Pub Date : 2015-04-22DOI: 10.1109/NER.2015.7146650
Chidrupi Inguva, Paul Wong, A. Sue, A. McEwan, P. Carter
Volume conduction models of the implanted cochlea are useful tools for investigating cochlear implant function. To date, however, all existing models have assumed that the tissues of the cochlea are purely resistive, despite evidence to the contrary. In this paper, a preliminary attempt to incorporate frequency-dependent effects is made using a simple, extruded finite element model of the cochlea. It was found that resistive and dispersive formulations exhibited marked differences in the pattern of current flow, especially later in the phase. The scala tympani response remained largely resistive as per published experimental evidence. However, injected current was also diverted away from higher impedance bone and neural tissue towards lower impedance pathways, particularly the cerebrospinal fluid in the modiolus. Further investigation of these effects is warranted to better understand these differences and how they might affect existing models of neural excitation.
{"title":"Frequency-dependent simulation of volume conduction in a linear model of the implanted cochlea","authors":"Chidrupi Inguva, Paul Wong, A. Sue, A. McEwan, P. Carter","doi":"10.1109/NER.2015.7146650","DOIUrl":"https://doi.org/10.1109/NER.2015.7146650","url":null,"abstract":"Volume conduction models of the implanted cochlea are useful tools for investigating cochlear implant function. To date, however, all existing models have assumed that the tissues of the cochlea are purely resistive, despite evidence to the contrary. In this paper, a preliminary attempt to incorporate frequency-dependent effects is made using a simple, extruded finite element model of the cochlea. It was found that resistive and dispersive formulations exhibited marked differences in the pattern of current flow, especially later in the phase. The scala tympani response remained largely resistive as per published experimental evidence. However, injected current was also diverted away from higher impedance bone and neural tissue towards lower impedance pathways, particularly the cerebrospinal fluid in the modiolus. Further investigation of these effects is warranted to better understand these differences and how they might affect existing models of neural excitation.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130554009","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 : 2015-04-22DOI: 10.1109/NER.2015.7146718
A. Richardson, Pauline K. Weigand, Srihari Y. Sritharan, T. Lucas
High-performance neuroprostheses designed to reanimate a paralyzed limb following spinal cord injury must restore both movement and sensation. For the latter goal, we are developing a novel strategy focused on encode sensations using microstimulation of the cuneate nucleus (CN) of the brainstem. Here, we characterized the temporal dynamics of downstream cortical excitation and inhibition in response to CN microstimulation in a macaque. A single CN stimulus pulse evoked a fast (7 ms) excitatory response in primary somatosensory cortex (S1) followed by an inhibitory period lasting until 50 ms. The S1 response to a second CN pulse within this inhibitory period was drastically attenuated. Following the inhibition, S1 unit activity rebounded with a prolonged excitatory phase lasting until 800 ms. Within this second excitatory phase were rhythmic peaks of increased unit activity with an alpha-band frequency (8-14 Hz). The rhythmic excitation was specific for perigranular laminae and was stimulus-amplitude dependent. The results show a complex cortical response to CN stimuli and can guide future design of CN stimulus patterns to evoke salient percepts.
