基于LSTM网络的视觉识别研究与实现——简化版

基于LSTM网络的视觉识别研究与实现详细版订阅本博客

https://blog.csdn.net/ccsss22/article/details/115429083

1.问题描述：

        以人脸图像的视觉识别为研究对象，研究了基于LSTM长短期记忆单元网络的视觉识别算法，通过使用卷积神经网络学习人脸图像的特征信息，然后使用LSTM网络建立序列知识，并生成描述性的句子，作为特征序列，建立一种基于句子描述的LSTM网络的视觉识别算法，最后通过MATLAB对该算法进行了仿真验证，对于不同姿态，不同干扰因素影响下的目标图像，本文所提出的基于LSTM的视觉识别正确率可以达到76%以上。
 

2.部分程序：
 
function nn = func_LSTM(train_x,train_y,test_x,test_y);

binary_dim     = 8;
largest_number = 2^binary_dim - 1;
binary         = cell(largest_number, 1);

for i = 1:largest_number + 1
    binary{i}      = dec2bin(i-1, binary_dim);
    int2binary{i}  = binary{i};
end

%input variables
alpha      = 0.000001;
input_dim  = 2;
hidden_dim = 32;
output_dim = 1;

%initialize neural network weights
%in_gate = sigmoid(X(t) * U_i + H(t-1) * W_i)
U_i        = 2 * rand(input_dim, hidden_dim) - 1;
W_i        = 2 * rand(hidden_dim, hidden_dim) - 1;
U_i_update = zeros(size(U_i));
W_i_update = zeros(size(W_i));

%forget_gate = sigmoid(X(t) * U_f + H(t-1) * W_f)
U_f        = 2 * rand(input_dim, hidden_dim) - 1;
W_f        = 2 * rand(hidden_dim, hidden_dim) - 1;
U_f_update = zeros(size(U_f));
W_f_update = zeros(size(W_f));

%out_gate    = sigmoid(X(t) * U_o + H(t-1) * W_o)
U_o = 2 * rand(input_dim, hidden_dim) - 1;
W_o = 2 * rand(hidden_dim, hidden_dim) - 1;
U_o_update = zeros(size(U_o));
W_o_update = zeros(size(W_o));

%g_gate      = tanh(X(t) * U_g + H(t-1) * W_g)
U_g = 2 * rand(input_dim, hidden_dim) - 1;
W_g = 2 * rand(hidden_dim, hidden_dim) - 1;
U_g_update = zeros(size(U_g));
W_g_update = zeros(size(W_g));

out_para = 2 * zeros(hidden_dim, output_dim) ;
out_para_update = zeros(size(out_para));
% C(t) = C(t-1) .* forget_gate + g_gate .* in_gate 
% S(t) = tanh(C(t)) .* out_gate                     
% Out  = sigmoid(S(t) * out_para)      

%train 
iter = 9999; % training iterations
for j = 1:iter
 
    % generate a simple addition problem (a + b = c)
    a_int = randi(round(largest_number/2));   % int version
    a     = int2binary{a_int+1};              % binary encoding
    
    b_int = randi(floor(largest_number/2));   % int version
    b     = int2binary{b_int+1};              % binary encoding
    
    % true answer
    c_int = a_int + b_int;                    % int version
    c     = int2binary{c_int+1};              % binary encoding
    
    % where we'll store our best guess (binary encoded)
    d     = zeros(size(c));
 
    
    % total error
    overallError = 0;
    
    % difference in output layer, i.e., (target - out)
    output_deltas = [];
    
    % values of hidden layer, i.e., S(t)
    hidden_layer_values = [];
    cell_gate_values    = [];
    % initialize S(0) as a zero-vector
    hidden_layer_values = [hidden_layer_values; zeros(1, hidden_dim)];
    cell_gate_values    = [cell_gate_values; zeros(1, hidden_dim)];
    
    % initialize memory gate
    % hidden layer
    H = [];
    H = [H; zeros(1, hidden_dim)];
    % cell gate
    C = [];
    C = [C; zeros(1, hidden_dim)];
    % in gate
    I = [];
    % forget gate
    F = [];
    % out gate
    O = [];
    % g gate
    G = [];
    
    % start to process a sequence, i.e., a forward pass
    % Note: the output of a LSTM cell is the hidden_layer, and you need to 
    for position = 0:binary_dim-1
        % X ------> input, size: 1 x input_dim
        X = [a(binary_dim - position)-'0' b(binary_dim - position)-'0'];
        % y ------> label, size: 1 x output_dim
        y = [c(binary_dim - position)-'0']';
        % use equations (1)-(7) in a forward pass. here we do not use bias
        in_gate     = sigmoid(X * U_i + H(end, :) * W_i);  % equation (1)
        forget_gate = sigmoid(X * U_f + H(end, :) * W_f);  % equation (2)
        out_gate    = sigmoid(X * U_o + H(end, :) * W_o);  % equation (3)
        g_gate      = tanh(X * U_g + H(end, :) * W_g);    % equation (4)
        C_t         = C(end, :) .* forget_gate + g_gate .* in_gate;    % equation (5)
        H_t         = tanh(C_t) .* out_gate;                          % equation (6)
        
