The source code and dataset for the RecGURU paper (WSDM 2022)

Last update: Jan 07, 2023

Overview

RecGURU

About The Project

Source code and baselines for the RecGURU paper "RecGURU: Adversarial Learning of Generalized User Representations for Cross-Domain Recommendation (WSDM 2022)"

Code Structure

RecGURU  
├── README.md                                 Read me file 
├── data_process                              Data processing methods
│   ├── __init__.py                           Package initialization file     
│   └── amazon_csv.py                         Code for processing the amazon data (in .csv format)
│   └── business_process.py                   Code for processing the collected data
│   └── item_frequency.py                     Calculate item frequency in each domain
│   └── run.sh                                Shell script to perform data processing  
├── GURU                                      Scripts for modeling, training, and testing 
│   ├── data                                  Dataloader package      
│     ├── __init__.py                         Package initialization file 
│     ├── data_loader.py                      Customized dataloaders 
│   └── tools                                 Tools such as loss function, evaluation metrics, etc.
│     ├── __init__.py                         Package initialization file
│     ├── lossfunction.py                     Customized loss functions
│     ├── metrics.py                          Evaluation metrics
│     ├── plot.py                             Plot function
│     ├── utils.py                            Other tools
│  ├── Transformer                            Transformer package
│     ├── __init__.py                         Package initialization 
│     ├── transformer.py                      transformer module
│  ├── AutoEnc4Rec.py                         Autoencoder based sequential recommender
│  ├── AutoEnc4Rec_cross.py                   Cross-domain recommender modules
│  ├── config_auto4rec.py                     Model configuration file
│  ├── gan_training.py                        Training methods of the GAN framework
│  ├── train_auto.py                          Main function for training and testing single-domain sequential recommender
│  ├── train_gan.py                           Main function for training and testing cross-domain sequential recommender
└── .gitignore                                gitignore file

Dataset

The public datasets: Amazon view dataset at: https://nijianmo.github.io/amazon/index.html
Collected datasets: https://drive.google.com/file/d/1NbP48emGPr80nL49oeDtPDR3R8YEfn4J/view
Data processing:

Amazon dataset:

```shell
cd ../data_process
python amazon_csv.py   
```

Collected dataset

```shell
cd ../data_process
python business_process.py --rate 0.1  # portion of overlapping user = 0.1   
```

After data process, for each cross-domain scenario we have a dataset folder:

."a_domain"-"b_domain"
├── a_only.pickle         # users in domain a only
├── b_only.pickle         # users in domain b only
├── a.pickle              # all users in domain a
├── b.pickle              # all users in domain b
├── a_b.pickle            # overlapped users of domain a and b

Note: see the code for processing details and make modifications accordingly.

Run

Single-domain Methods:

# SAS
python train_auto.py --sas "True"
# AutoRec (ours)
python train_auto.py

Cross-Domain Methods:

# RecGURU
python train_gan.py --cross "True"

The source code and dataset for the RecGURU paper (WSDM 2022)

Related tags

Overview

RecGURU

About The Project

Code Structure

Dataset

Amazon dataset:

Collected dataset

Run

Owner

Chenglin Li

An image base contains 490 images for learning (400 cars and 90 boats), and another 21 images for testingAn image base contains 490 images for learning (400 cars and 90 boats), and another 21 images for testing

NVTabular is a feature engineering and preprocessing library for tabular data designed to quickly and easily manipulate terabyte scale datasets used to train deep learning based recommender systems.

PyTorch implementation of Federated Learning with Non-IID Data, and federated learning algorithms, including FedAvg, FedProx.

The official codes for the ICCV2021 Oral presentation "Rethinking Counting and Localization in Crowds: A Purely Point-Based Framework"

Baseline model for "GraspNet-1Billion: A Large-Scale Benchmark for General Object Grasping" (CVPR 2020)

PyTorch implementation of Super SloMo by Jiang et al.

PyTorch code for the paper "Curriculum Graph Co-Teaching for Multi-target Domain Adaptation" (CVPR2021)

Code for ViTAS_Vision Transformer Architecture Search

PyTorch implementation of the WarpedGANSpace: Finding non-linear RBF paths in GAN latent space (ICCV 2021)

Instance Segmentation in 3D Scenes using Semantic Superpoint Tree Networks

A collection of inference modules for fastai2

A Domain-Agnostic Benchmark for Self-Supervised Learning

Metrics to evaluate quality and efficacy of synthetic datasets.

Face2webtoon - Despite its importance, there are few previous works applying I2I translation to webtoon.

A PyTorch implementation of "TokenLearner: What Can 8 Learned Tokens Do for Images and Videos?"

Pun Detection and Location

Pytorch Implementation for NeurIPS (oral) paper: Pixel Level Cycle Association: A New Perspective for Domain Adaptive Semantic Segmentation

Automatic 2D-to-3D Video Conversion with CNNs

Train Dense Passage Retriever (DPR) with a single GPU

A novel benchmark dataset for Monocular Layout prediction