Generating Energy Data for Machine Learning with Recurrent Generative Adversarial Networks

Fekri, Mohammad Navid; Ghosh, Ananda Mohon; Grolinger, Katarina

doi:10.3390/en13010130

Cited by 92 publications

(47 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A recurrent GAN preceded by core features pre-processing was also used to test the same dataset. The model trained with the synthetic data achieved a similar accuracy as the one trained with real data [43]. In another work, a conditional variational auto-encoder was developed to detect electricity theft in buildings.…”

Section: Learning-based Methodsmentioning

confidence: 77%

An Approach to Data Acquisition for Urban Building Energy Modeling Using a Gaussian Mixture Model and Expectation-Maximization Algorithm

2021

View full text Add to dashboard Cite

In recent years, a building’s energy performance is becoming uncertain because of factors such as climate change, the Covid-19 pandemic, stochastic occupant behavior and inefficient building control systems. Sufficient measurement data is essential to predict and manage a building’s performance levels. Assessing energy performance of buildings at an urban scale requires even larger data samples in order to perform an accurate analysis at an aggregated level. However, data are not only expensive, but it can also be a real challenge for communities to acquire large amounts of real energy data. This is despite the fact that inadequate knowledge of a full population will lead to biased learning and the failure to establish a data pipeline. Thus, this paper proposes a Gaussian mixture model (GMM) with an Expectation-Maximization (EM) algorithm that will produce synthetic building energy data. This method is tested on real datasets. The results show that the parameter estimates from the model are stable and close to the true values. The bivariate model gives better performance in classification accuracy. Synthetic data points generated by the models show a consistent representation of the real data. The approach developed here can be useful for building simulations and optimizations with spatio-temporal mapping.

show abstract

Section: Learning-based Methodsmentioning

confidence: 77%

An Approach to Data Acquisition for Urban Building Energy Modeling Using a Gaussian Mixture Model and Expectation-Maximization Algorithm

2021

View full text Add to dashboard Cite

show abstract

“…Recently, GANs have been used as a method for generating realistic energy consumption data. Under this framework [33] proposes an algorithm of using GANs to sufficiently learn from a limited number of real data, whereas the work of [34] proposes a synthetic appliance power signature generator, called PowerGAN, to mitigate the data limitations arising from the insufficient labeled appliances power data.…”

Section: )mentioning

confidence: 99%

EnerGAN++: A Generative Adversarial Gated Recurrent Network for Robust Energy Disaggregation

Kaselimi

Doulamis

Voulodimos

et al. 2021

IEEE Open J. Signal Process.

View full text Add to dashboard Cite

“…2) Sliding Window: At this point, one data sample consists of several readings, potentially from different sensors and different locations, for the same time step t. For the time series data, the windows sliding technique is applied to help the model capture time-dependencies or to prepare data in the format needed by the ML algorithms [30]. In HAR, the sliding window has achieved great success [30]. This study investigates if features can be reduced after the sliding window technique is applied and examines the effect on the degree of data reduction.…”

Section: A Preprocessingmentioning

confidence: 99%

Edge-Cloud Computing for IoT Data Analytics: Embedding Intelligence in the Edge with Deep Learning

Ghosh

Grolinger

2020

IEEE Trans. Ind. Inf.

Self Cite

View full text Add to dashboard Cite

Rapid growth in numbers of connected devices including sensors, mobile, wearable, and other Internet of Things (IoT) devices, is creating an explosion of data that are moving across the network. To carry out machine learning (ML), IoT data are typically transferred to the cloud or another centralized system for storage and processing; however, this causes latencies and increases network traffic. Edge computing has the potential to remedy those issues by moving computation closer to the network edge and data sources. On the other hand, edge computing is limited in terms of computational power and thus is not well suited for ML tasks. Consequently, this paper aims to combine edge and cloud computing for IoT data analytics by taking advantage of edge nodes to reduce data transfer. In order to process data close to the source, sensors are grouped according to locations, and feature learning is performed on the close by edge node. For comparison reasons, similarity-based processing is also considered. Feature learning is carried out with deep learning: the encoder part of the trained autoencoder is placed on the edge and the decoder part is placed on the cloud. The evaluation was performed on the task of human activity recognition from sensor data. The results show that when sliding windows are used in the preparation step, data can be reduced on the edge up to 80% without significant loss in accuracy.

show abstract

Generating Energy Data for Machine Learning with Recurrent Generative Adversarial Networks

Cited by 92 publications

References 37 publications

An Approach to Data Acquisition for Urban Building Energy Modeling Using a Gaussian Mixture Model and Expectation-Maximization Algorithm

An Approach to Data Acquisition for Urban Building Energy Modeling Using a Gaussian Mixture Model and Expectation-Maximization Algorithm

EnerGAN++: A Generative Adversarial Gated Recurrent Network for Robust Energy Disaggregation

Edge-Cloud Computing for IoT Data Analytics: Embedding Intelligence in the Edge with Deep Learning

Contact Info

Product

Resources

About