{"title":"Somatosensory encoding with cuneate nucleus microstimulation: Effects on downstream cortical activity","authors":"A. Richardson, Pauline K. Weigand, Srihari Y. Sritharan, T. Lucas","doi":"10.1109/NER.2015.7146718","DOIUrl":"https://doi.org/10.1109/NER.2015.7146718","url":null,"abstract":"High-performance neuroprostheses designed to reanimate a paralyzed limb following spinal cord injury must restore both movement and sensation. For the latter goal, we are developing a novel strategy focused on encode sensations using microstimulation of the cuneate nucleus (CN) of the brainstem. Here, we characterized the temporal dynamics of downstream cortical excitation and inhibition in response to CN microstimulation in a macaque. A single CN stimulus pulse evoked a fast (7 ms) excitatory response in primary somatosensory cortex (S1) followed by an inhibitory period lasting until 50 ms. The S1 response to a second CN pulse within this inhibitory period was drastically attenuated. Following the inhibition, S1 unit activity rebounded with a prolonged excitatory phase lasting until 800 ms. Within this second excitatory phase were rhythmic peaks of increased unit activity with an alpha-band frequency (8-14 Hz). The rhythmic excitation was specific for perigranular laminae and was stimulus-amplitude dependent. The results show a complex cortical response to CN stimuli and can guide future design of CN stimulus patterns to evoke salient percepts.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"133 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123231909","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 : 2015-04-22DOI: 10.1109/NER.2015.7146669
E. Brunton, E. Yan, Katherine L. Gillespie-Jones, Arthur James Lowery, R. Rajan
Stimulation of neural tissue for the remediation of brain and sensory deficits requires that stimulation paradigms are selected carefully to deliver the most stable, efficient and safest stimulation to evoke the desired therapeutic response. Here we asked two questions of a penetrating cortical prosthesis use to evoke sensory-guided behavior: 1) does the threshold charge required to evoke a behavioral response change over time? 2) what effect does changing the frequency of stimulation have on stimulus threshold? To answer these questions we implanted a 4-electrode array into the somatosensory (tactile) cortex of a Sprague Dawley rat. The threshold charge to evoke a behavioral response was measured weekly over a 9-week period. Stimulation frequencies of 50 or 200 Hz were used, while all other stimulus parameters were kept constant. Within a maximum current limit of 100 μA and with a pulse width of 200 μs, we reliably elicited a behavioral response on 2 electrodes. Over the 9 week implantation period there was an initial increase in threshold current at 4 weeks, followed by a decrease at week 5 post-implantation; by week 8 post-implantation, thresholds appeared to have stabilized. Although we could reliably evoke a response at both 50 and 200 Hz, the stimulus frequency of 50 Hz required on average a lower threshold charge to evoke a response.
{"title":"Chronic thresholds for evoking perceptual responses in the rat sensory cortex","authors":"E. Brunton, E. Yan, Katherine L. Gillespie-Jones, Arthur James Lowery, R. Rajan","doi":"10.1109/NER.2015.7146669","DOIUrl":"https://doi.org/10.1109/NER.2015.7146669","url":null,"abstract":"Stimulation of neural tissue for the remediation of brain and sensory deficits requires that stimulation paradigms are selected carefully to deliver the most stable, efficient and safest stimulation to evoke the desired therapeutic response. Here we asked two questions of a penetrating cortical prosthesis use to evoke sensory-guided behavior: 1) does the threshold charge required to evoke a behavioral response change over time? 2) what effect does changing the frequency of stimulation have on stimulus threshold? To answer these questions we implanted a 4-electrode array into the somatosensory (tactile) cortex of a Sprague Dawley rat. The threshold charge to evoke a behavioral response was measured weekly over a 9-week period. Stimulation frequencies of 50 or 200 Hz were used, while all other stimulus parameters were kept constant. Within a maximum current limit of 100 μA and with a pulse width of 200 μs, we reliably elicited a behavioral response on 2 electrodes. Over the 9 week implantation period there was an initial increase in threshold current at 4 weeks, followed by a decrease at week 5 post-implantation; by week 8 post-implantation, thresholds appeared to have stabilized. Although we could reliably evoke a response at both 50 and 200 Hz, the stimulus frequency of 50 Hz required on average a lower threshold charge to evoke a response.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121287950","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 : 2015-04-22DOI: 10.1109/NER.2015.7146819
Saeed Bamatraf, M. Hussain, Hatim Aboalsamh, H. Mathkour, A. Malik, H. Amin, Muhammad Ghulam, Emad-ul-Haq Qazi
Electroencephalography (EEG) has been widely adopted for investigating brain behavior in different cognitive tasks e.g. learning and memory. In this paper, we propose a pattern recognition system for discriminating the true and false memories in case of short-term memory (STM) for 3D and 2D educational contents by analyzing EEG signals. The EEG signals are converted to scalp-maps (topomaps) and city-block distance is applied to reduce the redundancy and select the most discriminative topomaps. Finally, statistical features are extracted from selected topomaps and passed to Support Vector Machine (SVM) to predict brain states corresponding to true and false memories. A sample of thirty four healthy subjects participated in the experiments, which consist of two tasks: learning and memory recall. In the learning task, half of the participants watched 2D educational contents and half of them watched the same contents in 3D mode. After 30 minutes of retention, they were asked to perform memory recall task, in which EEG signals were recorded. The classification accuracy of 97.5% was achieved for 3D as compared to 96.5% for 2D. The statistical analysis of the results suggest that there is no significant difference between 2D and 3D educational contents on STM in terms of true and false memory assessment.
