Trustworthy Building Fire Detection Framework With Simulation-Based Learning

Abstract

With the difficulty of collecting desirable training data due to the heterogeneities of IoT sensors in various buildings and the scarcity of fire events, it is time consuming and expensive to apply data-driven deep learning approaches to fire detection systems in specific building environments. Simulation-based learning has been actively researched to mitigate data scarcity problems by reproducing potential fire events. Since simulation-based learning mainly depends on synthetic training data, trained deep learning models may generate erroneous predictions in real-world scenarios that are unlike any of the training samples. In this paper, we propose a trustworthy building fire detection framework based on a multioutput encoder-decoder network, named MEDNet, which is designed for the practical usage of simulation-based learning in building fire detection. The fundamental steps of our approach are (1) modeling and simulating fire events to create realistic synthetic data that reflect data from actual buildings, (2) predicting a fire event and dissimilarities between real input data and synthetic training data based on the trained MEDNet model, and (3) operating a switching mechanism to use a knowledge-based method that does not depend on synthetic training data when dissimilarities exist. Finally, we perform simulation experiments based on a real building compartment where the proposed framework is compared with conventional time-series classification networks on various evaluation datasets. The proposed framework is trustworthy in practical usage because MEDNet with a switching mechanism achieves a 36.65% higher F1-score than conventional time-series classification networks and generates false-positive predictions lower than 0.02% even in unpredictable scenarios.INDEX TERMS Simulation-based learning, deep learning, modeling and simulation, encoder-decoder network, supervised learning-based rare event detection, building fire detection.

Uncertainty Assessment-Based Active Learning for Reliable Fire Detection Systems

“…To achieve these goals, data-driven information fusion methods such as deep learning automatically process their decision-making by learning from training datasets. However, collecting all of the required training datasets in the development phase for reliable fire detection may not be trivial due to dynamic changes in environmental factors such as unpredictable human behaviors [3,12,15]. Moreover, according to [3], the exact reasons for most false alarms are unknown.…”

“…In recent years, more advanced data-driven information fusion methods have been proposed [8][9][10][11][12][13]. Among these data-driven approaches, deep learning (DL) has received significant attention due to its effective pattern extraction and recognition capabilities by training from raw data itself.…”

“…Xu et al [11] proposed deep long short-term memory (LSTM) networks and variational autoencoders (VAEs) to improve the sensitivity and reliability of fire detection. Kim et al [12] proposed a simulation-based learning framework for detecting fires. Then, the three types of encoder-decoder networks based on the FNN, fully convolutional network (FCN), and residual network (ResNet) were designed to provide trustworthy fire detection.…”

“…Especially in the field of fire detection, collecting training data corresponding to various fire and nonfire scenarios in certain environments is not trivial [14]. To overcome this data scarcity problem, several studies have utilized synthetic data generated from fire simulators [12]. However, such simulators only reproduce fire scenarios, not nonfire ones.…”

Classification in Early Fire Detection Using Multi-Sensor Nodes—A Transfer Learning Approach

“…Generally, fire events are infrequent occurrences throughout a building’s lifespan. The scarcity of real event data poses challenges and necessitates reliance on data obtained from experimental setups or simulations [ 22 ]. However, conducting such (large-scale) experiments is expensive, and the availability of large-scale test rooms is very limited [ 21 ].…”