CN105844653A - Multilayer convolution neural network optimization system and method - Google Patents

Abstract

The invention relates to a multilayer convolution neural network optimization system and a multilayer convolution neural network optimization method. The multilayer convolution neural network optimization system comprises an image positioning module, a sampling module based on CP decomposition, a micro-sampling module, a parameter tuning module based on a BP algorithm, and a convolution neural network feature output module, wherein the image positioning module sets a parameter matrix theta by means of a regression function according to dimensionality of a convolutional layer; the sampling module based on CP decomposition performs tensor decomposition on a result after convolution operation to obtain two rank-one tensors p and q; the micro-sampling module adopts a bilinear interpolation algorithm for carrying out linear interpolation on pixel points in different directions of an image, and obtains network output V; the parameter tuning module based on a BP algorithm updates a parameter theta; and the convolution neural network feature output module is used for introducing an updated parameter theta<hat> into a network, carrying out iterative operation and outputting convolution neural network features. The multilayer convolution neural network optimization system and the multilayer convolution neural network optimization method are conductive to extracting space-invariant features, and improving operational efficiency.

Description

A kind of multilayer convolutional neural networks optimizes system and method

Technical field

The application relates to nerual network technique field, particularly to a kind of multilayer convolutional Neural net Network optimizes system and method.

Background technology

Convolutional neural networks (convolutional neural network, CNN) is a kind of feedforward Neutral net, the god unlike traditional algorithm, between the adjacent layer of convolutional neural networks It is not full connection through unit, but part connects, and for the convolution of a convolution kernel Computing weights are shared, thus decrease number of parameters, are reached with pond process by multiple convolution Purpose to feature extraction.Utilize convolution can realize image blurring process, rim detection Thus beneficially feature extraction, utilize pond computing more easily image to be dropped Dimension, thus reduce parameter and amount of calculation.

In tradition convolutional neural networks, network design is convolution pond, convolution pond Structure, need to set multilayer and just can extract the feature of space invariance.

Summary of the invention

This application provides a kind of multilayer convolutional neural networks and optimize system and method, to solve In prior art, network structure needs to set the skill of the feature that multilayer just can extract space invariance Art problem.

In order to solve the problems referred to above, the technical scheme is that

Embodiments provide a kind of multilayer convolutional neural networks and optimize system, including: Framing module, based on CP decompose sampling module, micro-sampling module, based on BP algorithm Arameter optimization module and convolutional neural networks feature output module, described framing module is led to Cross the regression function dimension set parameter matrix θ according to convolutional layer, and for next layer for The operation of coordinate transform；The described sampling module decomposed based on CP is to the knot through convolution operation Fruit carries out tensor resolution, obtains two order one tensor p and q, utilizes pixel corresponding simultaneously The parameter θ that coordinate produces with last layer carries out computing；Described micro-sampling module application bilinearity Interpolation algorithm carries out linear interpolation to the pixel of image different directions, it is thus achieved that network exports V；Described arameter optimization module based on BP algorithm is updated for parameter θ；Described convolution Neural network characteristics output module is for bringing the parameter θ ^ updated into network, and carries out repeatedly For computing, it is simultaneously introduced grader and feature is carried out classification prediction, export convolutional neural networks Feature.

The technical scheme that the embodiment of the present invention is taked also includes: described framing module is also used In receiving from result U after last convolution, wherein U is by carrying out with sample convolution kernel Image convolution calculates and obtains.

The technical scheme that the embodiment of the present invention is taked also includes: described framing module is passed through Regression function specifically includes according to the dimension set parameter matrix θ of convolutional layer: for convolution god A certain feature output U ∈ R through network^H*W*CAs the input of sample level, wherein, H, W The height of representative picture and width, C represents passage；Spatial transformation parameter is calculated by floc θ, floc comprise dimension set parameter matrix θ: the θ=f returning layer according to convolutional layer_loc(U)。

The technical scheme that the embodiment of the present invention is taked also includes: the application of described micro-sampling module is double Linear interpolation algorithm carries out linear interpolation to the pixel of image different directions, it is thus achieved that network is defeated Go out V particularly as follows: utilize the parameter θ tried to achieve to do imitative to the pixel respective coordinates (xt, yt) of U Penetrate conversion, obtain (xs, ys)；The coordinate (xs, ys) of pixel and p ° of q will be obtained through tensor computation It is input to micro-sample level, by bilinear interpolation algorithm, surrounding pixel is carried out in micro-sample level Compression substitutes, it is thus achieved that network output V.

