Related Papers
None of the papers listed below were written by me. Additionally, I have not contacted any of the authors, or obtained their permission to redistribute their work here. I collected these works for my own convenience, and quickly copied or wrote the abstract shown so that I could refer to them more easily. I am not sure if this page is a violation of copyright laws or not, but if you are the author of one of the papers provided here and wish to have it removed, please contact me and I will do so immediately.
Sincerely,
Jack 11/7/99
Analysis of EEG Signals with Wavelets and Knowledge Engineering Techniques
Saman K. Halgamnge, Christopher S. Herrmann, Lakhmi Jain
Neuro-Fuzzy analysis of EEG data, wavelet transformation. Compares three different Neuro-Fuzzy methods. 93.1% accuracy in detection of an artifact indicating some medical phenomenon.
Wavelet Transform as Feature Extraction for Medical Fuzzy Diagnosis
Christoph S. Herrmann, Frank Reine
Wavelet transformation as preprocessor to a fuzzy expert system.
Removing Electroencephalographic Artifacts: Comparison between ICA and PCA
Tzyy-Ping Jung, Colin Humphries, Te-Won Lee, Scott Makeig, Martin J.
Mckeown, Vincente Iragui, Terrence J. Sejnowski
Pervasive EEG artifacts associated with blinks, eye-movements, muscle noise, cardiac signals, and line noise poses a major challenge for EEG interpretation and analysis. A generally applicable method for removing a wide variety of artifacts from EEG records based on an extended version of an Independent Component Analysis algorithm for performing blind signal source separation on linear mixtures of independent source signals.
Blind Separation of Auditory Event-related Brain Reponses into Independent
Components
Scott Makeig, Tzyy-Ping Jung, Anthony J. Bell, Dara Ghahremani,
Terrence J. Sejnowski
A method based on information theory that decomposes one or more ERPs recorded at multiple scalp sensors into a sum of components with fixed scalp distributions and sparsely-activated, maximally independent time courses.
Estimating Alertness from the EEG Power Spectrum
Tzyy-Ping Jung, Scott Makeig, Magnus Stensmo, Terrence J. Sejnowski
Changes in the electroencephalographic power spectrum accompany fluctuations in the level of alertness, as assessed by measuring simultaneous changes in EEG and performance on an auditory monitoring task. Continuous, accurate, noninvasive, and near real-time estimation of an operator's global level of alertness is feasible using EEG measures recorded from as few as two central scalp sites.
Classification of EEG Signals from Four Subjects During Five Mental Tasks
Charles W. Anderson, Zlatko Sijercic
Neural networks are trained to classify half-second segments of six-channel, EEG data into one of five classes corresponding to five cognitive tasks performed by four subjects. Two and three-layer feedforward neural networks are trained using 10-fold cross validation and early stopping to control over-fitting. EEG signals were represented as autoregressive (AR) models. The average percentage of test segments correctly classified ranged from 71% for one subject to 38% for another subject. Cluster analysis of the resulting neural networks' hidden-unit weight vectors identifies which EEG channels are most relevant to this discrimination problem.
Discriminating Mental Tasks Using EEG Represented by AR Models
Charles W. Anderson, Erik A. Stolz, Sanyogita Shamsunder
EEG signals are modeled using single-channel and multi-channel autoregressive (AR) techniques. The coeffecients of these models are used to classify EEG data into one of two clases corresponding to the mental task the subject are performing. A neural network is trained to perform the classification. Multivariate AR representation performs slightly better, resulting in an average classification accuracty of about 91%.
Thesis: Analysis of LVQ in the Context of Spontaneous EEG Signal Classification
Daniel Kermit Ford
Learning Vector Quantization (LVQ) has proven to be an effective classification procedure. Investigates what and how LVQ learns in the context of EEG signal classification. LVQ is shown to be comparable with other neural network algorithms for the task of classifying EEG signals, yielding approximately 80% classification accuracy for three out of the four subjects tested when differentiating between two different mental tasks.
