More Related Content
The second seminar
- 1. EEG Brain Signal Classification for BCI Applications http://eegclassifyandrecognize.blogspot.com/
- 6. Assistants
- 8. Team Members
- 9. Ahmed KhaledAbd El-glil (Information Systems)
- 10. Ahmed Mohamed Ahmed Mahany (Computer Systems)
- 11. Islam Ahmed HamedElgarhy (Computer Systems)
- 12. KamalAshrafKamal El-deen (Information Systems)
- 13. Mohammed Saeed Ibrahim (Scientific Computing)AgendaProject Overview
- 14. Objectives
- 18. Methodology
- 19. Past, Present, Future Work
- 20. Time Plan
- 21. ReferencesProject Overview(Objectives)Develop a generic EEG Classification that can be used in different brain computer interface applications. Project Overview(Problem Statement)EEG Signal is a composite signal.EEG Signal OverviewAn electroencephalogram is a measure of the brain's voltage changes as detected from scalp electrodes.
- 22. Measured in microvolt (µV) .EEG Signal Overview(Cont.)BETA
- 23. 12-30
- 24. Thinking
- 25. Alert & working
- 26. GAMMA
- 27. 30-100
- 28. Short term memory of (sound , tactile,…..)
- 29. DELTA
- 30. Up to 4
- 31. These occur in deep sleep
- 32. Childhood and in serious organic brain diseasesTHETA
- 33. 4 – 7
- 34. Move of Hand
- 35. Idling
- 36. These occur during childhood
- 37. ALPHA
- 40. closing the eyesEEG Signal For Normal HumanHand Movement
- 41. Tongue MovementEEG Signal For Normal HumanWalking && Rotation
- 42. Eye BlinkSystem ArchitectureEEG Signal acquisitionEEG Signal Pre-processing EEG Signal Post-ProcessFeature Extraction Feature Selection Classification
- 43. System Architecture(cont.)The whole classification system contains four parts :
- 44. Preprocessing
- 45. Feature Extraction and Features Selection ( Dimensionality Reduction)
- 46. Classification
- 47. Post processingMethodologyIndependent Component Analysis(ICA) for Feature extraction.
- 48. Convolution Neural Network for classification. PreprocessingFilter out Noise And Remove The Artifacts
- 49. Removing eye blinks and muscular movements FeatureExtraction & SelectionPower Spectral Density
- 50. Temporal information from a window of data is extracted and then processed using a static classifier.
- 51. Spatial information alone is indeed powerful enough to produce state-of-the-art performance
- 52. Independents Components Analysis (EXPLAIN NEXT)FeatureExtraction & SelectionThe goal of feature subspace projections is to improve classifier robustness by reducing data dimensionality in order to facilitate better generalization, as well as reducing the learning and operating complexity of the classifiers.Independents Components AnalysisDefinition
- 53. Mixed Signals in Matrix Notation Independents Components AnalysisFind W using the ICA Algorithm
- 54. ICA Block Diagram (2 Signals)Signal #1Signal #2a11a22a12a21MultiplexedSignal #2MultiplexedSignal #1
- 55. Independents Components AnalysisICASensor #1Feature #1Sensor #2Feature #2Sensor #3Feature #3Sensor #4Feature #4
- 57. An information processing paradigm that is inspired by the way biological nervous systems, such as the brain.
- 58. It Composed of large number of neurons to solve a specific problem.How the Human Brain Learns?Neuron collects signals from others through dendrites.
- 59. The neuron sends out spikes of electrical activity through a long axon.
- 60. Synapse converts the activity from the axon into electrical effects.NeuralNetworks(cont.)General function can be formalized asXp : Input Pattern W :WeightX1W1 + X2W2 + X3W3 + ... > T
- 61. NeuralNetworks(cont.)Training Set FunctionTp: Desired outputFor calculating error from error function
- 62. For calculating optimal or near optimal weights NeuralNetworks(Multi-Layer Perceptron) Multi-layer Perceptron
- 63. Set (P) of several perceptrons , each of which classifies the input data differently, can be combined via an additional perceptron which receives all outputs from (P) as input.NeuralNetworks(Multi-Layer Perceptron)For calculating error from error function
- 64. For calculating optimal or near optimal weights
- 65. The output of neuron j in the output layerNeuralNetworks(Multi-Layer Perceptron)Multi-layer Perceptron Problems:
- 66. The number of trainable parameters becomes extremely large.
- 67. It offers little or no invariance to shifting , scaling and other forms of distortion.
- 68. The topology of the input data is completely ignored, yielding similar training results for all permutations of the input data.These problems are solved by convolution neural networks concepts.
- 69. ConvolutionNeuralNetworksIt divided input pattern to set of sub sampling layers .
- 70. Each of sub sampling consists of set of features maps that have a slightly smaller resolution than the input pattern.ConvolutionNeuralNetworks(cont.) It trained through back propagation because all neurons in one feature map share the same weights.
- 71. This lead to an implicit reduction of the gap.
- 72. The sub sampling layers have one trainable weight so number of free parameters in it is lower than in conventional layers.
- 73. It requires less computational effort than the training of multi-layers perceptrons. PostprocessingThe output of the classification system may be contains erroneous decisions.Past, Present and FuturePast
- 74. Research about EEG Signal
- 75. Present
- 76. Study feature extraction and classification algorithms .
- 77. Future
- 78. Implement the Classification System. Time Plan
- 79. References David Bouchain “Character Recognition Using Convolutional Neural Networks “. Seminar Statistical Learning Theory University of Ulm, Germany Institute for Neural Information Processing ,2006/2007.Christos Stergiou and DimitriosSiganos” Neural Networks ”. TianLan, DenizErdogmus, Andre Adami, and Michael Pavel, “Feature Selection by Independent Component Analysis and Mutual Information Maximization in EEG Signal Classification”, Department of Biomedical Engineering OGI School of Science and Engineering Oregon Health & Science University Beaverton, Oregon 97006, USA.4. Rave Harpaz ,” Independent Component Analysis: An Introduction” Pattern Recognition Laboratory The Graduate Center, City University of New York 365 Fifth Avenue New York, NY 10016, USA, Nov/15/ 2005.
- 80. Thanks