The technical scheme that the embodiment of the present invention is taked also includes: described parameter based on BP algorithm The concrete mode that tuning module is updated for θ is: utilize network output V to input network Order one p ° of q of exterior product of tensors seeks the sensitivity δ that local derviation obtains, and updates ginseng according to sensitivity δ Number.

The technical scheme that the embodiment of the present invention is taked also includes: described convolutional neural networks feature Output module is for by the output U of the space transformer through L iteration^LEnter softmax layer Classify, obtain the class probability of each class；Utilize the error function parameter to convolutional layer W seeks local derviation, successively undated parameter W, brings the W updated into each convolutional layer and again counts Calculate.

The embodiment of the present invention additionally provides a kind of multilayer convolutional neural networks optimization method, bag Include:

Step a: obtained result U of a certain convolutional layer by convolutional calculation, pass through framing Module obtain result U, using result U as sample level input calculate spatial transformation parameter θ；

Step b: U is carried out tensor resolution, it is thus achieved that two order one tensor p and q, utilization is tried to achieve Parameter θ pixel respective coordinates is done affine transformation, pixel will be obtained through tensor computation Coordinate and p ° of q be input to micro-sample level；

Step c: be combined with tensor operation by bilinear interpolation algorithm in micro-sample level, it is thus achieved that Network output V；V is asked the local derviation of order one p ° of q of tensor product, updates ginseng according to sensitivity δ Number θ；

Step d: the whole network of iterative computation, output convolutional neural networks feature output.

The technical scheme that the embodiment of the present invention is taked also includes: by convolution meter in described step a Calculate and obtain result U of a certain convolutional layer particularly as follows: the sample of convolution kernel and X*Y for d*d Carry out image convolution and calculate result U obtaining a certain convolutional layer.

The technical scheme that the embodiment of the present invention is taked also includes: in described step d, iterative computation is whole Individual network includes: bring the parameter updated into network, it is thus achieved that network output assignment, repeats Step a, to step c, is iterated computing.

The technical scheme that the embodiment of the present invention is taked also includes: described step d also includes: for Output U through the space transformer of L iteration^LEnter softmax layer to classify, obtain The class probability of each class；Definition quadratic loss function, utilizes error function to convolutional layer Parameter W seeks local derviation, successively undated parameter W；Bring the W updated into each convolutional layer again Calculate, it is thus achieved that the final network of CNN exports.

The multilayer convolutional neural networks of the embodiment of the present invention optimizes system and method through tensor Space transformer is added in convolutional layer output after decomposition, substitutes traditional pond layer, is conducive to Extract the feature of space invariance；Output simultaneously for convolutional layer carries out tensor resolution, Bigization retains under raw information and structural dependence premise, removes redundancy, improves computing Efficiency；Application bilinear interpolation algorithm is combined with tensor, the pixel to image different directions Carry out linear interpolation, between any two known pixels point, insert pixel, when amplifying Avoid distortion, replacement pixels point during dimensionality reduction, can be tried to achieve；Utilize the low order tensor after tensor resolution It is compressed processing to pixel in conjunction with bilinear interpolation algorithm, reduces computation complexity also Scaling at image processes and maintains antialiasing function.

Accompanying drawing explanation

Fig. 1 is the structural representation of the multilayer convolutional neural networks optimization system of the embodiment of the present invention Figure；

Fig. 2 is the flow chart of the multilayer convolutional neural networks optimization method of the embodiment of the present invention.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below tie Close drawings and Examples, the present invention is further elaborated.Should be appreciated that herein Described specific embodiment, only in order to explain the present invention, is not intended to limit the present invention.

