CDAE

Introduction

[paper]

Title: Collaborative Denoising Auto-Encoders for Top-N Recommender Systems

Authors: Yao Wu, Christopher DuBois, Alice X. Zheng, Martin Ester

Abstract: Most real-world recommender services measure their performance based on the top-N results shown to the end users. Thus, advances in top-N recommendation have far-ranging consequences in practical applications. In this paper, we present a novel method, called Collaborative Denoising Auto-Encoder (CDAE), for top-N recommendation that utilizes the idea of Denoising Auto-Encoders. We demonstrate that the proposed model is a generalization of several well-known collaborative filtering models but with more flexible components. Thorough experiments are conducted to understand the performance of CDAE under various component settings. Furthermore, experimental results on several public datasets demonstrate that CDAE consistently outperforms state-of-the-art top-N recommendation methods on a variety of common evaluation metrics.

../../../_images/cdae.png

Running with RecBole

Model Hyper-Parameters:

  • loss_type (str) : The loss function of model, now we support [BCE, MSE]. Defaults to BCE.

  • hid_activation (str) : The hidden layer activation function, now we support [sigmoid, relu, tanh] Defaults to relu.

  • out_activation (str) : The output layer activation function, now we support [sigmoid, relu]. Defaults to sigmoid.

  • corruption_ratio (float) : The corruption ratio of the input. Defaults to 0.5.

  • embedding_size (int) : The embedding size of user. Defaults to 64.

  • reg_weight_1 (float) : L1-regularization weight. Defaults to 0..

  • reg_weight_2 (float) : L2-regularization weight. Defaults to 0.01.

A Running Example:

Write the following code to a python file, such as run.py

from recbole.quick_start import run_recbole

run_recbole(model='CDAE', dataset='ml-100k')

And then:

python run.py

Note: Because this model is a non-sampling model, so you must set training_neg_sample=0 when you run this model.

Tuning Hyper Parameters

If you want to use HyperTuning to tune hyper parameters of this model, you can copy the following settings and name it as hyper.test.

learning_rate choice [0.01,0.005,0.001,0.0005,0.0001]

Note that we just provide these hyper parameter ranges for reference only, and we can not guarantee that they are the optimal range of this model.

Then, with the source code of RecBole (you can download it from GitHub), you can run the run_hyper.py to tuning:

python run_hyper.py --model=[model_name] --dataset=[dataset_name] --config_files=[config_files_path] --params_file=hyper.test

If you want to change parameters, dataset or evaluation settings, take a look at