WO2019151573A1 - Computing method for multi-cancer identification and computing device using same - Google Patents

Abstract

The present invention relates to a computing method for identifying a multi-cancer, characterized by performing binary classification on information on a biomarker group diagnosing k cancers so as to identify whether the information on biomarkers diagnosing k cancers corresponds to a cancer, and when a result of the binary classification on information on a biomarker group diagnosing k cancers is identified to be a cancer, the information on the biomarker group diagnosing k cancers is analyzed by using neural networks such that the information on the biomarker group diagnosing k cancers is identified as corresponding to a cancer from among the k cancers.

Description

Computing method for multi-arm discrimination and computing device using same

The present invention relates to a computing method and a computing device using the same for multi-cancer discrimination. More specifically, the information of the k cancer diagnostic biomarker is binary cancer by binary clashing information of k cancer diagnostic biomarkers. If it is determined that the cancer is a result of the binary classification of the information of k cancer diagnostic biomarkers, and k cancer diagnostic biomarker information is analyzed using a neural network, the information of k cancer diagnostic biomarkers is k A computing method for determining a multi-arm for determining which of the cancers corresponds to a cancer, and a computing device using the same.

Tumor metastasis is an important part of cancer-related deaths in patients with solid cancer, where a portion of the tumor falls off and travels through the blood to other parts of the body. Currently, a common method of diagnosing cancer is a biopsy in which a portion of tissue is removed from an initial metastasis state, which is difficult to determine an accurate biopsy site. On the other hand, liquid biopsy method, which is attracting attention in recent years, collects biological samples derived from the patient's body such as blood and urine of the patient and detects tumor cells in the biological sample. It is not only able to measure progress but also is useful for early detection and diagnosis.

As such, an indicator that can detect changes in the body using proteins, nucleic acids, metabolites, and the like contained in biological samples is called a biomarker.

For example, conventional biomarker techniques for early diagnosis of lung cancer include those that include A1AT, IGF-1, RANTES and TTR as disclosed in Korean Patent No. 10-1463588.

However, only a single biomarker has a limitation in diagnosing cancer. Currently, a method of diagnosing cancer using a complex biomarker that improves the sensitivity and specificity of cancer by using two or more biomarkers in the art is currently used. .

However, in the case of identifying a specific cancer using such a complex biomarker, the specific biomarker may not only indicate an indicator for a specific cancer but also an indicator for another cancer, and accordingly, a specific cancer may be used using the complex biomarker. Mis discrimination can occur in determining

 For example, in diagnosing adenocarcinoma (adenocarcinoma) using a composite biomarker, the result of the determination using the information of the complex biomarker may be determined as adenocarcinoma, but the actual patient's cancer is squamous cell carcinoma. )

In addition, in analyzing the biomarker information of each of the complex biomarkers, a problem such as different results may be generated depending on the discriminator in determining the cancer.

The present invention aims to solve all the above-mentioned problems.

Another object of the present invention is to be able to accurately determine the type of cancer using a biomarker group for multi-cancer diagnosis.

In addition, another object of the present invention is to be able to determine the multi-arm using a biomarker group for multi-cancer diagnosis.

In addition, another object of the present invention is to be able to accurately determine the type of cancer in the statistical determination method using a biomarker group for multi-cancer diagnosis.

In addition, another object of the present invention is to improve the reliability of cancer discrimination by a statistical discrimination method using a biomarker group for multi-cancer diagnosis.

The characteristic structure of this invention for achieving the objective of this invention mentioned above, and realizing the characteristic effect of this invention mentioned later is as follows.

According to an embodiment of the present invention, in the computing method for determining multi-arms, (a) when information of k cancer diagnosis biomarker groups is input, the computing device may determine the information of the k cancer diagnosis biomarker groups. Determining whether the information of the k cancer diagnosis biomarkers is cancer or not by performing binary cloning; And (b) if it is determined that the cancer of the k cancer diagnostic biomarker group is the result of the binary classification, the computing device analyzes the k cancer diagnostic biomarker group information using a neural network and k. Determining which of the k cancers information of the dog cancer diagnosis biomarker group corresponds to; There is provided a computing method for multi-arm determination comprising a.