Refer to the multilayer convolutional neural networks that Fig. 1, Fig. 1 are the embodiment of the present invention and optimize system Structural representation.The multilayer convolutional neural networks of the embodiment of the present application optimizes system and includes figure As locating module, the sampling module that decomposes based on CP, micro-sampling module, based on BP algorithm Arameter optimization module and convolutional neural networks feature output module.Framing module can set After convolutional layer, it is also possible to accept original image, framing module passes through regression function Dimension set parameter matrix θ according to convolutional layer, and for next layer for coordinate transform Operation；Sampling module based on CP decomposition carries out tensor to the result through convolution operation and divides Solve, obtain two order one tensor p and q, utilize coordinate and last layer that pixel is corresponding simultaneously The parameter θ produced carries out computing, particularly as follows: utilize the parameter θ the tried to achieve pixel to U Respective coordinates (xt, yt) does affine transformation, obtains (xs, ys):

$\left(\begin{array}{c}{x}^s\\ y^s\end{array}\right)=F_{\mathrm{\&theta;}}\left(U\right)=\left|\begin{array}{ccc}{\mathrm{\&theta;}}_{11}& {\mathrm{\&theta;}}_{12}& {\mathrm{\&theta;}}_{13}\\ {\mathrm{\&theta;}}_{21}& {\mathrm{\&theta;}}_{22}& {\mathrm{\&theta;}}_{23}\end{array}\right|\left(\begin{array}{c}{x}^t\\ y^t\end{array}\right)$

Tensor theories is a subdiscipline of mathematics, and tensor concept is pushing away of vector concept Extensively, vector is single order tensor, and tensor is one and can be used to represent at some vectors, scalar sum The polyteny function of the linear relationship between other tensors, the sample expressed for tensor form Prototype structure can be retained to the full extent, and then extract in feature extraction phases more robust Feature.

The pixel of image different directions is carried out by micro-sampling module application bilinear interpolation algorithm Linear interpolation, it is thus achieved that network output V, inserts pixel between any two known pixels point Point, avoids distortion when amplifying, can try to achieve replacement pixels point during dimensionality reduction.Wherein, bilinearity Interpolation algorithm: bilinear interpolation, is also called bilinear interpolation.Bilinear interpolation is to have two The linear interpolation extension of the interpolating function of variable, its core concept is to enter respectively in both direction Row once linear interpolation, is widely used in signal transacting, and digital picture and Video processing etc. should In with.

Arameter optimization module based on BP algorithm is updated for θ, concrete update mode For: utilize network output V to ask local derviation to obtain network input order one p ° of q of exterior product of tensors Sensitivity δ, according to sensitivity δ undated parameter θ；

Convolutional neural networks feature output module networks for the parameter θ ^ band that will have updated Network, and it is iterated computing, it is simultaneously introduced grader and feature is carried out classification prediction, output Convolutional neural networks feature.

Refer to the multilayer convolutional neural networks optimization side that Fig. 2, Fig. 2 are the embodiment of the present invention The flow chart of method.The multilayer convolutional neural networks optimization method of the embodiment of the present invention includes:

Step 10: image convolution calculating is carried out for the convolution kernel of d*d and the sample of X*Y Obtain result U of a certain convolutional layer；

In step 10, convolution kernel:

Sample X:

Obtain and export U:

Wherein to convolution results Um the most once, n

$U_{m,n}=\underset{j=n}{\overset{n+d-1}{\&Sigma;}}\underset{i=m}{\overset{m+d-1}{\&Sigma;}}{w}_{i-m+1j-n+1}*x_{ij}$

Step 20: framing module receives from result U after last convolution.

