Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679636
A. Rege, B. Tyler, M. J. Brooke, K. Murari
We present a memory-optimized, portable hardware implementation for temporal laser speckle imaging (tLSI) consisting of a complementary metal oxide semiconductor (CMOS) image sensor, a field programmable gate array (FPGA) processor and a microSD card for storage. The system is compared to a typical benchtop setup consisting of a charge coupled device (CCD) based image sensor and a computer for imaging rat brain vasculature. The hardware implementation enables faster operation, reduced data bandwidth, lower power consumption, smaller footprint and comparable performance in terms of contrast to noise ratio and flow-fidelity. We expect these advantages to enable portable, point-of-care diagnostic imaging devices for high-resolution, high-contrast measurement of haemodynamic parameters.
{"title":"A portable hardware implementation for temporal laser speckle imaging","authors":"A. Rege, B. Tyler, M. J. Brooke, K. Murari","doi":"10.1109/BioCAS.2013.6679636","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679636","url":null,"abstract":"We present a memory-optimized, portable hardware implementation for temporal laser speckle imaging (tLSI) consisting of a complementary metal oxide semiconductor (CMOS) image sensor, a field programmable gate array (FPGA) processor and a microSD card for storage. The system is compared to a typical benchtop setup consisting of a charge coupled device (CCD) based image sensor and a computer for imaging rat brain vasculature. The hardware implementation enables faster operation, reduced data bandwidth, lower power consumption, smaller footprint and comparable performance in terms of contrast to noise ratio and flow-fidelity. We expect these advantages to enable portable, point-of-care diagnostic imaging devices for high-resolution, high-contrast measurement of haemodynamic parameters.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"49 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114309659","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679689
S. Imtiaz, Siavash Saremi-Yarahmadi, E. Rodríguez-Villegas
Sleep spindles are the hallmark of N2 stage of sleep. They are transient waveforms observed on sleep electroencephalogram and their identification is required for sleep staging. Due to the large number of sleep spindles appearing on an overnight sleep EEG, automating the detection of sleep spindles would be desirable, not only to save specialist time but also for fully automated sleep staging systems. A simple algorithm for automatic sleep spindle detection is presented in this paper using only one channel of EEG input. This algorithm uses Teager energy and spectral edge frequency to mark sleep spindles and results in a sensitivity of 80% and specificity of about 98%. It is also shown that more than 91% of spindles detected by the algorithm were in N2 and N3 stages combined.
{"title":"Automatic detection of sleep spindles using Teager energy and spectral edge frequency","authors":"S. Imtiaz, Siavash Saremi-Yarahmadi, E. Rodríguez-Villegas","doi":"10.1109/BioCAS.2013.6679689","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679689","url":null,"abstract":"Sleep spindles are the hallmark of N2 stage of sleep. They are transient waveforms observed on sleep electroencephalogram and their identification is required for sleep staging. Due to the large number of sleep spindles appearing on an overnight sleep EEG, automating the detection of sleep spindles would be desirable, not only to save specialist time but also for fully automated sleep staging systems. A simple algorithm for automatic sleep spindle detection is presented in this paper using only one channel of EEG input. This algorithm uses Teager energy and spectral edge frequency to mark sleep spindles and results in a sensitivity of 80% and specificity of about 98%. It is also shown that more than 91% of spindles detected by the algorithm were in N2 and N3 stages combined.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114332129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679670
Alexandra-Maria Tăuțan, W. Serdijn, V. Mihajlović, B. Grundlehner, J. Penders
Dry electrodes provide the possibility of moving EEG usage from the research and clinical environment to real life applications. Having a framework for evaluating the performance of dry electrodes would facilitate this process and help EEG system developers to test their designs. This paper describes an evaluation method for dry electrode EEG recordings. The framework includes a setup for synchronous recordings with a parallel dry and gel electrode montage procedure. Several protocols are implemented to evaluate both the time and frequency content of the signal and to compute the setup and settling time. Signal quality was evaluated using signal correlations, SNR and P300 component characteristics. The preliminary data analysis and results show that a comparison between gel and dry electrodes is possible but improvements need to be made to the current evaluation framework.
