1Various Neural NetworksNetworksNeural Networks A mathematical model to solve engineering problems Group of connected neurons to realize compositions of non linear functions Tasks Classification Discrimination Estimation 2 types of networks Feed forward Neural Networks Recurrent Neural Networks Feed Forward Neural Networks The information is propagated from the inputs to the outputs Computations of2nd hiddenOutput layerfunctions from n input variables by compositions of algebraic functions Time has no role (NO cycle between outputs and inputs)x1x2 xn…..1st hidden layerlayerRecurrent Neural Networks Can have arbitrary topologies Can model systems with internal states (dynamic ones) Delays are associated to a specific weight100specific weight Training is more difficult Performance may be problematic Stable Outputs may be more difficult to evaluate Unexpected behavior (oscillation, chaos, …)x1 x201010Properties of Neural Networks Supervised networks are universal approximators networks) Theorem : Any limited function can be approximated by a neural network with a finite number of hidden neurons to an arbitrary precisionSupervised learning The desired response of the neural network in function of particular inputs is well known. A “Professor” may provide examples and teach the neural network how to fulfill a certain task2Unsupervised learning Idea : group typical input data in function of resemblance criteria un-known a priori Data clusteringNdffNo need of a professor The network finds itself the correlations between the data Examples of such networks : Kohonen feature mapsClassification (Discrimination) Class objects in defined categories Rough decision OREstimation of the probability for a certainEstimation of the probability for a certain object to belong to a specific classExample : Data mining Applications : Economy, speech and patterns recognition, sociology, etc. ExampleExamples of handwritten postal codes drawn from a database available from the US Postal serviceWhat needed to create NN ? Determination of relevant inputs Collection of data for the learning and testing phases of the neural networkFi di th ti b f hidd dFinding the optimum number of hidden nodes Learning the parameters Evaluate the performances of the network If performances are not satisfactory then review all the precedent pointsPopular neural architectures Perceptron Multi-Layer Perceptron (MLP)Radial Basis FunctionNetwork (RBFN)Radial Basis Function Network (RBFN) Time Delay Neural Network (TDNN) Other architecturesPerceptron Rosenblatt (1962) Linear separation Inputs :Vector of real valuesO++++++++++++++++++++++++++++1=yOutputs :1 or -1022110=++ xcxcc++++++++0=y0c1c2c∑1x2x122110xcxccv++=)(0 vstepy=3 The perceptron algorithm converges if examples are linearly separableMulti-Layer Perceptron One or more hidden layersOutput layer1st hidden layer2nd hiddenlayerInput dataStructureTypes ofDecision RegionsExclusive-ORProblemClasses withMeshed regionsMost GeneralRegion ShapesSingle-LayerHalf PlaneBounded ByHyperplaneAABBBADifferent non linearly separable problemsTwo-LayerThree-LayerHyperplaneConvex OpenOrClosed RegionsAbitrary(ComplexityLimited by No.of Nodes)ABAABBAABBBABA A radial basis function (RBF) is a real-valued function whose value depends only on the distance from some other point c, called a center, φ(x) = f(||x-c||)f f()Radial Basis Functions Any function φ that satisfies the property φ(x) = f(||x-c||) is a radial function. The distance is usually the Euclidean distance()∑=−=−Niii cxcx122|||| The popular output of radial basis functions is the Gaussian function:())exp(2⎟⎟⎞⎜⎜⎛−−=−ΦjjcxacxRadial Basis Functions ())exp(⎟⎟⎠⎜⎜⎝Φjjacxσa=1, c1=0.75, c2=3.25Radial Basis Functions Network (RBFN) Features One hidden layer The activation of a hidden unit is determined by a radial basis functionRadial unitsOutputsInputs4 Generally, the hidden unit function is the Gaussian function The output Layer is linear:()∑Φ=KcxWxs)(()∑=−Φ=jjjcxWxs1)(())exp(2⎟⎟⎟⎠⎞⎜⎜⎜⎝⎛−−=−ΦjjjcxwcxjσRBFN Learning The training is performed by deciding on How many hidden nodes there should be The centers and the sharpness of the Gaussians2t2 steps In the 1st stage, the input data set is used to determine the parameters of the RBF In the 2nd stage, RBFs are kept fixed while the second layer weights are learned ( Simple BP algorithm like for MLPs)Time Delay Neural Network (TDNN) Introduced by Waibel in 1989 Properties Local, shift invariant feature extractionN ti f ti fi ld bi i l l i f tiNotion of receptive fields combining local information into more abstract patterns at a higher level Weight sharing concept (All neurons in a feature share the same weights) All neurons detect the same feature but in different position Principal Applications Speech recognition Image analysis TDNNs (cont’d) Objects recognition in an image Each hidden unit receive inputs only from a small HiddenLayer 2region of the input space : receptive field Shared weights for all receptive fields => translation invariance in the response of the networkInputsHiddenLayer 1 Advantages Reduced number of weights Require fewer examples in the training setqpg Faster learning Invariance under time or space translation Faster execution of the net (in comparison of full connected MLP)Summary Neural networks are utilized as statistical tools Adjust non linear functions to fulfill a task Need of multiple and representative examples but fewer than in other methods Neural networks enable to model complex static phenomena (Feed-Forward)as well as dynamic ones (RecurentNN)Forward) as well as dynamic ones (RecurentNN) NN are good classifiers BUT Good representations of data have to be formulated Training vectors must be statistically representative of the entire input space Unsupervised techniques can help The use of NN needs a good comprehension of the
View Full Document