Step 30: a certain feature for CNN exports U ∈ R^H*W*C(H, W representative picture Height and width, C represents passage), as the input of sample level, calculate space by floc and become Changing parameter θ, floc comprises recurrence layer and the parameter of convolutional layer becomes θ matrix:

θ=f_loc(U)

Obtaining result is:

$\left|\begin{array}{ccc}{\mathrm{\&theta;}}_{11}& {\mathrm{\&theta;}}_{12}& {\mathrm{\&theta;}}_{13}\\ {\mathrm{\&theta;}}_{21}& {\mathrm{\&theta;}}_{22}& {\mathrm{\&theta;}}_{23}\end{array}\right|$

Step 40: result U after convolution is carried out tensor resolution, obtains two order one tensors P and q:

In step 40, in tradition convolutional neural networks, network design is convolution pond The structure in convolution pond, needs to set multilayer and just can extract the feature of space invariance. And in the multilayer convolutional neural networks optimization method of this embodiment of the present invention, through tensor Add space transformer after convolutional layer output after decomposition, substitute traditional pond layer, favorably In the feature extracting space invariance；Output simultaneously for convolutional layer carries out tensor resolution, Maximize and retain under raw information and structural dependence premise, remove redundancy, improve fortune Calculate efficiency.

Step 50: utilize the parameter θ tried to achieve that the pixel respective coordinates (xt, yt) of U is done Affine transformation, obtains (xs, ys):

$\left(\begin{array}{c}{x}^s\\ y^s\end{array}\right)=F_{\mathrm{\&theta;}}\left(U\right)=\left|\begin{array}{ccc}{\mathrm{\&theta;}}_{11}& {\mathrm{\&theta;}}_{12}& {\mathrm{\&theta;}}_{13}\\ {\mathrm{\&theta;}}_{21}& {\mathrm{\&theta;}}_{22}& {\mathrm{\&theta;}}_{23}\end{array}\right|\left(\begin{array}{c}{x}^t\\ y^t\end{array}\right)$

Step 60: the coordinate (xs, ys) obtaining pixel through tensor computation is input to p ° of q Micro-sample level, is compressed replacing to surrounding pixel by bilinear interpolation algorithm in micro-sample level Generation, it is thus achieved that network output V.

In a step 60, micro-sample level application bilinear interpolation algorithm is combined with tensor, to figure As the pixel of different directions carries out linear interpolation, interleaving of any two known pixels point Enter pixel, avoid distortion when amplifying, replacement pixels point during dimensionality reduction, can be tried to achieve, at this specially Profit is be combined with tensor by bilinear interpolation algorithm to be compressed pixel processing, fall While low computation complexity, in the scaling of image processes, maintain antialiasing function

Step 70: utilize network output V that network input order one p ° of q of exterior product of tensors is asked inclined Lead the sensitivity δ of acquisition, according to sensitivity δ undated parameter θ.

$\frac{\&part;x}{\&part;{\mathrm{\&theta;}}_i}=\left(\begin{array}{c}{x}^t\\ y^t\end{array}\right){\mathrm{\&delta;}}_i$

$\mathrm{\&Delta;}{\left(\mathrm{\&theta;}\right)}^I=-\mathrm{\&eta;}\frac{\&part;x}{\&part;{\left(\mathrm{\&theta;}\right)}^I}$

$\frac{\&part;y}{\&part;{\mathrm{\&theta;}}_i}=\left(\begin{array}{c}{x}^t\\ y^t\end{array}\right){\mathrm{\&delta;}}_i$

$\mathrm{\&Delta;}{\left(\mathrm{\&theta;}\right)}^I=-\mathrm{\&eta;}\frac{\&part;y}{\&part;{\left(\mathrm{\&theta;}\right)}^I}$

Step 80: bring the parameter θ ^ updated into network, it is thus achieved that network output is entered as U2, as the input of the first step, repeats step 10 and arrives step 70, be iterated computing.

In step 80, after space transformer regulates parameter, can arbitrarily add and convolution The optional position of neutral net, after can being arranged on original sample input layer, it is also possible to arranges After convolutional layer (sample-convolution-space transformer ... or sample-space transformer-convolution ...) Finally realizing effect is the feature that different space transformers can extract different levels, such as One space transformer can extract the feature of bird head, second feature etc. that can extract bird body Deng.