{"title":"Framework for evaluating EEG signal quality of dry electrode recordings","authors":"Alexandra-Maria Tăuțan, W. Serdijn, V. Mihajlović, B. Grundlehner, J. Penders","doi":"10.1109/BioCAS.2013.6679670","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679670","url":null,"abstract":"Dry electrodes provide the possibility of moving EEG usage from the research and clinical environment to real life applications. Having a framework for evaluating the performance of dry electrodes would facilitate this process and help EEG system developers to test their designs. This paper describes an evaluation method for dry electrode EEG recordings. The framework includes a setup for synchronous recordings with a parallel dry and gel electrode montage procedure. Several protocols are implemented to evaluate both the time and frequency content of the signal and to compute the setup and settling time. Signal quality was evaluated using signal correlations, SNR and P300 component characteristics. The preliminary data analysis and results show that a comparison between gel and dry electrodes is possible but improvements need to be made to the current evaluation framework.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"155 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132492235","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679630
Bo-Yu Shiu, Shuohan Wang, Y. Chu, T. Tsai
In this paper, we implemented a low-power low-noise ECG acquisition system for heart disease identifications. The ECG signal is acquired with an analog front-end circuit, and the offset and baseline drift is eliminated at the same time. We also implemented a digital signal processing (DSP) unit to effectively remove EMG interferences. Interception of ST segment is proposed to achieve the identification of heart diseases. The ECG front-end chip has been designed and fabricated by using a TSMC 90nm CMOS technology, the total power consumption was measured at 40.3μW. DSP algorithms are carried out in the FPGA. Experimental results show that the sensitivity and specificity of ST segment classification after EMG elimination is 96.6% and 93.1%, respectively.
{"title":"Low-power low-noise ECG acquisition system with dsp for heart disease identification","authors":"Bo-Yu Shiu, Shuohan Wang, Y. Chu, T. Tsai","doi":"10.1109/BioCAS.2013.6679630","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679630","url":null,"abstract":"In this paper, we implemented a low-power low-noise ECG acquisition system for heart disease identifications. The ECG signal is acquired with an analog front-end circuit, and the offset and baseline drift is eliminated at the same time. We also implemented a digital signal processing (DSP) unit to effectively remove EMG interferences. Interception of ST segment is proposed to achieve the identification of heart diseases. The ECG front-end chip has been designed and fabricated by using a TSMC 90nm CMOS technology, the total power consumption was measured at 40.3μW. DSP algorithms are carried out in the FPGA. Experimental results show that the sensitivity and specificity of ST segment classification after EMG elimination is 96.6% and 93.1%, respectively.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132811220","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679679
J. Besnoff, M. Reynolds
We consider the use of insulated single wires as transmission lines to carry 100 MHz - 3 GHz radio frequency (RF) signals among devices implanted in biological tissue. In contrast to near-field magnetically coupled links, the use of transmission lines to carry RF signals results in higher efficiency for a given implant package size once the antenna is included, albeit with the disadvantage of tissue displacement along the path of the wire. We present a theory based on the work of Goubau and Rao that describes the transmission line loss of a single insulated wire in a lossy dielectric medium. We experimentally verify the characteristic impedance and insertion loss of transmission lines formed by thin wires insulated with Teflon fluorinated ethylene propylene (FEP). We consider media including 0.91% saline (a homogeneous tissue proxy), muscle tissue, and brain tissue, and present a launcher design based on a dielectric loaded coaxial sleeve. For example, in the saline proxy, a single FEP-insulated conductor of only 0.127 mm diameter presents a measured return loss of 10 dB in a 50Ω system, with a measured insertion loss of only 1 dB/cm at 1 GHz.
