Cross-correlation aided support vector machine classifier for classification of EEG signals

Chandaka, Suryannarayana; Chatterjee, Amitava; Munshi, Sugata

doi:10.1016/j.eswa.2007.11.017

Cited by 238 publications

(92 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…On the other hand, the fixed-point conversion causes a slight loss of precision due to rounding and saturation during the execution of the processing chain. The results obtained from the execution of the fixed-point application lead to an average loss of precision of 6.5% compared to the floating point version and to an average accuracy of 92% for the floating-point application and 89% from the fixed-point version, which is aligned with the state of the art [46][47][48]. Tables 3 and 4 summarize the execution time (clock cycles) of the seizure detection application on the PULP platform, in its floating-point and fixed-pint embodiment, respectively.…”

Section: Methodssupporting

confidence: 52%

Flexible, Scalable and Energy Efficient Bio-Signals Processing on the PULP Platform: A Case Study on Seizure Detection

Montagna

Benatti

Rossi

2017

JLPEA

View full text Add to dashboard Cite

Ultra-low power operation and extreme energy efficiency are strong requirements for a number of high-growth application areas requiring near-sensor processing, including elaboration of biosignals. Parallel near-threshold computing is emerging as an approach to achieve significant improvements in energy efficiency while overcoming the performance degradation typical of low-voltage operations. In this paper, we demonstrate the capabilities of the PULP (Parallel Ultra-Low Power) platform on an algorithm for seizure detection, representative of a wide range of EEG signal processing applications. Starting from the 28-nm FD-SOI (Fully Depleted Silicon On Insulator) technology implementation of the third embodiment of the PULP architecture, we analyze the energy-efficient implementation of the seizure detection algorithm on PULP. The proposed parallel implementation exploits the dynamic voltage and frequency scaling capabilities, as well as the embedded power knobs of the PULP platform, reducing energy consumption for a seizure detection by up to 10× with respect to a sequential implementation at the nominal supply voltage and by 4.2× with respect to a sequential implementation with voltage scaling. Moreover, we analyze the trans-precision optimization of the algorithm on PULP, by means of a hybrid fixed-and floating-point implementation. This approach reduces the energy consumption by up to 43% with respect to the plain fixed-point and floating-point implementations, leveraging the requirements in terms of the precision of the kernels composing the processing chain to improve energy efficiency. Thanks to the proposed architecture and system-level approach for optimization, we demonstrate that PULP reduces energy consumption by up to 140× with respect to commercial low-power microcontrollers, being able to satisfy the real-time constraints typical of bio-medical applications, breaking the barrier of microwatts for a 50-ms complete seizure detection and a few milliwatts for a 5-ms detection latency on a fully-programmable architecture.

show abstract

Section: Methodssupporting

confidence: 52%

Flexible, Scalable and Energy Efficient Bio-Signals Processing on the PULP Platform: A Case Study on Seizure Detection

Montagna

Benatti

Rossi

2017

JLPEA

View full text Add to dashboard Cite

show abstract

“…Their results showed that the most discriminative features for neonatal seizure detection 1 are morphological based features, such as amplitude, shape and duration of waveforms. In addition, time domain features such as statistical features (Adjouadi et al, 2005), Hjorth's descriptors (Hjorth, 1970), nonlinear features (Kannathal, Acharya, Lim, & Sadasivan, 2005;McSharry, et al, 2002)-correlation dimension (Elger & Lehnertz, 1998), Lyapunov exponent Ubeyli, 2006;Ubeyli, 2010b) and other features obtained from convolution kernels (Adjouadi et al, 2004), eigenvector methods (Naghsh-Nilchi & Aghashahi, 2010 ; Ubeyli, 2008aUbeyli, , 2008bUbeyli, , 2009a, principal component analysis (PCA) (Ghosh-Dastidar, Adeli, & Dadmehr, 2008;Hesse & James, 2007;James & Hesse, 2005;Polat & Gunes, 2008a;Subasi & Gursoy, 2010), ICA (Hesse & James, 2007;James & Hesse, 2005;Subasi & Gursoy, 2010), crosscorrelation function (Chandaka, Chatterjee, & Munshi, 2009;Iscan, et al, 2011), and entropy (Guo, Rivero, Dorado, et al, 2010;Kannathal, Choo, Acharya, & Sadasivan, 2005;Liang, Wang, & Chang, 2010;Naghsh-Nilchi & Aghashahi, 2010 ;H. Ocak, 2009;Srinivasan, Eswaran, & Sriraam, 2007;Wang, et al, 2011) have been proposed to characterize the EEG signal.…”

Section: Automated Epileptic Seizure Analysismentioning

confidence: 99%

“…The classification methods varied from simple threshold (Altunay, Telatar, & Erogul, 2010;Martinez-Vargas, et al, 2011), rule based decisions (Gotman, 1990(Gotman, , 1999, or linear classifiers (Ghosh-Dastidar, Iscan, et al, 2011;Liang, et al, 2010;Subasi & Gursoy, 2010) to ANNs , 2008Mousavi, et al, 2008;Nigam & Graupe, 2004;Srinivasan, et al, 2005Srinivasan, et al, , 2007Tzallas, et al, 2007aTzallas, et al, , 2007bTzallas, et al, , 2009Ubeyli, 2006Ubeyli, , 2009cUbeyli, 2010b) that have a complex shaped decision boundary. Other classification methods have been used using SVMs (Chandaka, et al, 2009;Iscan, et al, 2011;Liang, et al, 2010;Lima, et al, 2010;Subasi & Gursoy, 2010;Ubeyli, 2008a;Ubeyli, 2010a), k-nearest neighbour classifiers (Guo, et al, 2011;Iscan, et al, 2011;Liang, et al, 2010;Orhan, et al, 2011;Tzallas, et al, 2009), quadratic analysis (Iscan, et al, 2011), logistic regression Tzallas, et al, 2009), naive Bayes classifiers (Iscan, et al, 2011;Tzallas, et al, 2009), decision trees (Iscan, et al, 2011;Polat & Gunes, 2007;Tzallas, et al, 2009), Gaussian mixture model …”

Section: Automated Epileptic Seizure Analysismentioning

confidence: 99%

Automated Epileptic Seizure Detection Methods: A Review Study

Tzallas¹,

Tsipouras²,

Tsalikakis³

et al. 2012

Epilepsy - Histological, Electroencephalographic and Psychological Aspects

View full text Add to dashboard Cite

“…mu or beta-rhythm amplitudes) . A BCI could conceivably use both timedomain and frequency-domain signal features, and might thereby improve performance) (Schalk et al , 2000).…”

Section: Feature Extractionmentioning

confidence: 99%