LVQ学习矢量量化

    现在已经2011年五月份了，今年的两个多月几乎没什么成果，想到水样年华这个词，额！总结过往，翘首未来，心中不免些许迷茫,程序员之路在何方？靡靡之音，不绝于耳，然何时我也能奏一首阳春白雪，不为流传于世，只求一心灵安静之所！！

 

     LVQ原理不再赘述，其本质可以看成是最近邻分类，所谓竞争性神经网络，即把欧氏距离最小的神经元当做胜出神经元，针对正确分类和不正确分类情况进行调整权值。后来又发展了LVQ2，LVQ2.1,LVQ3等算法，在本质上没有改变，只不过又考虑了次获胜神经元。

     其中LVQ1，LVQ2.1在matlab神经网络工具箱中已经实现，但是LVQ3目前为止还没有找到较好的资源，我根据网上现存的程序进行了修改。

function [dw,ls] = learnlv3(w,p,z,n,a,t,e,gW,gA,d,lp,ls) %LEARNLV2 LVQ2 weight learning function. % % Syntax % % [dW,LS] = learnlv3(w,p,n,a,T,lp,ls,Ttrain,C) % info = learnlv2(code) % % Description % % LEARNLV3 is the OLVQ weight learning function. % % LEARNLV2(W,P,Z,N,A,T,E,gW,gA,D,LP,LS) takes several inputs, % W - SxR weight matrix (or Sx1 bias vector). % P - RxQ input vectors (or ones(1,Q)). % Z - SxQ weighted input vectors. % N - SxQ net input vectors. % A - SxQ output vectors. % T - SxQ layer target vectors. % E - SxQ layer error vectors. % gW - SxR weight gradient with respect to performance. % gA - SxQ output gradient with respect to performance. % D - SxS neuron distances. % LP - Learning parameters, none, LP = []. % LS - Learning state, initially should be = []. % and returns, % dW - SxR weight (or bias) change matrix. % LS - New learning state. % % Learning occurs according to LEARNLV1's learning parameter, % shown here with its default value. % LP.lr - 0.01 - Learning rate % % LEARNLV2(CODE) returns useful information for each CODE string: % 'pnames' - Returns names of learning parameters. % 'pdefaults' - Returns default learning parameters. % 'needg' - Returns 1 if this function uses gW or gA. % % Examples % % Here we define a sample input P, output A, weight matrix W, and % output gradient gA for a layer with a 2-element input and 3 neurons. % We also define the learning rate LR. % % p = rand(2,1); % w = rand(3,2); % n = negdist(w,p); % a = compet(n); % gA = [-1;1; 1]; % lp.lr = 0.5; % % Since LEARNLV2 only needs these values to calculate a weight % change (see Algorithm below), we will use them to do so. % % dW = learnlv3(w,p,n,a,lp,Ttrain,C) % % Network Use % % You can create a standard network that uses LEARNLV2 with NEWLVQ. % % To prepare the weights of layer i of a custom network % to learn with LEARNLV1: % 1) Set NET.trainFcn to 'trainwb1'. % (NET.trainParam will automatically become TRAINWB1's default parameters.) % 2) Set NET.adaptFcn to 'adaptwb'. % (NET.adaptParam will automatically become TRAINWB1's default parameters.) % 3) Set each NET.inputWeights{i,j}.learnFcn to 'learnlv2'. % Set each NET.layerWeights{i,j}.learnFcn to 'learnlv2'. % (Each weight learning parameter property will automatically % be set to LEARNLV2's default parameters.) % % To train the network (or enable it to adapt): % 1) Set NET.trainParam (or NET.adaptParam) properties as desired. % 2) Call TRAIN (or ADAPT). % % Algorithm % % LEARNLV3 calculates the weight change dW for a given neuron from % the neuron's input P, output A, train vector target T train, output % conexion matrix C and learning rate LR % according to the OLVQ rule, given i the index of the neuron whose % output a(i) is 1: % % dw(i,:) = +lr*(p-w(i,:)) if C(:,i) = Ttrain % = -lr*(p-w(i,:)) if C(:,i) ~= Ttrain % % if C(:,i) ~= Ttrain then the index j is found of the neuron with the % greatest net input n(k), from the neurons whose C(:,k)=Ttrain. This % neuron's weights are updated as follows: % % dw(j,:) = +lr*(p-w(i,:)) % % See also LEARNLV1, ADAPTWB, TRAINWB, ADAPT, TRAIN. % Mark Beale, 11-31-97 % Copyright (c) 1992-1998 by The MathWorks, Inc. % $Revision: 1.1.1.1 $ % FUNCTION INFO % ============= if isstr(w) switch lower(w) case 'name' dw = 'Learning Vector Quantization 3'; case 'pnames' dw = {'lr';'window'}; case 'pdefaults' lp.lr = 0.01; lp.window = 0.25; dw = lp; case 'needg' dw = 1; otherwise error('NNET:Arguments','Unrecognized code.') end return end % CALCULATION % =========== [S,R] = size(w); Q = size(p,2); pt = p'; dw = zeros(S,R); % For each q... for q=1:Q % Find closest neuron k1 找到获胜神经元 nq = n(:,q); k1 = find(nq == max(nq)); k1 = k1(1); % Find next closest neuron k2 次获胜神经元 nq(k1) = -inf; k2 = find(nq == max(nq)); k2 = k2(1); % and if x falls into the window... d1 = abs(n(k1,q)); % Shorter distance d2 = abs(n(k2,q)); % Greater distance if d2/d1 > ((1-lp.window)/(1+lp.window)) % then move incorrect neuron away from input, % and the correct neuron towards the input ptq = pt(q,:); if gA(k1,q) ~= gA(k2,q) % indicate the incorrect neuron with i, the other with j if gA(k1,q) ~= 0 i = k1; j = k2; else i = k2; j = k1; end dw(i,:) = dw(i,:) - lp.lr*(ptq - w(i,:)); dw(j,:) = dw(j,:) + lp.lr*(ptq - w(j,:)); else dw(k1,:) = dw(k1,:) + 0.11*lp.window*(ptq-w(k1,:)); % dw(k2,:) = dw(k2,:) + 0.11*lp.window*(ptq-w(k2,:)); end end end