{"title":"A system based on 3D and 2D educational contents for true and false memory prediction using EEG signals","authors":"Saeed Bamatraf, M. Hussain, Hatim Aboalsamh, H. Mathkour, A. Malik, H. Amin, Muhammad Ghulam, Emad-ul-Haq Qazi","doi":"10.1109/NER.2015.7146819","DOIUrl":"https://doi.org/10.1109/NER.2015.7146819","url":null,"abstract":"Electroencephalography (EEG) has been widely adopted for investigating brain behavior in different cognitive tasks e.g. learning and memory. In this paper, we propose a pattern recognition system for discriminating the true and false memories in case of short-term memory (STM) for 3D and 2D educational contents by analyzing EEG signals. The EEG signals are converted to scalp-maps (topomaps) and city-block distance is applied to reduce the redundancy and select the most discriminative topomaps. Finally, statistical features are extracted from selected topomaps and passed to Support Vector Machine (SVM) to predict brain states corresponding to true and false memories. A sample of thirty four healthy subjects participated in the experiments, which consist of two tasks: learning and memory recall. In the learning task, half of the participants watched 2D educational contents and half of them watched the same contents in 3D mode. After 30 minutes of retention, they were asked to perform memory recall task, in which EEG signals were recorded. The classification accuracy of 97.5% was achieved for 3D as compared to 96.5% for 2D. The statistical analysis of the results suggest that there is no significant difference between 2D and 3D educational contents on STM in terms of true and false memory assessment.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116177967","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 : 2015-04-22DOI: 10.1109/NER.2015.7146740
Mariah W. Whitmore, L. Hargrove, E. Perreault
Falls initiated by slipping are a major cause for concern for lower-limb amputees, due to their lacking the distal musculature that aids in avoiding the initiation of a slip. It has been previously demonstrated that able-bodied individuals can interact safely with slippery surfaces by adapting limb kinematics and altering muscle activity to minimize slipping. Newly developed prosthetic devices have the potential to restore specific gait modes to the user, such as walking on a slippery surface, if only more was known about how the mechanical properties should be regulated in each mode. As a first step towards understanding the mechanics relevant to slip prevention, this study sought to quantify lower-limb muscle activity during steady state walking on a range of slippery surfaces. A specific goal was to quantify how people walk on moderately slippery surfaces that pose a hazard, but are more likely to be found on an everyday basis than some of the surfaces previously studied. Our results showed a significant trend (p<;0.001) towards decreasing the level of activity used at the ankle as the floor becomes more slippery. In contrast, there is a significant trend (p<;0.001) towards increasing the level of activity used at the knee. These findings suggest a strategy in which the ankle becomes increasingly compliant to maximize the surface area in contact with the floor, while increased activity in proximal muscles is used to help stabilize the legs and trunk for increased safety.
