Advances Toward the Next Generation Fire Detection: Deep LSTM Variational Autoencoder for Improved Sensitivity and Reliability

Xu, Zhaoyi; Guo, Yanjie; Saleh, Joseph H.

doi:10.1109/access.2021.3060338

Cited by 20 publications

(10 citation statements)

References 33 publications

(66 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The recent work [73] has proposed the usage of an unsupervised model, which is the LSTM based Variational Autoencoder, for fire detection and has presented the performance for each individual experiments in the data set that is considered in this paper. Since the results of this paper are obtained with the supervised models for fire detection, the comparison of LSTM based Variational Autoencoder against rTPNN is not fair.…”

Section: ) Long-short Term Memory (Lstm)mentioning

confidence: 99%

“…4 (In this comparison, the supervised models (including rTPNN) would be advantageous.) On the other hand, considering the success of LSTM based Variational Autoencoder for fire detection in [73] and that of LSTM for other problems on multivariate time series data, we use LSTM for performance comparison.…”

Section: ) Long-short Term Memory (Lstm)mentioning

confidence: 99%

“…In this way, LSTM-VAE does not require fire data (which is one of the most important properties of the unsupervised approach). However, such an unsupervised method detects fire later than the supervised models, where the recent work[73] has reported that time to alarm of LSTM-VAE is about 180 seconds on average for the experiments that are considered in this paper 5. The headers for these sensor data in the data set are TCB_1, SMB_1, GASB_1, GASB_3, GASB_5.…”

mentioning

confidence: 96%

See 2 more Smart Citations

Recurrent Trend Predictive Neural Network for Multi-Sensor Fire Detection

Nakıp

Güzeliş

Yıldız³

2021

IEEE Access

View full text Add to dashboard Cite

We propose a Recurrent Trend Predictive Neural Network (rTPNN) for multi-sensor fire detection based on the trend as well as level prediction and fusion of sensor readings. The rTPNN model significantly differs from the existing methods due to recurrent sensor data processing employed in its architecture. rTPNN performs trend prediction and level prediction for the time series of each sensor reading and captures trends on multivariate time series data produced by multi-sensor detector. We compare the performance of the rTPNN model with that of each of the Linear Regression (LR), Nonlinear Perceptron (NP), Multi-Layer Perceptron (MLP), Kendall-τ combined with MLP, Probabilistic Bayesian Neural Network (PBNN), Long-Short Term Memory (LSTM), and Support Vector Machine (SVM) on a publicly available fire data set. Our results show that rTPNN model significantly outperforms all of the other models (with 96% accuracy) while it is the only model that achieves high True Positive and True Negative rates (both above 92%) at the same time. rTPNN also triggers an alarm in only 11 s from the start of the fire, where this duration is 22 s for the second-best model. Moreover, we present that the execution time of rTPNN is acceptable for real-time applications.

show abstract

Section: ) Long-short Term Memory (Lstm)mentioning

confidence: 99%

Section: ) Long-short Term Memory (Lstm)mentioning

confidence: 99%

mentioning

confidence: 96%

See 1 more Smart Citation

Recurrent Trend Predictive Neural Network for Multi-Sensor Fire Detection

Nakıp

Güzeliş

Yıldız³

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…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.…”

Section: Introductionmentioning

confidence: 99%

“…Pack et al [10] also used a feed-forward neural network (FNN) and a convolutional neural network (CNN) to analyze high-dimensional multimodal data, including image and sensor signals. 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.…”

Section: Introductionmentioning

confidence: 99%

Uncertainty Assessment-Based Active Learning for Reliable Fire Detection Systems

Kim

2022

IEEE Access

View full text Add to dashboard Cite

Deep learning technologies, due to their advanced pattern extraction and recognition of highdimensional data, have been widely adopted into multisensor-based fire detection systems. Since deep learning approaches can generate erroneous predictions due to incomplete training datasets, a retraining process over unseen observations is needed. However, storing a large amount of data from continuous multisensor streams and labeling them to create a retraining dataset are costly and time-consuming. In this paper, we propose an active learning framework based on an informative experience memory that is populated with meaningful retraining data by assessing the uncertainty of the data. In the proposed framework, the deep learning model predicts fire occurrence and estimates model uncertainty by taking advantage of a Bayesian neural network using Monte Carlo dropout. By storing only higher uncertain data points into the fixed-size informative experience memory and querying them to the system managers, the storage and labeling costs are minimized while improving performance. To evaluate our active learning framework with different neural network structures, we develop three Bayesian neural networks based on conventional classification networks, including the feedforward neural network, fully convolutional network, and long short-term memory. We further investigate various uncertainty assessment scoring methods for classification tasks such as entropy, BALD, variation ratios, and mean STD. Experiments on a real dataset show that the Bayesian FCN using the BALD assessment method has the highest performance gain with an F1 score of 0.95, with an improvement of 24% using only 700 data points.

show abstract

Fire Detection Model Using Deep Learning Techniques

Pundir

Raman²

2022

Communications in Computer and Information Science

View full text Add to dashboard Cite

Advances Toward the Next Generation Fire Detection: Deep LSTM Variational Autoencoder for Improved Sensitivity and Reliability

Cited by 20 publications

References 33 publications

Recurrent Trend Predictive Neural Network for Multi-Sensor Fire Detection

Recurrent Trend Predictive Neural Network for Multi-Sensor Fire Detection

Uncertainty Assessment-Based Active Learning for Reliable Fire Detection Systems

Fire Detection Model Using Deep Learning Techniques

Contact Info

Product

Resources

About