{"title":"Single-wire RF transmission lines for implanted devices","authors":"J. Besnoff, M. Reynolds","doi":"10.1109/BioCAS.2013.6679679","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679679","url":null,"abstract":"We consider the use of insulated single wires as transmission lines to carry 100 MHz - 3 GHz radio frequency (RF) signals among devices implanted in biological tissue. In contrast to near-field magnetically coupled links, the use of transmission lines to carry RF signals results in higher efficiency for a given implant package size once the antenna is included, albeit with the disadvantage of tissue displacement along the path of the wire. We present a theory based on the work of Goubau and Rao that describes the transmission line loss of a single insulated wire in a lossy dielectric medium. We experimentally verify the characteristic impedance and insertion loss of transmission lines formed by thin wires insulated with Teflon fluorinated ethylene propylene (FEP). We consider media including 0.91% saline (a homogeneous tissue proxy), muscle tissue, and brain tissue, and present a launcher design based on a dielectric loaded coaxial sleeve. For example, in the saline proxy, a single FEP-insulated conductor of only 0.127 mm diameter presents a measured return loss of 10 dB in a 50Ω system, with a measured insertion loss of only 1 dB/cm at 1 GHz.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134002856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BIOCAS.2013.6679710
Tianyi Liu, J. Anders, M. Ortmanns
This paper presents a design methodology for an optimization of the optical receiver in a neural recording system. In the transcutaneous optical telemetric link (TOTL) system, light/tissue interaction produces significant scattering of the transmitted photons. Therefore, in order to increase the transmission efficiency, it is in principle advantageous to use a photodiode with a larger size to collect more scattered photons. The photodiode size, however, is limited by the bandwidth requirement. In this paper, we derive a mathematical expression of the maximum photodiode size in the presence of constraints imposed by its bias voltage and the receiver bandwidth. A numerical example using real-world TOTL system parameters is given to demonstrate the design methodology. The trade off between transmission efficiency, noise and inter-symbol interference (ISI) in terms of the 3-dB bandwidth of the receiver is investigated. To this end, receivers with bandwidth about 4/3, 3/3, 2/3 and 1/3 of the data rate are compared with respect to the sensitivity. It was found that the receiver bandwidth of 2/3 of the data rate achieves the highest sensitivity for thin tissue (2 mm) and that a receiver bandwidth of 1/3 of the data rate is preferable for thicker tissues (5 and 8 mm).
{"title":"Design optimization of the optical receiver in transcutaneous telemetric links","authors":"Tianyi Liu, J. Anders, M. Ortmanns","doi":"10.1109/BIOCAS.2013.6679710","DOIUrl":"https://doi.org/10.1109/BIOCAS.2013.6679710","url":null,"abstract":"This paper presents a design methodology for an optimization of the optical receiver in a neural recording system. In the transcutaneous optical telemetric link (TOTL) system, light/tissue interaction produces significant scattering of the transmitted photons. Therefore, in order to increase the transmission efficiency, it is in principle advantageous to use a photodiode with a larger size to collect more scattered photons. The photodiode size, however, is limited by the bandwidth requirement. In this paper, we derive a mathematical expression of the maximum photodiode size in the presence of constraints imposed by its bias voltage and the receiver bandwidth. A numerical example using real-world TOTL system parameters is given to demonstrate the design methodology. The trade off between transmission efficiency, noise and inter-symbol interference (ISI) in terms of the 3-dB bandwidth of the receiver is investigated. To this end, receivers with bandwidth about 4/3, 3/3, 2/3 and 1/3 of the data rate are compared with respect to the sensitivity. It was found that the receiver bandwidth of 2/3 of the data rate achieves the highest sensitivity for thin tissue (2 mm) and that a receiver bandwidth of 1/3 of the data rate is preferable for thicker tissues (5 and 8 mm).","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"47 5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134344950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679640
Go Nakamura, T. Shibanoki, K. Shima, Y. Kurita, Masaki Hasegawa, A. Otsuka, Y. Honda, T. Chin, T. Tsuji
This paper proposes a novel EMG-based MyoBock training system that consistently provides a variety of functions ranging from EMG signal control training to task training. Using the proposed training sytem, a trainee controls a virtual hand (VH) in a 3D virtual reality (VR) environment using EMG signals and position/posture information recorded from the trainee. The trainee can also perform tasks such as holding and moving virtual objects using the system. In the experiments of this study, virtual task training developed with reference to the Box and Block Test (BBT) used to evaluate myoelectric prostheses was conducted with two healthy subjects, who repeatedly performed 10 one-minute tasks involving grasping a ball in one box and transporting it to another. The BBT experiments were also conducted in a real environment before and after the virtual training, with results showing an improvement in the number of tasks successfully completed. It was therefore confirmed that the proposed system could be used for myoelectric prosthesis control training.
