1.軟件版本

matlab2021a

2.本算法理論知識(shí)

??? 長(zhǎng)短時(shí)記憶模型LSTM是由Hochreiter等人在1997年首次提出的，其主要原理是通過(guò)一種特殊的神經(jīng)元結(jié)構(gòu)用來(lái)長(zhǎng)時(shí)間存儲(chǔ)信息。LSTM網(wǎng)絡(luò)模型的基本結(jié)構(gòu)如下圖所示：

圖1 LSTM網(wǎng)絡(luò)的基本結(jié)構(gòu)

??? 從圖1的結(jié)構(gòu)圖可知，LSMT網(wǎng)絡(luò)結(jié)構(gòu)包括輸入層，記憶模塊以及輸出層三個(gè)部分，其中記憶模塊由輸入門（Input Gate）、遺忘門（Forget Gate）以及輸出門（Output Gate）。LSTM模型通過(guò)這三個(gè)控制門來(lái)控制神經(jīng)網(wǎng)絡(luò)中所有的神經(jīng)元的讀寫操作。

??? LSTM模型的基本原理是通過(guò)多個(gè)控制門來(lái)抑制RNN神經(jīng)網(wǎng)絡(luò)梯度消失的缺陷。通過(guò)LSTM模型可以在較長(zhǎng)的時(shí)間內(nèi)保存梯度信息，延長(zhǎng)信號(hào)的處理時(shí)間，因此LSTM模型適合處理各種頻率大小的信號(hào)以及高低頻混合信號(hào)。LSTM模型中的記憶單元中輸入門（Input Gate）、遺忘門（Forget Gate）以及輸出門（Output Gate）通過(guò)控制單元組成非線性求和單元。其中輸入門、遺忘門以及輸出門三個(gè)控制門的激活函數(shù)為Sigmoid函數(shù)，通過(guò)該函數(shù)實(shí)現(xiàn)控制門“開(kāi)”和“關(guān)”狀態(tài)的改變。

??? 下圖為L(zhǎng)STM模型中記憶模塊的內(nèi)部結(jié)構(gòu)圖：

?

圖2 LSTM網(wǎng)絡(luò)的記憶單元內(nèi)部結(jié)構(gòu)

??? 從圖2的結(jié)構(gòu)圖可知，LSTM的記憶單元的工作原理為，當(dāng)輸入門進(jìn)入”開(kāi)“狀態(tài)，那么外部信息由記憶單元讀取信息，當(dāng)輸入門進(jìn)入“關(guān)”狀態(tài)，那么外部信息無(wú)法進(jìn)入記憶單元。同理，遺忘門和輸出門也有著相似的控制功能。LSTM模型通過(guò)這三個(gè)控制門將各種梯度信息長(zhǎng)久的保存在記憶單元中。當(dāng)記憶單元進(jìn)行信息的長(zhǎng)時(shí)間保存的時(shí)候，其遺忘門處于“開(kāi)”狀態(tài)，輸入門處于“關(guān)”狀態(tài)。

??? 當(dāng)輸入門進(jìn)入“開(kāi)”狀態(tài)之后，記憶單元開(kāi)始接受到外部信息并進(jìn)行存儲(chǔ)。當(dāng)輸入門進(jìn)入“關(guān)”狀態(tài)之后，記憶單元暫停接受外部信息，同時(shí)，輸出門進(jìn)入“開(kāi)”狀態(tài)，記憶單元中保存的信息傳輸?shù)胶笠粚印６z忘門的功能則是在必要的時(shí)候?qū)ι窠?jīng)元的狀態(tài)進(jìn)行重置。

??? 對(duì)于LSTM網(wǎng)絡(luò)模型的前向傳播過(guò)程，其涉及到的各個(gè)數(shù)學(xué)原理如下：

?

?2.遺忘門計(jì)算過(guò)程如下所示：

? ? ? ?