Step 90: the output UL for the space transformer through L iteration enters Softmax layer is classified, and obtains the class probability of each class.

$Y=h_{\mathrm{\&theta;}}\left(U^{L\left(i\right)}\right)=\left[\begin{array}{c}p(y^{\left(i\right)}=1|U^L\left(i\right);\mathrm{\&theta;})\\ p(y^{\left(i\right)}=2|U^L\left(i\right);\mathrm{\&theta;})\\ \mathrm{......}\\ p(y^{\left(i\right)}=k|U^L\left(i\right);\mathrm{\&theta;})\end{array}\right]=\frac{1}{{\&Sigma;}_{j=1}^k{e_j}^{{\mathrm{\&theta;}}^T{U}^L\left(i\right)}}\left[\begin{array}{c}{e_1}^{{\mathrm{\&theta;}}^T{U}^L\left(i\right)}\\ {e_2}^{{\mathrm{\&theta;}}^T{U}^L\left(i\right)}\\ \mathrm{...}\\ {e_k}^{{\mathrm{\&theta;}}^T{U}^L\left(i\right)}\end{array}\right]$

Step 100: definition quadratic loss function,Represent the label that the n-th sample is corresponding Kth is tieed up,Represent the kth output of the network output that the n-th sample is corresponding

$E^n=\frac{1}{2}\underset{k=1}{\overset{c}{\&Sigma;}}(t_k^n-y_k^n)=\frac{1}{2}||t^n-y^n||$

Step 110: utilize error function that parameter W of convolutional layer is asked local derviation, successively update Parameter W.

$\frac{\&part;E}{\&part;{\left(W\right)}^I}={\left(W\right)}^{I-1}{\left(\mathrm{\&delta;}\right)}^T$

$\mathrm{\&Delta;}{\left(W\right)}^I=-\mathrm{\&eta;}\frac{\&part;E}{\&part;{\left(W\right)}^I}$

Step 120: bring the W updated into each convolutional layer and recalculate, it is thus achieved that CNN is Whole network output, the feature i.e. extracted.

The multilayer convolutional neural networks of the embodiment of the present invention optimizes system and method through tensor Space transformer is added in convolutional layer output after decomposition, substitutes traditional pond layer, is conducive to Extract the feature of space invariance；Output simultaneously for convolutional layer carries out tensor resolution, Bigization retains under raw information and structural dependence premise, removes redundancy, improves computing Efficiency；Application bilinear interpolation algorithm is combined with tensor, the pixel to image different directions Carry out linear interpolation, between any two known pixels point, insert pixel, when amplifying Avoid distortion, replacement pixels point during dimensionality reduction, can be tried to achieve；Utilize the low order tensor after tensor resolution It is compressed processing to pixel in conjunction with bilinear interpolation algorithm, reduces computation complexity also Scaling at image processes and maintains antialiasing function.

Described above to the disclosed embodiments, enables professional and technical personnel in the field real Now or use the present invention.To the multiple amendment of these embodiments professional technique people to this area Will be apparent from for Yuan, generic principles defined herein can be without departing from this In the case of the spirit or scope of invention, realize in other embodiments.Therefore, the present invention It is not intended to be limited to the embodiments shown herein, and is to fit to disclosed herein Principle and the consistent the widest scope of features of novelty.

Claims (10)

1. a multilayer convolutional neural networks optimizes system, it is characterised in that including: framing mould Block, based on CP decompose sampling module, micro-sampling module, based on BP algorithm parameter tuning module and volume Long-pending neural network characteristics output module, described framing module passes through the regression function dimension according to convolutional layer Degree setup parameter matrix θ, and it is used for next layer operation for coordinate transform；Described based on CP decomposition Sampling module carries out tensor resolution to the result through convolution operation, obtains two order one tensor p and q, The parameter θ simultaneously utilizing coordinate that pixel is corresponding and last layer to produce carries out computing；Described micro-sampling mould Block application bilinear interpolation algorithm carries out linear interpolation to the pixel of image different directions, it is thus achieved that network is defeated Go out V；Described arameter optimization module based on BP algorithm is updated for parameter θ；Described convolutional Neural Network characterization output module is for bringing the parameter θ ^ updated into network, and is iterated computing, simultaneously Add grader and feature is carried out classification prediction, export convolutional neural networks feature.