{"title":"A training system for the MyoBock hand in a virtual reality environment","authors":"Go Nakamura, T. Shibanoki, K. Shima, Y. Kurita, Masaki Hasegawa, A. Otsuka, Y. Honda, T. Chin, T. Tsuji","doi":"10.1109/BioCAS.2013.6679640","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679640","url":null,"abstract":"This paper proposes a novel EMG-based MyoBock training system that consistently provides a variety of functions ranging from EMG signal control training to task training. Using the proposed training sytem, a trainee controls a virtual hand (VH) in a 3D virtual reality (VR) environment using EMG signals and position/posture information recorded from the trainee. The trainee can also perform tasks such as holding and moving virtual objects using the system. In the experiments of this study, virtual task training developed with reference to the Box and Block Test (BBT) used to evaluate myoelectric prostheses was conducted with two healthy subjects, who repeatedly performed 10 one-minute tasks involving grasping a ball in one box and transporting it to another. The BBT experiments were also conducted in a real environment before and after the virtual training, with results showing an improvement in the number of tasks successfully completed. It was therefore confirmed that the proposed system could be used for myoelectric prosthesis control training.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132042536","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BioCAS.2013.6679663
V. Cambareri, Mauro Mangia, R. Rovatti, G. Setti
In this work, we propose an analog-to-information conversion based sensing protocol for the joint acquisition of signals in a typical body sensor network composed of different kinds of sensing devices. The proposed system is based on the compressive sensing paradigm, and jointly encodes sparse and non-sparse signals with an advantage in terms of required transmission bits and achieved signal-to-noise ratio with respect to the standard case of disjoint signal acquisition. A theoretical analysis is supported by some practical examples which show how the proposed system is able to outperform traditional approaches.
{"title":"Joint analog-to-information conversion of heterogeneous biosignals","authors":"V. Cambareri, Mauro Mangia, R. Rovatti, G. Setti","doi":"10.1109/BioCAS.2013.6679663","DOIUrl":"https://doi.org/10.1109/BioCAS.2013.6679663","url":null,"abstract":"In this work, we propose an analog-to-information conversion based sensing protocol for the joint acquisition of signals in a typical body sensor network composed of different kinds of sensing devices. The proposed system is based on the compressive sensing paradigm, and jointly encodes sparse and non-sparse signals with an advantage in terms of required transmission bits and achieved signal-to-noise ratio with respect to the standard case of disjoint signal acquisition. A theoretical analysis is supported by some practical examples which show how the proposed system is able to outperform traditional approaches.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132136621","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BIOCAS.2013.6679656
Ismael Rattalino, P. Ros, I. Taurino, F. Cortés-Salazar, G. Piccinini, D. Demarchi, G. Micheli, S. Carrara
Although non-enzymatic glucose sensors have demonstrated better stability and reproducibility with respect to enzymatic ones, so far they have been inappropriate for most applications, since they require alkaline conditions to achieve the necessary sensitivity. In this work, we propose a gold nanogap-based non-enzymatic sensor to localize the generation of alkaline conditions inside the gap, thus preserving the overall pH in the media during glucose detection. The working principle is based on an electrochemical bi-potentiostatic measurement, where an alkaline aqueous condition is locally generated at one side of nanogap, while glucose detection is performed at the counterpart. To this purpose, a nanogap array platform was fabricated by means of standard lithography and controlled electromigration. Mono-potentiostatic electrochemical detection of ascorbic acid was successfully performed to preliminary test the platform prior to measuring glucose in bi-potentiostatic mode. Cyclic voltammetries reveal that two oxidation peaks are sensitive to glucose concentration, making nanogap glucose detection possible in principle. This promising proof of concept could be innovative in bio-applications with implantable devices or direct monitoring of cell culture, where neutral pH in contact with living tissue is required. Further geometrical improvements of the system to increase the durability of the sensor are currently still in progress, and are briefly discussed in the final part of the paper.