?3.記憶單元計(jì)算過(guò)程如下所示：

?4.輸出門計(jì)算過(guò)程如下所示：

?5.記憶單元輸出計(jì)算過(guò)程如下所示：

對(duì)于LSTM網(wǎng)絡(luò)模型的反向傳播過(guò)程，其涉及到的各個(gè)數(shù)學(xué)原理如下：

?6.輸入門計(jì)算過(guò)程如下所示：

??? 基于LSTM網(wǎng)絡(luò)的視覺(jué)識(shí)別算法，其整體算法流程圖如下圖所示：

????????????????????????? ??????

?

圖3基于LSTM網(wǎng)絡(luò)的視覺(jué)識(shí)別算法流程圖

根據(jù)圖3的算法流程圖，本文所要研究的基于LSTM網(wǎng)絡(luò)的視覺(jué)識(shí)別算法步驟為：

??? 步驟一：圖像的采集，本文以人臉圖像為研究對(duì)象。

??? 步驟二：圖像預(yù)處理，根據(jù)本章2節(jié)的內(nèi)容對(duì)所需要識(shí)別的視覺(jué)圖像進(jìn)行預(yù)處理，獲得較為清晰的圖像。

??? 步驟三：圖像分割，將圖像進(jìn)行分割，分割大小根據(jù)采集圖像的識(shí)別目標(biāo)和整體場(chǎng)景大小關(guān)系進(jìn)行確定，將原始的圖像分割為大小的子圖像。

??? 步驟四：子圖幾何元素提取，通過(guò)邊緣提取方法，獲得每個(gè)子圖中所包含的幾何元素，并將各個(gè)幾何元素構(gòu)成句子信息。

??? 步驟五：將句子信息輸入到LSTM網(wǎng)絡(luò)，這個(gè)步驟也是核心環(huán)節(jié)，下面對(duì)LSTM網(wǎng)絡(luò)的識(shí)別過(guò)程進(jìn)行介紹。首先，將句子信息通過(guò)LSTM的輸入層輸入到LSTM網(wǎng)絡(luò)中，基本結(jié)構(gòu)圖如下圖所示：

圖3基于LSTM網(wǎng)絡(luò)的識(shí)別結(jié)構(gòu)圖

??? 這里假設(shè)LSTM某一時(shí)刻的輸入特征信息和輸出結(jié)果為和，其記憶模塊中的輸入和輸出為和，和表示LSTM神經(jīng)元的激活函數(shù)的輸出和隱含層的輸出，整個(gè)LSTM的訓(xùn)練流程為：

3.核心代碼

---

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 + 1binary{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 versiona = int2binary{a_int+1}; % binary encodingb_int = randi(floor(largest_number/2)); % int versionb = int2binary{b_int+1}; % binary encoding% true answerc_int = a_int + b_int; % int versionc = int2binary{c_int+1}; % binary encoding% where we'll store our best guess (binary encoded)d = zeros(size(c));% total erroroverallError = 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-vectorhidden_layer_values = [hidden_layer_values; zeros(1, hidden_dim)];cell_gate_values = [cell_gate_values; zeros(1, hidden_dim)];% initialize memory gate% hidden layerH = [];H = [H; zeros(1, hidden_dim)];% cell gateC = [];C = [C; zeros(1, hidden_dim)];% in gateI = [];% forget gateF = [];% out gateO = [];% g gateG = [];% 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_dimX = [a(binary_dim - position)-'0' b(binary_dim - position)-'0'];% y ------> label, size: 1 x output_dimy = [c(binary_dim - position)-'0']';% use equations (1)-(7) in a forward pass. here we do not use biasin_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 gatesI = [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 outputpred_out = sigmoid(H_t * out_para);% compute error in output layeroutput_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 erroroverallError = overallError + abs(output_error(1));% decode estimate so we can print it outd(binary_dim - position) = round(pred_out);end% from the last LSTM cell, you need a initial hidden layer differencefuture_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 cellfor position = 0:binary_dim-1X = [a(position+1)-'0' b(position+1)-'0'];% hidden layerH_t = H(end-position, :); % H(t)% previous hidden layerH_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 differenceoutput_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 diferenceO_t_diff = H_t_diff .* tanh(C_t) .* sigmoid_output_to_derivative(O_t);% C_t differenceC_t_diff = H_t_diff .* O_t .* tan_h_output_to_derivative(C_t);% forget_gate_diffeenceF_t_diff = C_t_diff .* C_t_1 .* sigmoid_output_to_derivative(F_t);% in_gate differenceI_t_diff = C_t_diff .* G_t .* sigmoid_output_to_derivative(I_t);% g_gate differenceG_t_diff = C_t_diff .* I_t .* tan_h_output_to_derivative(G_t);% differences of U_i and W_iU_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_oU_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_oU_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_oU_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);% updateU_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;endU_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;endnn = newgrnn(train_x',train_y(:,1)',mean(mean(abs(out_para)))/2);