{"title":"Lower-limb muscle activity when walking on different slippery surfaces","authors":"Mariah W. Whitmore, L. Hargrove, E. Perreault","doi":"10.1109/NER.2015.7146740","DOIUrl":"https://doi.org/10.1109/NER.2015.7146740","url":null,"abstract":"Falls initiated by slipping are a major cause for concern for lower-limb amputees, due to their lacking the distal musculature that aids in avoiding the initiation of a slip. It has been previously demonstrated that able-bodied individuals can interact safely with slippery surfaces by adapting limb kinematics and altering muscle activity to minimize slipping. Newly developed prosthetic devices have the potential to restore specific gait modes to the user, such as walking on a slippery surface, if only more was known about how the mechanical properties should be regulated in each mode. As a first step towards understanding the mechanics relevant to slip prevention, this study sought to quantify lower-limb muscle activity during steady state walking on a range of slippery surfaces. A specific goal was to quantify how people walk on moderately slippery surfaces that pose a hazard, but are more likely to be found on an everyday basis than some of the surfaces previously studied. Our results showed a significant trend (p<;0.001) towards decreasing the level of activity used at the ankle as the floor becomes more slippery. In contrast, there is a significant trend (p<;0.001) towards increasing the level of activity used at the knee. These findings suggest a strategy in which the ankle becomes increasingly compliant to maximize the surface area in contact with the floor, while increased activity in proximal muscles is used to help stabilize the legs and trunk for increased safety.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"98 6","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113944094","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 : 2015-04-22DOI: 10.1109/NER.2015.7146736
Xiang-Yu Gao, Yu-Fei Zhang, Wei-Long Zheng, Bao-Liang Lu
Fatigue is a status of human brain activities, and driving fatigue detection is a topic of great interest all over the world. In this paper, we propose a measure of fatigue produced by eye tracking glasses, and use it as the ground truth to evaluate driving fatigue detection algorithms. Particularly, PERCLOS, which is the percentage of eye closure over the pupil over time, was calculated from eyelid movement data provided by eye tracking glasses. Experiments of a vigilance task were carried out in which both EOG signals and eyelid movement were recorded. The evaluation results of an effective EOG-based fatigue detection algorithm convinced us that our proposed measure is an appropriate candidate for evaluating driving fatigue detection algorithms.
{"title":"Evaluating driving fatigue detection algorithms using eye tracking glasses","authors":"Xiang-Yu Gao, Yu-Fei Zhang, Wei-Long Zheng, Bao-Liang Lu","doi":"10.1109/NER.2015.7146736","DOIUrl":"https://doi.org/10.1109/NER.2015.7146736","url":null,"abstract":"Fatigue is a status of human brain activities, and driving fatigue detection is a topic of great interest all over the world. In this paper, we propose a measure of fatigue produced by eye tracking glasses, and use it as the ground truth to evaluate driving fatigue detection algorithms. Particularly, PERCLOS, which is the percentage of eye closure over the pupil over time, was calculated from eyelid movement data provided by eye tracking glasses. Experiments of a vigilance task were carried out in which both EOG signals and eyelid movement were recorded. The evaluation results of an effective EOG-based fatigue detection algorithm convinced us that our proposed measure is an appropriate candidate for evaluating driving fatigue detection algorithms.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"67 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126311754","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 : 2015-04-22DOI: 10.1109/NER.2015.7146784
Ankit Sethi, C. Kemere
In the CA1 region of the rat hippocampus, fast field oscillations termed sharp wave ripples have been identified as playing a crucial role in memory formation and learning. During ripple activity, particular sequences of neurons fire in a phenomena called replay. So termed because the spiking encodes patterns of past experiences, the exact role of the content of replay is an active subject of investigation in order to determines its relationship with learning and memory guided decision making. A need arises for systems that can decode replay activity during ripples in real time. This necessitates fast algorithms for both spike sorting and ripple detection with the lowest possible latency. A low latency implementation makes possible feedback experiments where decoded ripple sequences can, with minimal delay, trigger stimulating pulses that can disrupt particular kinds of decoded information before they can contribute to behavior. In this study, we optimize and implement a recently proposed online spike sorting algorithm for an increasingly popular electrophysiological software suite and measure improvements that greatly enhance its multi-tetrode decoding capabilities. Synchronizing with online ripple detection, this novel framework will allows experimenters to study the effects of disrupting replay activity with a degree of granularity hitherto unavailable.
