Variational Autoencoder Generative Adversarial Network for Synthetic Data Generation in Smart Home

Razghandi, Mina; Zhang, Hao; Erol‐Kantarci, Melike; Turgut, Damla

doi:10.48550/arxiv.2201.07387

Cited by 4 publications

(5 citation statements)

References 15 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A GAN was used to learn the underlying physical model that affects the internal relationships of data. While many studies adopted GANs for data generation, Razghandi et al [20] proposed a model combining a VAE and a GAN to generate smart home synthetic time series data. The model learns the distribution of various data types in a smart home without prior knowledge and generates plausible samples.…”

Section: Synthetic Data Generationmentioning

confidence: 99%

Data-Driven ICS Network Simulation for Synthetic Data Generation

Kim,

Jeon,

Cho

et al. 2024

Electronics

View full text Add to dashboard Cite

Industrial control systems (ICSs) are integral to managing and optimizing processes in various industries, including manufacturing, power generation, and more. However, the scarcity of widely adopted ICS datasets hampers research efforts in areas like optimization and security. This scarcity arises due to the substantial cost and technical expertise required to create physical ICS environments. In response to these challenges, this paper presents a groundbreaking approach to generating synthetic ICS data through a data-driven ICS network simulation. We circumvent the need for expensive hardware by recreating the entire ICS environment in software. Moreover, rather than manually replicating the control logic of ICS components, we leverage existing data to autonomously generate control logic. The core of our method involves the stochastic setting of setpoints, which introduces randomness into the generated data. Setpoints serve as target values for controlling the operation of the ICS process. This approach enables us to augment existing ICS datasets and cater to the data requirements of machine learning-based ICS intrusion detection systems and other data-driven applications. Our simulated ICS environment employs virtualized containers to mimic the behavior of real-world PLCs and SCADA systems, while control logic is deduced from publicly available ICS datasets. Setpoints are generated probabilistically to ensure data diversity. Experimental results validate the fidelity of our synthetic data, emphasizing their ability to closely replicate temporal and statistical characteristics of real-world ICS networks. In conclusion, this innovative data-driven ICS network simulation offers a cost-effective and scalable solution for generating synthetic ICS data. It empowers researchers in the field of ICS optimization and security with diverse, realistic datasets, furthering advancements in this critical domain. Future work may involve refining the simulation model and exploring additional applications for synthetic ICS data.

show abstract

Section: Synthetic Data Generationmentioning

confidence: 99%

Data-Driven ICS Network Simulation for Synthetic Data Generation

Kim,

Jeon,

Cho

et al. 2024

Electronics

View full text Add to dashboard Cite

show abstract

“…Data synthesis, a more advanced data augmentation method [ 39 ], shows promise and is the subject of much interest. Among the many approaches for creating synthetic data, variational autoencoders (VAEs) [ 40 ] and generative adversarial networks (GANs) [ 41 ] stand out as the most common. The latent representation of VAEs is highly structured and continuous, making them simple to train.…”