In addition, according to an embodiment of the present invention, a computing device for determining the multi-arms, Communication device for obtaining information of the k cancer diagnostic biomarker group; And (i) a process of determining binary information of the k cancer diagnosis biomarker groups to determine whether the information of the k cancer diagnosis biomarker groups is cancer or not cancer, and (ii) diagnosing the k cancer diagnosis biomarkers. When it is determined that the biomarker group information is cancer, the k cancer diagnosis biomarker group information is analyzed using a neural network, and the information of the k cancer diagnosis biomarker group is analyzed among the k cancers. A processor which performs a process of determining which arm corresponds to which arm; There is provided a computing device for multi-arm determination comprising a.

In addition to this, a computer readable recording medium for recording a computer program for executing the method of the present invention is further provided.

The present invention can accurately determine the type of cancer using a biomarker group for multi-cancer diagnosis.

In addition, the present invention can determine the multi-arm using the biomarker group for multi-cancer diagnosis.

In addition, the present invention can accurately determine the type of cancer in the statistical determination method using a biomarker group for multi-cancer diagnosis.

In addition, the present invention can improve the reliability of cancer discrimination by a statistical discrimination method using a biomarker group for multi-cancer diagnosis.

1 schematically illustrates a computing device for determining a multi-arm in accordance with one embodiment of the present invention,

2 schematically illustrates a computing method for determining multi-arms according to an embodiment of the present invention.

3 schematically illustrates a model of an expert network for binary replication in a computing method for determining multi-arms according to an embodiment of the present invention.

FIG. 4 schematically illustrates a method of selecting an expert in an expert network for binary replication in a computing method of determining a multi-arm according to an embodiment of the present invention.

FIG. 5 schematically illustrates a neural network model for multi-closure in a computing method for determining multi-arms according to an embodiment of the present invention.

DETAILED DESCRIPTION The following detailed description of the invention refers to the accompanying drawings that show, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that the various embodiments of the present invention are different but need not be mutually exclusive. For example, certain shapes, structures, and characteristics described herein may be embodied in other embodiments without departing from the spirit and scope of the invention with respect to one embodiment. In addition, it is to be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the invention. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present invention, if properly described, is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. Like reference numerals in the drawings refer to the same or similar functions throughout the several aspects.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention.

FIG. 1 schematically illustrates a computing device 1000 for determining a multi arm according to an embodiment of the present invention. Referring to FIG. 1, the computing device 1000 for determining a multi arm may be connected to a communication unit 100. It may include a processor 200.

The communication unit 100 performs a function of acquiring information of k cancer diagnosis biomarker groups, and calculates an index value of each biomarker from k cancer diagnosis biomarker kits composed of biomarkers for k cancer diagnosis. Each biomarker indicator value may be received from an external device that receives or receives each biomarker indicator value or measures each biomarker indicator value.

The processor 200 may binary process the information of the k cancer diagnosis biomarker groups obtained by the communication unit 100 to determine whether the information of the k cancer diagnosis biomarker groups is cancer or cancer, and k If it is determined that cancer is the result of binary clashing of the cancer diagnosis biomarker group information, the k cancer diagnosis biomarker group information is analyzed using a neural network, and the information of the k cancer diagnosis biomarker group is any cancer among k cancers. You can follow the process to determine if

A method for determining a multi-arm according to an embodiment of the present invention by using the computing device 1000 for determining the multi-arm according to an embodiment of the present invention configured as described above is as follows.

First, as can be seen in FIG. 2, the computing device 1000 obtains k cancer diagnosis biomarker group information through the communication unit 100 (210), and the processor 200 of the computing device 1000 communicates with the communication unit 100. The k cancer diagnosis biomarker group information obtained through the operation 100 is binary classified to determine whether the information of the k cancer diagnosis biomarkers is cancer or cancer (220).