{"title":"Mulitchannel real time spike sorting for decoding ripple sequences","authors":"Ankit Sethi, C. Kemere","doi":"10.1109/NER.2015.7146784","DOIUrl":"https://doi.org/10.1109/NER.2015.7146784","url":null,"abstract":"In the CA1 region of the rat hippocampus, fast field oscillations termed sharp wave ripples have been identified as playing a crucial role in memory formation and learning. During ripple activity, particular sequences of neurons fire in a phenomena called replay. So termed because the spiking encodes patterns of past experiences, the exact role of the content of replay is an active subject of investigation in order to determines its relationship with learning and memory guided decision making. A need arises for systems that can decode replay activity during ripples in real time. This necessitates fast algorithms for both spike sorting and ripple detection with the lowest possible latency. A low latency implementation makes possible feedback experiments where decoded ripple sequences can, with minimal delay, trigger stimulating pulses that can disrupt particular kinds of decoded information before they can contribute to behavior. In this study, we optimize and implement a recently proposed online spike sorting algorithm for an increasingly popular electrophysiological software suite and measure improvements that greatly enhance its multi-tetrode decoding capabilities. Synchronizing with online ripple detection, this novel framework will allows experimenters to study the effects of disrupting replay activity with a degree of granularity hitherto unavailable.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"101 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126433297","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 : 2015-04-22DOI: 10.1109/NER.2015.7146658
Oscar F. Cota, D. Plachta, T. Stieglitz, Y. Manoli, M. Kuhl
The following work presents a CMOS-integrated low-noise pre-amplifier (LNA) for bio-potential recordings, which is part of a multi-channel neural recording system. The versatile pre-amplifier channel features a tunable lower cut-off frequency from 0.2 Hz to 10 kHz, an upper cut-off frequency from 37.9 Hz to 11 kHz, and a middle-band gain from 41 to 45 dB. With its variable power consumption from 3.3 μW to 1 mW, the input-referred noise can be set from 2 down to 0.8 μVRMS. The pre-amplifier, fabricated in a 0.35 μm CMOS process, was successfully tested for ECG, EMG, and EEG applications.
{"title":"In-vivo characterization of a 0.8 – 3 µVRMS input-noise versatile CMOS pre-amplifier","authors":"Oscar F. Cota, D. Plachta, T. Stieglitz, Y. Manoli, M. Kuhl","doi":"10.1109/NER.2015.7146658","DOIUrl":"https://doi.org/10.1109/NER.2015.7146658","url":null,"abstract":"The following work presents a CMOS-integrated low-noise pre-amplifier (LNA) for bio-potential recordings, which is part of a multi-channel neural recording system. The versatile pre-amplifier channel features a tunable lower cut-off frequency from 0.2 Hz to 10 kHz, an upper cut-off frequency from 37.9 Hz to 11 kHz, and a middle-band gain from 41 to 45 dB. With its variable power consumption from 3.3 μW to 1 mW, the input-referred noise can be set from 2 down to 0.8 μVRMS. The pre-amplifier, fabricated in a 0.35 μm CMOS process, was successfully tested for ECG, EMG, and EEG applications.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126229339","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 : 2015-04-22DOI: 10.1109/NER.2015.7146576
Sami Dalhoumi, G. Dray, J. Montmain, G. Derosière, S. Perrey
Learning from other subjects and/or sessions led to considerable reduction of calibration time in EEG-based BCIs. However, such learning scheme is not straightforward because of the non-stationary nature of EEG signals. In this paper, we propose an adaptive accuracy-weighted ensemble (AAWE) approach that allows tracking non-stationarity in EEG signals and effectively learning from other subjects. It consists of an ensemble of classifiers, each of which is trained using data recorded from one BCI user. Classifiers' weights are initialized according to their accuracy in classifying calibration data of current BCI user. These weights are updated using ensemble decision during feedback phase, when there is no information about true class labels. The effectiveness of our approach is demonstrated through an empirical comparison with other state of the art classifiers combination strategies.
