DOC PREVIEW
URI CSC 481 - Artificial Neural Networks

This preview shows page 1-2-3-4-5-6 out of 18 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

1page 1 of 34Class Presentation of CSC481- Artificial Neural NetworksTiegeng RenDept. of Computer Science04/19/2004page 2 of 34Outline• Problems in classification systems• Introduction of neural networks• How to use neural networks• Applications • Summary2page 3 of 34Classification SystemsMappingRelationshipMethods:Statistical classifierANN-based classifierObservation spaceSolutionspace0~255 0~255 0~255 0~255 0~255 0~255 0~255Category 1Category ..Category …Category …Category …Category …Category Npage 4 of 34Classification Process for Remote Sensing Image MappingRelationshipMethods:Statistical classifierANN-based classifierObservation spaceSolutionspaceLandsat TMBand1Band2Band3Band4Band5Band6Band70~255 0~255 0~255 0~255 0~255 0~255 0~255Category 1Category ..Category …Category …Category …Category …Category NWaterwetlandForestAgri.UrbanResidential4045611938011225Category: Forest(Pattern)3page 5 of 34Supervised Remote Sensing Image ClassificationVegetationWaterSoilBand 1 (0 ~ 255)Band 2 (0 ~ 255)page 6 of 34Statistical Methods- Need Gaussian (Normal) distribution on the input data which is required by Bayesian classifier.- Restrictions about the format of input data.Band 1 (0 ~ 255)Band 2 (0 ~ 255)VegetationWaterSoil4page 7 of 34Statistical Methods:Deciduous forestConifer forestForest WetlandMixed forestBrush LandNon-forest WetlandHow to find a boundary for the following patterns?page 8 of 34Artificial Neural Network Approach• No need for normal distribution on input data• Flexibility on input data format• Improved classification accuracy• Robust and reliability5page 9 of 34………………………Introduction to Neural Networks-- Artificial Neural Network Is Defined by ...• Processing elements• Organized topological structure• Training/Learning algorithmspage 10 of 34Processing Element (PE)f(x)OutputInputPEArtificial counterparts of neurons in a brainWj1Wj2Wj4Wj3Wj56page 11 of 34PE’s OutputFunction of Processing Elements………………………unitjo1o2o3fwj1wj2wj3ojPE’s Inputs• Receive outputs from each PEslocate in previous layer.• Compute the output with a Sigmoid activation function F(Sumof(Oi*Wji))• Transfer the output to all thePEs in next layerpage 12 of 34Organized topological structures7page 13 of 34Input layer Hidden layer Output layer………………………Input vector i(x1, x2, … xn)Output vector i(o1, o2, … om)ANN Structure- A multi-layer feedforward NNpage 14 of 34Input layer Hidden layer Output layer………………………Input vector i(x1, x2, … xn)Output vector i(o1, o2, … om)Training/Learning Algorithm- Backpropagation (BP)Feed Information Analysis Result8page 15 of 34Training/Learning AlgorithmBack-propagation Mechanism• Compute total error• Compute the partial derivatives• Update the weights and go to next epoch W (t+1) = W(t) + ∑∑−∈ npnpnPpst2)(21ijwE∂∂W∆page 16 of 34Back-propagation MechanismWijErrorw∆Global minimum9page 17 of 34BPANNError Space, Weight AdaptiveTotal ErrorStepsWijWklpage 18 of 34How to use a neural network • Analysis the problem domain• ANN design– What structure of ANN to choose– What Algorithm to use– Input and Output• Training• Applying the well-trained neural network to your problem10page 19 of 34Pattern Recognition- Understand the problemVegetation: (10, 89) ------> (1,0,0) (11,70) ------> (1,0,0) … … … (1,0,0)Water:(10, 21) ------> (0,1,0) (15, 32) ------> (0,1,0) … … … (0,1,0)Soil:(50, 40) -------> (0,0,1) (52, 40) -------> (0,0,1) … … … (0,0,1)Band 1 (0 ~ 255)Band 2 (0 ~ 255)VegetationWaterSoilWhat ANN structure to choose? – Multi-layer feed-forwardWhat ANN training algorithm?– Back-propagationpage 20 of 34ANN DesignPatternInput layer Hidden layer Output layer…………How many PEs we need - Basic rules in designing an ANN.• Input layer PEs - by dimension of input vector• Output layer PEs - by total number of patterns (classes)…………Feed ForwardBack-Propagate(0 ~ 255)(0 ~ 255)(0 ~ 1)(0 ~ 1)(0 ~ 1)Band 1 (0 ~ 255)Band 2 (0 ~ 255)VegetationWaterSoil11page 21 of 34ANN Training - From Pattern to Land Cover CategoryPattern……………………Feed ForwardBack-PropagateVegetation: (10, 89) ------> (1,0,0) (11,70) ------> (1,0,0) … … … (1,0,0)Water:(10, 21) ------> (0,1,0) (15, 32) ------> (0,1,0) … … … (0,1,0)Soil:(50, 40) -------> (0,0,1) (52, 40) -------> (0,0,1) … … … (0,0,1)Land Cover Category10891 (Vegetation)0 (Water)0 (Soil)A vegetation pixel(10, 89) ------> (1,0,0)page 22 of 34After Training waterVegetationSoil……………………xy1 (Vegetation)0 (Water)0 (Soil)A new pixel (x,y), x in band 1, y in band 2A Well-trained Neural Network12page 23 of 34Real-world Applications• Pattern recognition - Remote sensing image classification• Banking – credit evaluation• $tock market data analysis and predictionpage 24 of 34Band 4,3,2In RGBBand 5,4,3In RGBRemote sensing image classificationRhode Island 1999 Landsat-7 Enhanced Thematic MapperPlus (ETM+) Image13page 25 of 34ANN Design What ANN structure to choose? – Multi-layer feed-forwardWhat ANN training algorithm? – Back-propagation & RPROPPE in each layer? -- 6 – X – 10 page 26 of 34Training sample and Testing sampleClass Name Training SampleSize (pixels)Testing Sample Size(pixels)Agriculture 116 153Barren Land 146 125Conifer Forest 173 146Deciduous Forest 343 217Mixed Forest 265 155Brush Land 75 75Urban Area 287 108Water 238 133Non-forest Wetland 248 163Forest Wetland 52 57Total Pixels 1943 1332Training and Testing Pattern14page 27 of 34Classification ResultDeciduous forestTurf / GrassBarren landConifer forestWetland 2Mixed forestBrush LandWetland 1WaterUrban areaSome stats on the classification: Training: 1943 pixel, ANN structure: 48-350-11, Training time: 5 hours, final error: (100% - 92.7%)Classification: over 9 Million pixels, takes 6 hours to get the land-cover map.page 28 of 34Classification Result- A Close LookRhode Island 1999 ETM+Rhode Island 1999 Land-use and Land-cover mapDeciduous forestTurf / GrassBarren landConifer forestWetland 2Mixed forestBrush LandWetland 1WaterUrban area15page 29 of 34Banking – credit evaluation• 10 ~ 20 attributes as input– Yearly income, marriage status, credit history, residence, children, etc• Expert to choose typical training data set• Choose NN structure and training algorithm– A dynamic NN structure applied– Self-growing algorithm


View Full Document

URI CSC 481 - Artificial Neural Networks

Download Artificial Neural Networks
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Artificial Neural Networks and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Artificial Neural Networks 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?