In this case, the k cancer diagnostic biomarker group information includes information of each biomarker for diagnosing k cancers, and includes proteins, nucleic acids, and the like contained in biological samples derived from the patient's body such as blood and urine of the patient. It may be an index value for diagnosing cancer using metabolites. In addition, the k cancer diagnosis biomarker group information may include information such as the age, sex, medical history of the patient.

In addition, the processor 200 of the computing device 1000 may use a mixture of experts to perform binary replication to determine whether the cancer is cancer or not using the cancer diagnosis biomarker group information. ), Decision tree, tree ensemble, random forest, Bayesian network, support vector machine, neural network, and logistic regression At least one method of logistic regression may be used. However, hereinafter, the description will focus on the Mixture of Experts method.

That is, the processor 200 of the computing device 1000 determines whether the patient possesses at least one of the k cancers by using an indicator value of each of the k cancer diagnostic biomarkers affected by the multi-arms. Do.

For example, referring to FIG. 3, the processor 200 of the computing device 1000 may include k expert networks 221 (221-1, 221-2,...) Trained on k cancer diagnosis biomarker group information x. , 221-k) to each of the trained k expert networks 221 to infer whether they are respective cancers. In detail, the first expert network 221-1 analyzes k cancer diagnosis biomarker group information to infer whether the cancer is the first cancer, and the second expert network 221-2 analyzes the k cancer diagnosis biomarker group. By analyzing the information, it is inferred whether it is the second cancer, and the k-th expert network 221-k infers whether or not the k cancer diagnosis biomarker group information is the k-th cancer.

In addition, the processor 200 of the computing device 1000 may include gating functions g1, g2, and gating of the gating network 222 in which the contribution to each of the k expert networks is determined in response to the k cancer diagnosis biomarker group information x. …, Gk) is used to recalculate the inference values of each of the k expert networks 221, and the recalculated result is used to determine whether the information (x) of the k cancer diagnosis biomarker groups is cancer. (Y).

In this case, the processor 200 of the computing device 1000 may use the softmax algorithm to determine whether the information of the k cancer diagnosis biomarker groups is cancer using the recalculated result.

의 확률로 j번째 엑스퍼트 네트워크가 선택되고 출력 y는

의 확률로 선택됨을 알 수 있다.In addition, the binary replication model of FIG. 3 may be expressed as Equation 1 below, and given an input value x

Where the jth expert network is selected and the output y is

It can be seen that the probability is selected.

[Equation 1]

In Equation 1, g (-) is a gating function of the gating network 222, which determines which expert network to use.

In this case, the gating function may be represented by Equation 2 below.

[Equation 2]

의 조건을 가진다.And the gating function

Has the condition

For example, referring to FIG. 4, (a) e2 is used at input value 0, and the output is [0, 1, 0]. (B) e2 and e3 are used at input value 0.5. [0, 0.5, 0.5].

Next, as can be seen in FIG. 2, if the processor 200 of the computing device 1000 is determined to be cancer as a result of the binary procedure of the information of the k cancer diagnostic biomarker group, k cancer diagnostic biomarkers. The group information may be analyzed using a neural network to determine which cancer among k cancer biomarker groups corresponds to k cancer (230).

For example, as shown in FIG. 5, the processor 200 of the computing device 1000 may learn when the k cancer diagnosis biomarker group information determined as cancer is input to the input layer 231 as a result of binary classification. K cancer diagnostic biomarker information is calculated using one connection intensity (Wj) and input into the learned hidden layer (232) consisting of a plurality of neurons, and the first connection intensity (Wj) in the learned hidden layer (232). The value computed using is computed using the hidden layer activation function. In this case, the hidden layer activation function may use any one of a sigmoid function (σ), a hyperbolic tangent function (tanh), and a rectified linear unit function (ReLU). In addition, although the hidden layer 232 may be formed as a single layer or a multilayer, it is assumed in FIG. 5 that the hidden layer 232 is a single layer. Then, the trained second connection intensity Wk is applied to the output value of the output layer 233 by applying the learned second connection strength Wk to the calculated value of each neuron of the hidden layer 232, and to each of k arms in the output layer 233. It is possible to determine whether the cancer of the patient from the k cancer diagnosis biomarker group information from the k cancer diagnosis biomarker group information. At this time, for each of the k cancers can be used for the softmax algorithm.