        % store these memory gates
        I = [I; in_gate];
        F = [F; forget_gate];
        O = [O; out_gate];
        G = [G; g_gate];
        C = [C; C_t];
        H = [H; H_t];
        
        % compute predict output
        pred_out = sigmoid(H_t * out_para);
        
        % compute error in output layer
        output_error = y - pred_out;
        
        % compute difference in output layer using derivative
        % output_diff = output_error * sigmoid_output_to_derivative(pred_out);
        output_deltas = [output_deltas; output_error];
        
        % compute total error
        overallError = overallError + abs(output_error(1));
        
        % decode estimate so we can print it out
        d(binary_dim - position) = round(pred_out);
    end
    
    % from the last LSTM cell, you need a initial hidden layer difference
    future_H_diff = zeros(1, hidden_dim);
    
    % stare back-propagation, i.e., a backward pass
    % the goal is to compute differences and use them to update weights
    % start from the last LSTM cell
    for position = 0:binary_dim-1
        X = [a(position+1)-'0' b(position+1)-'0'];
        % hidden layer
        H_t = H(end-position, :);         % H(t)
        % previous hidden layer
        H_t_1 = H(end-position-1, :);     % H(t-1)
        C_t = C(end-position, :);         % C(t)
        C_t_1 = C(end-position-1, :);     % C(t-1)
        O_t = O(end-position, :);
        F_t = F(end-position, :);
        G_t = G(end-position, :);
        I_t = I(end-position, :);
        
        % output layer difference
        output_diff = output_deltas(end-position, :);
%         H_t_diff = (future_H_diff * (W_i' + W_o' + W_f' + W_g') + output_diff * out_para') ...
%                    .* sigmoid_output_to_derivative(H_t);

%         H_t_diff = output_diff * (out_para') .* sigmoid_output_to_derivative(H_t);
        H_t_diff = output_diff * (out_para') .* sigmoid_output_to_derivative(H_t);
        
%         out_para_diff = output_diff * (H_t) * sigmoid_output_to_derivative(out_para);
        out_para_diff =  (H_t') * output_diff;

        % out_gate diference
        O_t_diff = H_t_diff .* tanh(C_t) .* sigmoid_output_to_derivative(O_t);
        
        % C_t difference
        C_t_diff = H_t_diff .* O_t .* tan_h_output_to_derivative(C_t);
 
        % forget_gate_diffeence
        F_t_diff = C_t_diff .* C_t_1 .* sigmoid_output_to_derivative(F_t);
        
        % in_gate difference
        I_t_diff = C_t_diff .* G_t .* sigmoid_output_to_derivative(I_t);
        
        % g_gate difference
        G_t_diff = C_t_diff .* I_t .* tan_h_output_to_derivative(G_t);
        
        % differences of U_i and W_i
        U_i_diff =  X' * I_t_diff .* sigmoid_output_to_derivative(U_i);
        W_i_diff =  (H_t_1)' * I_t_diff .* sigmoid_output_to_derivative(W_i);
        
        % differences of U_o and W_o
        U_o_diff = X' * O_t_diff .* sigmoid_output_to_derivative(U_o);
        W_o_diff = (H_t_1)' * O_t_diff .* sigmoid_output_to_derivative(W_o);
        
        % differences of U_o and W_o
        U_f_diff = X' * F_t_diff .* sigmoid_output_to_derivative(U_f);
        W_f_diff = (H_t_1)' * F_t_diff .* sigmoid_output_to_derivative(W_f);
        
        % differences of U_o and W_o
        U_g_diff = X' * G_t_diff .* tan_h_output_to_derivative(U_g);
        W_g_diff = (H_t_1)' * G_t_diff .* tan_h_output_to_derivative(W_g);
        
        % update
        U_i_update = U_i_update + U_i_diff;
        W_i_update = W_i_update + W_i_diff;
        U_o_update = U_o_update + U_o_diff;
        W_o_update = W_o_update + W_o_diff;
        U_f_update = U_f_update + U_f_diff;
        W_f_update = W_f_update + W_f_diff;
        U_g_update = U_g_update + U_g_diff;
        W_g_update = W_g_update + W_g_diff;
        out_para_update = out_para_update + out_para_diff;
    end
 
    U_i = U_i + U_i_update * alpha; 
    W_i = W_i + W_i_update * alpha;
    U_o = U_o + U_o_update * alpha; 
    W_o = W_o + W_o_update * alpha;
    U_f = U_f + U_f_update * alpha; 
    W_f = W_f + W_f_update * alpha;
    U_g = U_g + U_g_update * alpha; 
    W_g = W_g + W_g_update * alpha;
    out_para = out_para + out_para_update * alpha;
    
    U_i_update = U_i_update * 0; 
    W_i_update = W_i_update * 0;
    U_o_update = U_o_update * 0; 
    W_o_update = W_o_update * 0;
    U_f_update = U_f_update * 0; 
    W_f_update = W_f_update * 0;
    U_g_update = U_g_update * 0; 
    W_g_update = W_g_update * 0;
    out_para_update = out_para_update * 0;
    
     
end
————————————————

A-05-40