Section: Related Workmentioning

confidence: 99%

Skin Lesion Synthesis and Classification Using an Improved DCGAN Classifier

2023

View full text Add to dashboard Cite

The prognosis for patients with skin cancer improves with regular screening and checkups. Unfortunately, many people with skin cancer do not receive a diagnosis until the disease has advanced beyond the point of effective therapy. Early detection is critical, and automated diagnostic technologies like dermoscopy, an imaging device that detects skin lesions early in the disease, are a driving factor. The lack of annotated data and class-imbalance datasets makes using automated diagnostic methods challenging for skin lesion classification. In recent years, deep learning models have performed well in medical diagnosis. Unfortunately, such models require a substantial amount of annotated data for training. Applying a data augmentation method based on generative adversarial networks (GANs) to classify skin lesions is a plausible solution by generating synthetic images to address the problem. This article proposes a skin lesion synthesis and classification model based on an Improved Deep Convolutional Generative Adversarial Network (DCGAN). The proposed system generates realistic images using several convolutional neural networks, making training easier. Scaling, normalization, sharpening, color transformation, and median filters enhance image details during training. The proposed model uses generator and discriminator networks, global average pooling with 2 × 2 fractional-stride, backpropagation with a constant learning rate of 0.01 instead of 0.0002, and the most effective hyperparameters for optimization to efficiently generate high-quality synthetic skin lesion images. As for the classification, the final layer of the Discriminator is labeled as a classifier for predicting the target class. This study deals with a binary classification predicting two classes—benign and malignant—in the ISIC2017 dataset: accuracy, recall, precision, and F1-score model classification performance. BAS measures classifier accuracy on imbalanced datasets. The DCGAN Classifier model demonstrated superior performance with a notable accuracy of 99.38% and 99% for recall, precision, F1 score, and BAS, outperforming the state-of-the-art deep learning models. These results show that the DCGAN Classifier can generate high-quality skin lesion images and accurately classify them, making it a promising tool for deep learning-based medical image analysis.

show abstract

“…D minimizes L D , equation (12), in the training process, which has three terms. L real , equation ( 9), is the likelihood of original input data being classified as fake data, L f ake , equation (10), is the likelihood of reconstructed data sample being classified as real, and L noise , equation (11), is the likelihood of classifying a random noise input as a real data which is added to L D to improve the convergence of the discriminator.…”

Section: Data-driven Generative Modelmentioning

confidence: 99%

“…In our recent paper [10], our approach involved generating electric load profiles and PV generation synthetic data for a smart home using variational autoencoder-generative adversarial networks (VAE-GAN), and we compared the distributions of synthetic data for the VAE-GAN model and vanilla GAN network to the real data distributions. According to the statistical metrics such as KL divergence, Wasserstein distance, and maximum mean discrepancy, the distribution of data points was extremely close between synthetic data and real data for both PV power generation and energy consumption load profiles.…”

Section: Introductionmentioning

confidence: 99%

Smart Home Energy Management: Sequence-to-Sequence Load Forecasting and Q-Learning

Razghandi

Zhou

Erol‐Kantarci

et al. 2021

2021 IEEE Global Communications Conference (GLOBECOM)

View full text Add to dashboard Cite

Recent years have noticed an increasing interest among academia and industry towards analyzing the electrical consumption of residential buildings and employing smart home energy management systems (HEMS) to reduce household energy consumption and costs. HEMS have been developed to simulate the statistical and functional properties of actual smart grids. Access to publicly available datasets is a major challenge in this type of research. The potential of artificial HEMS applications will be further enhanced with the development of time series that represent different operating conditions of the synthetic systems.In this paper, we propose a novel variational auto-encodergenerative adversarial network (VAE-GAN) technique for generating time-series data on energy consumption in smart homes. We also explore how the generative model performs when combined with a Q-learning-based HEMS. We tested the online performance of Q-learning-based HEMS with real-world smart home data. To test the generated dataset, we measure the Kullback-Leibler (KL) divergence, maximum mean discrepancy (MMD), and the Wasserstein distance between the probability distributions of the real and synthetic data. Our experiments show that VAE-GAN-generated synthetic data closely matches the real data distribution. Finally, we show that the generated data allows for the training of a higher-performance Q-learning-based HEMS compared to datasets generated with baseline approaches.

show abstract

Variational Autoencoder Generative Adversarial Network for Synthetic Data Generation in Smart Home

Cited by 4 publications

References 15 publications

Data-Driven ICS Network Simulation for Synthetic Data Generation

Data-Driven ICS Network Simulation for Synthetic Data Generation

Skin Lesion Synthesis and Classification Using an Improved DCGAN Classifier

Smart Home Energy Management: Sequence-to-Sequence Load Forecasting and Q-Learning

Contact Info

Product

Resources

About