Multilayer convolutional neural networks the most according to claim 1 optimizes system, it is characterised in that institute Stating framing module to be additionally operable to receive result U from after last convolution, wherein U is by convolution Core and sample carry out image convolution and calculate acquisition.

Multilayer convolutional neural networks the most according to claim 1 and 2 optimizes system, and its feature exists In, described framing module is concrete according to the dimension set parameter matrix θ of convolutional layer by regression function Including: a certain feature for convolutional neural networks exports U ∈ R^H*W*CAs the input of sample level, its In, the height of H, W representative picture and width, C represents passage；Spatial transformation parameter is calculated by floc θ, floc comprise dimension set parameter matrix θ: the θ=f returning layer according to convolutional layer_loc(U)。

Multilayer convolutional neural networks the most according to claim 1 optimizes system, it is characterised in that institute State micro-sampling module application bilinear interpolation algorithm and the pixel of image different directions carried out linear interpolation, Obtain network output V particularly as follows: utilize the parameter θ tried to achieve that the pixel respective coordinates (xt, yt) of U is done Affine transformation, obtains (xs, ys)；The coordinate (xs, ys) obtaining pixel through tensor computation is input to p o q Micro-sample level, is compressed substituting to surrounding pixel by bilinear interpolation algorithm in micro-sample level, it is thus achieved that Network output V.

Multilayer convolutional neural networks the most according to claim 1 optimizes system, it is characterised in that The concrete mode that described arameter optimization module based on BP algorithm is updated for θ is: utilize network defeated Go out the sensitivity δ that network input order one exterior product of tensors p o q is asked local derviation to obtain by V, according to sensitivity δ Undated parameter.

Multilayer convolutional neural networks the most according to claim 1 optimizes system, it is characterised in that: Described convolutional neural networks feature output module is for by the output of the space transformer through L iteration U^LEnter softmax layer to classify, obtain the class probability of each class；Utilize error function to convolution Parameter W of layer seeks local derviation, and successively undated parameter W is brought the W updated into each convolutional layer and again counted Calculate.

7. a multilayer convolutional neural networks optimization method, including:

Step a: obtained result U of a certain convolutional layer by convolutional calculation, is obtained by framing module Result U, using result U as sample level input calculate spatial transformation parameter θ；

Step b: U is carried out tensor resolution, it is thus achieved that two order one tensor p and q, utilizes the ginseng tried to achieve Pixel respective coordinates is done affine transformation by number θ, will obtain coordinate and the p o q of pixel through tensor computation It is input to micro-sample level；

Step c: be combined with tensor operation by bilinear interpolation algorithm in micro-sample level, it is thus achieved that network is defeated Go out V；V is asked the local derviation of order one tensor product p o q, according to sensitivity δ undated parameter θ；

Step d: the whole network of iterative computation, output convolutional neural networks feature output.

Multilayer convolutional neural networks optimization method the most according to claim 7, it is characterised in that Described step a obtains result U of a certain convolutional layer by convolutional calculation particularly as follows: for the volume of d*d The sample of long-pending core and X*Y carries out image convolution and calculates result U obtaining a certain convolutional layer.

9. optimizing system according to the multilayer convolutional neural networks described in claim 7 or 8, its feature exists In, in described step d, the whole network of iterative computation includes: bring the parameter updated into network, it is thus achieved that Network output assignment, repetition step a, to step c, is iterated computing.

Multilayer convolutional neural networks the most according to claim 9 optimizes system, it is characterised in that Described step d also includes: for the output U of the space transformer through L iteration^LEnter softmax layer Classify, obtain the class probability of each class；Definition quadratic loss function, utilizes error function to volume Parameter W of lamination seeks local derviation, successively undated parameter W；Bring the W updated into each convolutional layer again to count Calculate, it is thus achieved that the final network of CNN exports.