{"title":"An adaptive accuracy-weighted ensemble for inter-subjects classification in brain-computer interfacing","authors":"Sami Dalhoumi, G. Dray, J. Montmain, G. Derosière, S. Perrey","doi":"10.1109/NER.2015.7146576","DOIUrl":"https://doi.org/10.1109/NER.2015.7146576","url":null,"abstract":"Learning from other subjects and/or sessions led to considerable reduction of calibration time in EEG-based BCIs. However, such learning scheme is not straightforward because of the non-stationary nature of EEG signals. In this paper, we propose an adaptive accuracy-weighted ensemble (AAWE) approach that allows tracking non-stationarity in EEG signals and effectively learning from other subjects. It consists of an ensemble of classifiers, each of which is trained using data recorded from one BCI user. Classifiers' weights are initialized according to their accuracy in classifying calibration data of current BCI user. These weights are updated using ensemble decision during feedback phase, when there is no information about true class labels. The effectiveness of our approach is demonstrated through an empirical comparison with other state of the art classifiers combination strategies.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125602707","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 : 2015-04-22DOI: 10.1109/NER.2015.7146630
Yuichiro Yada, Takeshi Mita, R. Kanzaki, D. Bakkum, Hirokazu Takahashi
Synchrony in a neuronal network is not just a spontaneous event but rather a representation of inner information. In this point of view, the variety of synchrony patterns is considered to be related to inner capacity of the network. However, evaluating and comparing the variety of synchrony patterns, especially between different samples or different times, is difficult. In this paper, we proposed to identify the variety of synchrony based on Bayesian model selection. Hypothesizing that globally synchronized activity consists of partial synchrony, we attempted to identify reproducible-spatial pattern bases in spontaneous bursting activities of dissociated cortical cultures using Bayesian non-negative matrix factorization. Neuronal activity was recorded with high-density CMOS electrode arrays. Bayesian treatment provides evidence for selection of the number of bases based on marginal likelihood. We compared model evidence of the activity in juvenile and matured cultures. Our results suggested that the variety of synchrony patterns diversify through maturation.
{"title":"Identification of diverse synchrony patterns in dissociated cortical culture using Bayesian non-negative matrix factorization","authors":"Yuichiro Yada, Takeshi Mita, R. Kanzaki, D. Bakkum, Hirokazu Takahashi","doi":"10.1109/NER.2015.7146630","DOIUrl":"https://doi.org/10.1109/NER.2015.7146630","url":null,"abstract":"Synchrony in a neuronal network is not just a spontaneous event but rather a representation of inner information. In this point of view, the variety of synchrony patterns is considered to be related to inner capacity of the network. However, evaluating and comparing the variety of synchrony patterns, especially between different samples or different times, is difficult. In this paper, we proposed to identify the variety of synchrony based on Bayesian model selection. Hypothesizing that globally synchronized activity consists of partial synchrony, we attempted to identify reproducible-spatial pattern bases in spontaneous bursting activities of dissociated cortical cultures using Bayesian non-negative matrix factorization. Neuronal activity was recorded with high-density CMOS electrode arrays. Bayesian treatment provides evidence for selection of the number of bases based on marginal likelihood. We compared model evidence of the activity in juvenile and matured cultures. Our results suggested that the variety of synchrony patterns diversify through maturation.","PeriodicalId":137451,"journal":{"name":"2015 7th International IEEE/EMBS Conference on Neural Engineering (NER)","volume":"115 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2015-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126004313","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：