DGDATA: Deep Generative Domain Adaptation with Temporal Relation Attention Mechanism for Cross-User Activity Recognition

Authors

Xiaozhou Ye, Email: xye685@aucklanduni.ac.nz
Kevin Wang, Email: kevin.wang@aucklanduni.ac.nz

Affiliation

Department of Electrical, Computer, and Software Engineering, The University of Auckland

Overview

This repository contains the code and data for our paper: "Deep Generative Domain Adaptation with Temporal Relation Attention Mechanism for Cross-User Activity Recognition". The goal of this project is to enhance cross-user Human Activity Recognition (HAR) by integrating temporal dependency relations during domain adaptation.

Paper Link

[https://www.sciencedirect.com/science/article/pii/S0031320324005624]

Reference

@article{ye2024deep, title={Deep generative domain adaptation with temporal relation attention mechanism for cross-user activity recognition}, author={Ye, Xiaozhou and Kevin, I and Wang, Kai}, journal={Pattern Recognition}, pages={110811}, year={2024}, publisher={Elsevier} }

Setup Environment

Python 3.10.11

pip install -r requirements.txt

asttokens==2.4.1 colorama==0.4.6 comm==0.2.2 contourpy==1.2.0 cycler==0.12.1 decorator==5.1.1 exceptiongroup==1.2.1 executing==2.0.1 filelock==3.14.0 fonttools==4.50.0 fsspec==2024.5.0 giotto-ph==0.2.2 giotto-tda==0.6.0 igraph==0.11.5 intel-openmp==2021.4.0 ipython==8.24.0 ipywidgets==8.1.2 jedi==0.19.1 Jinja2==3.1.4 joblib==1.4.2 jupyterlab_widgets==3.0.10 kiwisolver==1.4.5 MarkupSafe==2.1.5 matplotlib==3.8.3 matplotlib-inline==0.1.7 mkl==2021.4.0 mpmath==1.3.0 networkx==3.3 numpy==1.26.4 packaging==24.0 pandas==2.2.1 parso==0.8.4 patsy==0.5.6 pillow==10.2.0 plotly==5.22.0 POT==0.9.3 prompt-toolkit==3.0.43 pure-eval==0.2.2 pyflagser==0.4.5 Pygments==2.18.0 pyparsing==3.1.2 python-dateutil==2.9.0.post0 pytz==2024.1 scikit-learn==1.4.2 scipy==1.13.0 six==1.16.0 stack-data==0.6.3 statsmodels==0.14.2 sympy==1.12 tbb==2021.12.0 tenacity==8.3.0 texttable==1.7.0 threadpoolctl==3.5.0 torch==2.3.0 traitlets==5.14.3 typing_extensions==4.11.0 tzdata==2024.1 wcwidth==0.2.13 widgetsnbextension==4.0.10

Data Preparation

The datasets used in this project are OPPT, PAMAP2, and DSADS. Download the datasets from their respective sources.

http://www.opportunity-project.eu/node/56.html

https://archive.ics.uci.edu/dataset/231/pamap2+physical+activity+monitoring

https://archive.ics.uci.edu/dataset/256/daily+and+sports+activities

File Overview

This repository contains various scripts and modules essential for training and running a deep generative temporal relation model with autoregression across different datasets (DSADS, OPPT, and PAMAP2).

Main Training Scripts

GPU_DSADS_deepGenTRNet_main_with_autoregression.py: Main script for training the model using the DSADS dataset.
GPU_OPPT_deepGenTRNet_main_with_autoregression.py: Main script for training the model using the OPPT dataset.
GPU_PAMAP2_deepGenTRNet_main_with_autoregression.py: Main script for training the model using the PAMAP2 dataset.

Utility Modules

utils.py: Contains helper functions for data processing and visualization.

Model Training and Definitions

gen_model/train.py: Script for the training phase of deep learning models.

Algorithms

gen_model/alg/DeepGenTempRelaNet.py: Defines the deep generative temporal relation model class.
gen_model/alg/linear_regression.py: Contains the definition and training function for a linear regression model.
gen_model/alg/modelopera.py: Utility module for various model operations.
gen_model/alg/opt.py: Includes functions for obtaining and setting the optimizer.

Loss Functions

gen_model/loss/common_loss.py: Implements various loss functions used within the models.

Network Models

gen_model/network/Adver_network.py: Defines a neural network model including a backpropagation function and a discriminator model.
gen_model/network/common_network.py: Contains definitions for common neural network models.
gen_model/network/feature_extraction_network.py: Definitions for feature extraction models in neural networks.

Utilities

gen_model/utils/util.py: Provides helper functions for logging, formatting, and setting random seeds.

Data process

OPPT_get_features_samples.py: reads the OPPT dataset, selects specific activities and sensor channels, and extracts data for a source user and a target user
PAMAP2_get_features_samples.py: reads the PAMAP2 dataset, selects specific activities and sensor channels, and extracts data for a source user and a target user
DSADS_get_features_samples.py: reads the DSADS dataset, selects specific activities and sensor channels, and extracts data for a source user and a target user
read_dataset/read_OPPT_dataset.py: read and process OPPT data. It initializes with parameters for users, window sizes, and sampling frequency, and includes methods to map activities, identify sensor channels, and segment data
read_dataset/read_PAMAP2_dataset.py: read and process PAMAP2 data. It initializes with parameters for users, window sizes, and sampling frequency, and includes methods to map activities, identify sensor channels, and segment data
read_dataset/read_DSADS_dataset.py: read and process DSADS data. It initializes with parameters for users, window sizes, and sampling frequency, and includes methods to map activities, identify sensor channels, and segment data

Hyperparameters

The hyperparameters used in the model can significantly impact performance. Here are the key hyperparameters and their default values:

Hyperparameter	Default Value
Training Epochs	100
Adam Optimizer Weight Decay	0.0005
Adam Optimizer Beta	0.2
Reconstruction Loss Coefficient (α)	1.0
Mean-Variance Loss Coefficient (ζ)	10.0
Class Constraint Loss Coefficient (γ)	30.0
Domain Constraint Loss Coefficient (δ)	1.0
Temporal State Constraint Loss Coefficient (η)	10.0

Basic Example

Here is a basic example to get you started with training the model on the OPPT dataset:

python GPU_OPPT_deepGenTRNet_main_with_autoregression.py

This script will load the OPPT dataset, initialize the model, and start training.

helloworld1973 / DGDATA