Probabilistic Rank Score Coding: A Robust Rank-Order Based Classifier for Electronic Nose Applications

Hassan, Muhammad; Belhaouari, Samir Brahim; Bermak, Amine

doi:10.1109/jsen.2015.2405296

Cited by 18 publications

(7 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This enables us to verify the classifier’s robustness to shuffling of spikes. The learning data consisting of six rank-order patterns were determined using the probabilistic rank-score coding, described in [17], which calculates the probability of a sensor spiking at a specific rank in the pattern for a target gas using Pij(k)=Nij(k)N(k) where Pij(k) is the probability of sensor i spiking at rank j for a target gas k , N ( k ) is the total number of rank-order patterns generated for target gas k , and Nij(k) is the number of times the sensor i in the array has spiked at rank j when exposed to target gas k . As a result, the unique learning set consisted of a reference rank-order signature for each target gas derived from this method.…”

Section: Resultsmentioning

confidence: 99%

“…These techniques require an entire frame of the rank order signatures to commence the pattern-matching process, resulting in a latency build-up in result classification. Furthermore, limitations of such an approach include susceptibility to inconsistent spiking order [17], and high computational and power requirements. Recent development in a neuro-inspired spike-pattern classifier was reported in [26], but its dependency on futuristic 3D technology to apply hyper-dimensional computing principles may not be viable for current artificial olfactory systems.…”

Section: Rank-order Encoding and Classifiers—overviewmentioning

confidence: 99%

“…Furthermore, a 4 × 4 metal-oxide sensor array design implemented resistance-to-time conversion in order to generate unique rank-order signatures for classification and identification of target gases. While the use of rank-order spiking patterns was not effective for encoding information, such as changing odor concentration, they have provided a simplified platform for detecting the presence of a target gas in feature space [5,17]. However, the utilization of traditional pattern-matching methods in the presence of shuffled spike sequences imposed limitations on the performance of the classifier [17].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Real-Time Classification of Multivariate Olfaction Data Using Spiking Neural Networks

Vanarse

Osseiran

Rassau

2019

Sensors

View full text Add to dashboard Cite

Recent studies in bioinspired artificial olfaction, especially those detailing the application of spike-based neuromorphic methods, have led to promising developments towards overcoming the limitations of traditional approaches, such as complexity in handling multivariate data, computational and power requirements, poor accuracy, and substantial delay for processing and classification of odors. Rank-order-based olfactory systems provide an interesting approach for detection of target gases by encoding multi-variate data generated by artificial olfactory systems into temporal signatures. However, the utilization of traditional pattern-matching methods and unpredictable shuffling of spikes in the rank-order impedes the performance of the system. In this paper, we present an SNN-based solution for the classification of rank-order spiking patterns to provide continuous recognition results in real-time. The SNN classifier is deployed on a neuromorphic hardware system that enables massively parallel and low-power processing on incoming rank-order patterns. Offline learning is used to store the reference rank-order patterns, and an inbuilt nearest neighbor classification logic is applied by the neurons to provide recognition results. The proposed system was evaluated using two different datasets including rank-order spiking data from previously established olfactory systems. The continuous classification that was achieved required a maximum of 12.82% of the total pattern frame to provide 96.5% accuracy in identifying corresponding target gases. Recognition results were obtained at a nominal processing latency of 16ms for each incoming spike. In addition to the clear advantages in terms of real-time operation and robustness to inconsistent rank-orders, the SNN classifier can also detect anomalies in rank-order patterns arising due to drift in sensing arrays.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Rank-order Encoding and Classifiers—overviewmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-Time Classification of Multivariate Olfaction Data Using Spiking Neural Networks

Vanarse

Osseiran

Rassau

2019

Sensors

View full text Add to dashboard Cite

show abstract

“…The Moreover, a comparison study has been performed in [48] PCA is used for preprocessing and GMM for classification reaching an accuracy of 92.7%. Authors in [49] presented an improved technique for EN based on the rank order (RO) by using probabilistic rank score coding (PRSC). The RO methods uses spikes that represent a unique signature for each gas.…”

Section: Accepted Manuscriptmentioning

confidence: 99%

An interactive software tool for gas identification

Djelouat

Ali

Amira

et al. 2018

Journal of Natural Gas Science and Engineering

View full text Add to dashboard Cite

This paper presents the design of an interactive graphical user interface (GUI) to monitor and quantify a developed electronic nose (EN) platform for gas identification. To this end, an EN system has been implemented using a multi-sensing embedded platform comprised of a data acquisition unit, an RFID module and a signal processing unit. The gas data are collected using two different types of gas sensors, namely, seven commercial Figaro sensors and in-house fabricated 4 × 4 tin-oxide gas array sensor. The collected gas data are processed for identification by means of dimensionality reduction algorithms and classification techniques where the software implementation and the quantification of these algorithms have been carried out. Subsequently, the GUI was designed to enable several operations. The GUI allows the user to visualize the sensors responses for any selected gas at any point of the acquisition process as well as visualizing the data distribution. Beside, it provides an easy approach to evaluate the EN system performance in terms of data identification and execution time by computing the classification accuracy using a 10-fold cross validation technique. Furthermore, the GUI, which is freely distributed, grants the users the privilege to upload other types of data to enable different pattern recognition applications.

show abstract

“…Probabilistic rank score coding is used in [25] to implement an improved EN which uses rank orders. In the rank order based technique, the gas signature is represented by temporal spikes.…”

Section: Related Workmentioning

confidence: 99%

Embedded Platform for Gas Applications Using Hardware/Software Co-Design and RFID

et al. 2018

Self Cite

View full text Add to dashboard Cite

This paper presents the development of a wireless low power reconfigurable self-calibrated multi-sensing platform for gas sensing applications. The proposed electronic nose (EN) system monitors gas temperatures, concentrations and mixtures wirelessly using the radio-frequency identification (RFID) technology. The EN takes the form of a set of gas and temperature sensors as well as multiple pattern recognition algorithms implemented on the Zynq system on chip (SoC) platform. The gas and temperature sensors are integrated on a semi-passive RFID tag to reduce the consumed power. Various gas sensors are tested including an in-house fabricated 4 × 4 SnO2 based sensor and 7 commercial Figaro sensors. The Data is transmitted to the Zynq based processing unit using a RFID reader where it is processed using multiple pattern recognition algorithms for dimensionality reduction and classification. Multiple algorithms are explored for optimum performance including principal component analysis (PCA) and linear discriminant analysis (LDA) for dimensionality reduction while decision tree (DT) and k-nearest neighbors (KNN) are assessed for classification purpose. Different gases are targeted at diverse concentration including carbon monoxide (CO), ethanol (C2H6O), carbon dioxide (CO2), propane (C3H8), ammonia (N H3) and hydrogen (H2). An accuracy of 100% is achieved in many cases with an overall accuracy above 90% in most scenarios. Finally, the hardware/software heterogeneous solution to implementation PCA, LDA, DT and KNN on the Zynq SoC shows promising results in terms of resources usage, power consumption and processing time.

show abstract

Probabilistic Rank Score Coding: A Robust Rank-Order Based Classifier for Electronic Nose Applications

Cited by 18 publications

References 35 publications

Real-Time Classification of Multivariate Olfaction Data Using Spiking Neural Networks

Real-Time Classification of Multivariate Olfaction Data Using Spiking Neural Networks

An interactive software tool for gas identification

Embedded Platform for Gas Applications Using Hardware/Software Co-Design and RFID

Contact Info

Product

Resources

About