Deep Neuroevolution: Training Deep Neural Networks for False Alarm Detection in Intensive Care Units

Hooman, Oroojeni M. J.; al-Rifaie, Mohammad Majid; Nicolaou, Mihalis A.

doi:10.23919/eusipco.2018.8552944

Cited by 14 publications

(8 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It can be seen the proposed approach leads to a state of the art with a considerable margin. These results are obtained while we just used one lead of information but the most approaches are based on three leads [62]- [66], [68], we do not use expert knowledge about the rules while some approaches are based on that [62], [64], [66], and we do not use the expert knowledge about the possible place of arrhythmia while some of the methods rely on this information [62]- [66], [68]. Results on 2017 PhysioNet computing in Cardiology Dataset: We show in Table.…”

Section: Resultsmentioning

confidence: 99%

“…The proposed method in [68] used a deep neuroevolution method that utilized genetic algorithms and [69] used neural networks for arrhythmia classification. The former method is based on using handcrafted features introduced by [70], which includes morphological and frequency features extracted from the ECG, ABP, and PPG signals.…”

Section: A Review Of Related Work For False Alarm Reduction In Icusmentioning

confidence: 99%

See 1 more Smart Citation

Piece-wise Matching Layer in Representation Learning for ECG Classification

Ghazanfari¹,

Afghah²,

Zhang³

2020

Preprint

View full text Add to dashboard Cite

This paper proposes piece-wise matching layer as a novel layer in representation learning methods for electrocardiogram (ECG) classification. Despite the remarkable performance of representation learning methods in the analysis of time series, there are still several challenges associated with these methods ranging from the complex structures of methods, the lack of generality of solutions, the need for expert knowledge, and large-scale training datasets. We introduce the piece-wise matching layer that works based on two levels to address some of the aforementioned challenges. At the first level, a set of morphological, statistical, and frequency features and comparative forms of them are computed based on each periodic part and its neighbors. At the second level, these features are modified by predefined transformation functions based on a receptive field scenario. Several scenarios of offline processing, incremental processing, fixed sliding receptive field, and event-based triggering receptive field can be implemented based on the choice of length and mechanism of indicating the receptive field. We propose dynamic time wrapping as a mechanism that indicates a receptive field based on event triggering tactics. The proposed layer outperforms the common kernels and pooling on high-level representations of ECG classification. This layer offers several advantages including 1) it enables the learning from few instances for training and handling long quasi-periodic time series, 2) the layer followed by a simultaneous multiple feature tracking mechanism that does not require any annotated data or expert knowledge in quasi-periodic time series, and 3) the proposed framework can provide and handle several scenarios like online processing and dynamic data-driven application systems. To evaluate the performance of this method in time series analysis, we applied the proposed layer in two publicly available datasets of PhysioNet competitions in 2015 and 2017 where the input data is ECG signal. These datasets present a good example of quasiperiodic time series signals with a small number of training samples, highly imbalanced classes, long time series recordings, and non-linear and complex patterns. We compared the performance of our method against a variety of known tuned methods from expert knowledge, machine learning, deep learning methods, and the combination of them. The proposed approach improves the state of the art in two known completions 2015 and 2017 around 4% and 7% correspondingly while it does not rely on in advance knowledge of the classes or the possible places of arrhythmia.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: A Review Of Related Work For False Alarm Reduction In Icusmentioning

confidence: 99%

Piece-wise Matching Layer in Representation Learning for ECG Classification

Ghazanfari¹,

Afghah²,

Zhang³

2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Dispersive flies optimization, introduced in 2014 [19], is inspired from two behaviours of flies: their swarming behaviour when they find a food source and their retreating and dispersing behaviour when encountered a threat. It has been employed in several discrete and continuous search spaces problems in the domain of medical imaging [20], training of deep neural network [21], optimization of machine learning algorithms [22]. DFO's implementation in optimization problem is illustrated by the pseudo-code.…”

Section: Dispersive Flies Optimizationmentioning

confidence: 99%

PID speed control of DC motor using meta-heuristic algorithms

Acharya

Dhakal

Bhattarai

et al. 2021

IJPEDS

View full text Add to dashboard Cite

This paper presents archimedes optimization algorithm(AOA) and dispersive flies optimization(DFO) to optimally tune gain parameters of PID control scheme in order to regulate DC motor’s speed. These suggested techniques tune the controller by the minimization of the fitness function represented by the integral of time multiplied by absolute error (ITAE). The modelling and simulation are carried out in MATLAB/Simulink. The transient response of unit step input obtained from AOA-PID-ITAE andDFO-PID-ITAE controllers were compared to those obtained from Ziegler-Nichols (ZN) method and particle swarm optimization(PSO). The results indicate that AOA-PID-ITAE and DFO-PID-ITAE are more efficient than ZN method and PSO in reducing rise time and settling time. Likewise, DFOconverge faster to the optimal solution with lower overshoot than AOA and PSO.

show abstract

“…In recent years, EAs performed much better than traditional optimizing method such as gradient descend in various domains [31], [32]. This advantage of EA is becoming more important in deep neural network because of the diversity and complexity it provides [33], [34]. Martní et al [35] created an Android malware detection system with evolutionary strategies to leverage third-party calls to bypass the effects of concealment strategies.…”

Section: Evolutionary Algorithmmentioning

confidence: 99%

MDEA: Malware Detection with Evolutionary Adversarial Learning

Wang

Miikkulainen

2020

Preprint

View full text Add to dashboard Cite

Malware detection have used machine learning to detect malware in programs. These applications take in raw or processed binary data to neural network models to classify as benign or malicious files. Even though this approach has proven effective against dynamic changes, such as encrypting, obfuscating and packing techniques, it is vulnerable to specific evasion attacks where that small changes in the input data cause misclassification at test time. This paper proposes a new approach: MDEA, an Adversarial Malware Detection model uses evolutionary optimization to create attack samples to make the network robust against evasion attacks. By retraining the model with the evolved malware samples, its performance improves a significant margin.

show abstract

Deep Neuroevolution: Training Deep Neural Networks for False Alarm Detection in Intensive Care Units

Cited by 14 publications

References 26 publications

Piece-wise Matching Layer in Representation Learning for ECG Classification

Piece-wise Matching Layer in Representation Learning for ECG Classification

PID speed control of DC motor using meta-heuristic algorithms

MDEA: Malware Detection with Evolutionary Adversarial Learning

Contact Info

Product

Resources

About