CF is to recommend products based on behaviour of similar users, and the Item-Based CF is to recommend similar products from products that user purchased. No matter which method is used, the user-user or item-item similarity matrix, which could be sizable, is required to compute.

While on the contrast, a model based approach could refer to converting recommendation problem to regression, classification, learning to rank problem. Matrix Factorization, which is also known as latent factor model,SVD, is one of the most commonly used model based methods. In this post, a variety methods of CF will be discussed, including:

• Gradient Descent CF (GD)
• Alternating Least Square (ALS)