4.操作步驟與仿真結(jié)論

??? 通過(guò)本文的LSTM網(wǎng)絡(luò)識(shí)別算法，對(duì)不同干擾大小采集得到的人臉進(jìn)行識(shí)別，其識(shí)別正確率曲線如下圖所示：

?

??? 從圖2的仿真結(jié)果可知，隨著對(duì)采集圖像干擾的減少，本文所研究的LSTM識(shí)別算法具有最好的識(shí)別準(zhǔn)確率，RNN神經(jīng)網(wǎng)絡(luò)與基于卷積的深度神經(jīng)網(wǎng)絡(luò)，其識(shí)別率相當(dāng)，普通的神經(jīng)網(wǎng)絡(luò)，其識(shí)別率性能明顯較差。具體的識(shí)別率大小如下表所示：

表1 四種對(duì)比算法的識(shí)別率

算法	-15db	-10db	-5db	0db	5db	10db	15db
NN	17.5250	30.9500	45.0000	52.6000	55.4750	57.5750	57.6000
RBM	19.4000	40.4500	58.4750	67.9500	70.4000	72.2750	71.8750
RNN	20.6750	41.1500	60.0750	68.6000	72.5500	73.3500	73.3500
LSTM	23.1000	46.3500	65.0250	72.9500	75.6000	76.1000	76.3250

5.參考文獻(xiàn)

[01]米良川,楊子夫,李德升等.自動(dòng)機(jī)器人視覺(jué)控制系統(tǒng)[J].工業(yè)控制計(jì)算機(jī).2003.3.

[02]Or1ando,Fla.Digital Image Processing Techniques.Academic Pr,Inc.1984

[03]K.Fukushima.A neural network model for selective attention in visual pattern recognition. Biological Cybernetics[J]October 1986?55(1):5-15.

[04]T.H.Hidebrandt Optimal Training of Thresholded Linear Correlation Classifiers[J]. IEEE Transaction Neural Networks.1991?2(6):577-588.

[05]Van Ooyen B.Nienhuis Pattern Recognition in the Neocognitron Is Improved by Neural Adaption[J].Biological Cybernetics.1993,70:47-53.

[06]Bao Qing Li BaoXinLi. Building pattern classifiers using convolutional neural networks[J]. Neural.Networks?vol.5(3): 3081-3085.

[07]E S ackinger?,B boser,Y lecun?,L jaclel. Application of the ANNA Neural Network Chip to High Speed Character Recognition[J]. IEEE Transactions on Neural Networks 1992.3:498-505.

A05-40

6.完整源碼獲得方式

方式1：微信或者QQ聯(lián)系博主

方式2：訂閱MATLAB/FPGA教程，免費(fèi)獲得教程案例以及任意2份完整源碼

總結(jié)

以上是生活随笔為你收集整理的【LSTM】基于LSTM网络的人脸识别算法的MATLAB仿真的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問(wèn)題。

如果覺(jué)得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。