Effects of Variations in Neural Network Topology and Output Averaging on the
Discrimination of Mental Tasks from Spontaneous Electroencephalogram
Charles W. Anderson
EEG from one subject who performed three mental tasks are classified by neural networks. Using a sixth-order autoregressive (AR) model of half-second windows of six-chanell EEG, a classification accuracy of 89% on test data is achieved. A cross-validation study of a variety of neural network topologies showed that a network with one hidden layer of 20 units produced the best performance. It was also found that averaging the output of the network over consecutive inputs improved performance.
Using Feedforward Neural Networks to Monitor Alertness from Changes in EEG
Correlation and Coherence
Scott Makeig
Changes in the normalized EEG cross-spectrum can be used in conjuction with feedforward neural networks to monitor changes in alertness of operators continuously and in near-real time. Increases in the frequency of detection errors in this task are also accompanied by patterns of increased and decreased spectral coherence in several frequency hand and EEG channel pairs.
Independent Component Analysis of Electroencephalographic Data
Scott Makeig, Anthony J. Bell, Tzyy-Ping Jung, Terrence J. Sejnowski
Because of the distance between the skull and brain and their different resistivities, electroencephalogramic data collected from any point on the human scalp includes activity generated within a large brain area. This spatial smearing of EEG data by volume conduction does not involve significant time delays, however, suggesting that the Independent Component Analysis algorithm of Bell and Sejnowski is suitable for performing blind source separation on EEG data. Results: (1) ICA training is insensitive to different random seeds. (2) ICA may be used to segregate obvious artifactual EEG components (line and muscle noise, eye movements) from other sources. (3) ICA is capable of isolating overlapping EEG phenomena, including alpha and theta bursts and spatially-separable ERP components, to separate ICA channels. (4) Nonstationarities in EEG and behavioral state can be tracked using ICA via changes in the amount of residual correlation between ICA-filtered output channels.
EEG Signal Classification with Different Signal Representations
Charles W. Anderson, Saikumar V. Devulapalli, Erik A. Stolz
Comparison of four representations of EEG signals and their classification by a two-layer neural network with sigmoid activation functions. Unprocessed representation @ 250 Hz: 53.2% accuracy. Unprocessed representation @ 125 Hz: 51.1% accuracy. Karhunen-Loeve representation: 51.7% accuracy. Frequency-Band representation: 79.3% accuracy.
Changes in alertness are a principal component of variance in the EEG spectrum
Scott Makeig, Tzyy-Ping Jung
Correlate minute-scale fluctuations in the normalized EEG log spectrum during the drowsiness with concurrent changes in level of performance on a sustained auditory detection task, and show that single principal component of EEG variance is linearly related to minutes-scale changes in detection performance.
Thesis: Non-Linear Principal Component Analysis and Classification of EEG
During Mental Tasks
Saikumar Devulapalli
Explores the effectiveness of Non-Linear Principal Component Analysis (NLPCA) as a technique for reducing the dimensionality of human electroencephalogram (EEG) for enabling it to be classified into two different mental tasks. EEG signals from a single subject recorded through six channels was studied during the performance of two mental tasks. Standard backpropagation network. Results: 86.22% accuracy for distinguishing between two tasks.
Determining Mental State from EEG Signals Using Parallel Implementations of
Neural Networks
Charles W. Anderson, Saikumar V. Devulapalli, Erik A. Stolz
Best classifications: 73% using frequency-based representation.
Multivariate Autoregressive Models for Classification of Spontaneous
Electroencephalogram During Mental Tasks
Charles W. Anderson, Erik A. Stolz, Sanyogita Shamsunder
Explores the use of scalar and multivariate AR models to extract features from EEG with which mental tasks can be discriminated. Multivariate AR coefficients performed best with average classification accuracy of 91.4% on novel, untrained EEG signals.
EEG Signal Compression with ADPCM Subband Coding
Zlatko Sijercic, Gyan C. Agarwal, Charles W. Anderson
EEG signal compression method that combines both octave-band filter bank frequency decomposition and coding in subbands using adaptive differential pulse code modulation (ADPCM). Besides being computationally effective, this compression method in its simplest form yields 70% data reductions with very little distorsion. Higher compression rates are obtained by increasing the order of the predictor used.