{"title":"Nanogap-based enzymatic-free electrochemical detection of glucose","authors":"Ismael Rattalino, P. Ros, I. Taurino, F. Cortés-Salazar, G. Piccinini, D. Demarchi, G. Micheli, S. Carrara","doi":"10.1109/BIOCAS.2013.6679656","DOIUrl":"https://doi.org/10.1109/BIOCAS.2013.6679656","url":null,"abstract":"Although non-enzymatic glucose sensors have demonstrated better stability and reproducibility with respect to enzymatic ones, so far they have been inappropriate for most applications, since they require alkaline conditions to achieve the necessary sensitivity. In this work, we propose a gold nanogap-based non-enzymatic sensor to localize the generation of alkaline conditions inside the gap, thus preserving the overall pH in the media during glucose detection. The working principle is based on an electrochemical bi-potentiostatic measurement, where an alkaline aqueous condition is locally generated at one side of nanogap, while glucose detection is performed at the counterpart. To this purpose, a nanogap array platform was fabricated by means of standard lithography and controlled electromigration. Mono-potentiostatic electrochemical detection of ascorbic acid was successfully performed to preliminary test the platform prior to measuring glucose in bi-potentiostatic mode. Cyclic voltammetries reveal that two oxidation peaks are sensitive to glucose concentration, making nanogap glucose detection possible in principle. This promising proof of concept could be innovative in bio-applications with implantable devices or direct monitoring of cell culture, where neutral pH in contact with living tissue is required. Further geometrical improvements of the system to increase the durability of the sensor are currently still in progress, and are briefly discussed in the final part of the paper.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"55 34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114509936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2013-12-12DOI: 10.1109/BIOCAS.2013.6679697
Enyi Yao, Shaista Hussain, A. Basu, G. Huang
In this paper, we describe a low power neuromorphic machine learner that utilizes device mismatch prevalent in today's VLSI processes to perform a significant part of the computation while a digital back end enables precision in the final output. The particular machine learning algorithm we use is extreme learning machine (ELM). Mismatch in silicon spiking neurons and synapses are used to perform the vector-matrix multiplication (VMM) that forms the first stage of this classifier and is the most computationally intensive. System simulations are presented to evaluate the dependence of performance (in a classification and a regression task) on analog and digital parameters like weight resolution, maximum spike frequency etc. SPICE simulations show that the proposed implementation is ≈ 92X more energy efficient as opposed to custom digital implementations for a classification task with 100 dimensional inputs. Measurement results for a regression task from a field programmable analog array (FPAA) fabricated in 0.35μm CMOS are presented as a proof of concept.
{"title":"Computation using mismatch: Neuromorphic extreme learning machines","authors":"Enyi Yao, Shaista Hussain, A. Basu, G. Huang","doi":"10.1109/BIOCAS.2013.6679697","DOIUrl":"https://doi.org/10.1109/BIOCAS.2013.6679697","url":null,"abstract":"In this paper, we describe a low power neuromorphic machine learner that utilizes device mismatch prevalent in today's VLSI processes to perform a significant part of the computation while a digital back end enables precision in the final output. The particular machine learning algorithm we use is extreme learning machine (ELM). Mismatch in silicon spiking neurons and synapses are used to perform the vector-matrix multiplication (VMM) that forms the first stage of this classifier and is the most computationally intensive. System simulations are presented to evaluate the dependence of performance (in a classification and a regression task) on analog and digital parameters like weight resolution, maximum spike frequency etc. SPICE simulations show that the proposed implementation is ≈ 92X more energy efficient as opposed to custom digital implementations for a classification task with 100 dimensional inputs. Measurement results for a regression task from a field programmable analog array (FPAA) fabricated in 0.35μm CMOS are presented as a proof of concept.","PeriodicalId":344317,"journal":{"name":"2013 IEEE Biomedical Circuits and Systems Conference (BioCAS)","volume":"165 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2013-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125969265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：