In the neural network model of FIG. 5, x _i is an i-th input variable, h _m is a neuron of the m-th hidden layer, t _k is a k-th output neuron, and the number of dimensions of the input layer 231 is p, a hidden layer. If the number of neurons of 232 is M and the number of neurons of the output layer 233 is K, the model of FIG. 5 may be expressed as Equation 3 below.

[Equation 3]

In the above, it is determined whether the cancer is a binary cancer through the use of k cancer diagnosis biomarker group information through the learned expert network and the trained neural network, and in the case of cancer, among the k cancers of the cancer through multi-closure. How to determine which cancer has been described, and how to learn the expert network and neural network in the following briefly.

First, the learning method of the expert network will be described. The processor 200 of the computing device 1000 inputs k cancer diagnosis biomarker group information for training into each of the k expert networks before training to diagnose k cancers for training. Infer whether the biomarker group information is cancer and use the gating function of the gating network whose contribution to each of the k expert networks corresponding to the k cancer diagnosis biomarker group information for training is determined. Inferences are recalculated and the recalculated results are used to determine whether the information of the k cancer diagnostic biomarker group for training is cancer. Then, the parameters of the k expert networks before learning and the parameters of the gating network before learning are learned with reference to the resultant result.

In this case, one of the gradient ascent, the EM algorithm, and the quasi-Newton method that maximizes the log likelihood may be used to learn the parameters of the k expert networks before learning and the gating networks before learning. have.

이며,

이므로, 수학식 1에서

로 로지스틱 모형을 사용할 수 있다.That is, in Equation 2

Is,

In Equation 1

Logistic models can be used.

이며, 추정하고자 하는 모수는

이다.here,

The parameter to be estimated is

to be.

Next, when the processor 200 of the computing device 1000 learns the neural network, when the k cancer diagnostic biomarker group information for training is input to the input layer before the training, the first connection strength before the training is acquired. The k cancer diagnostic biomarker group information for training is calculated and input into a hidden layer of a plurality of neurons before training, and the value calculated using the first connection strength before the training is hidden in the hidden layer before the training. Computational for each of the k arms by operating using a layer activation function and inputting the computed value in each neuron of the hidden layer before learning to the output layer before learning by applying the second connection strength before learning. do. In addition, a loss value is obtained by comparing the resultant value and the reference value corresponding to the k cancer diagnosis biomarker group information for training, and the first connection strength and learning before learning through backpropagation using the loss value. The second connection strength before it is learned, the parameters of the hidden layer before being learned, and the parameters of the output layer before being learned are learned.

의 웨이트(weight) 행렬과 b(M 차원), c(K 차원)의 바이어스 벡터이며, 손실 함수를 크로스 엔트로피 손실 함수로 하고, 이것을 최소화하도록 백프로퍼게이션 알고리즘을 사용하여 모수를 업데이트한다.That is, in Equation 3, the parameter to be estimated is

It is a weight matrix of and a bias vector of b (M dimension) and c (K dimension), and the loss function is a cross-entropy loss function, and the parameter is updated using a backpropagation algorithm to minimize this.

In the above, the method of learning the expert network and the neural network using k cancer diagnosis biomarker group information for training was described.However, the multi-clamping result value using k cancer diagnosis biomarker group information is different. You can also learn neural networks and expert networks at the same time.

In addition, in the above, the multi-arm is determined based on the learned and learned parameters using the same computing device. Alternatively, the multi-arm may be learned using a separate learning device, and the learned result may be used in the computing device for multi-arm determination. .

Embodiments according to the present invention described above may be implemented in the form of program instructions that may be executed by various computer components, and may be recorded in a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the computer-readable recording medium may be those specially designed and configured for the present invention, or may be known and available to those skilled in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs, DVDs, and magneto-optical media such as floptical disks. media), and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device may be configured to operate as one or more software modules to perform the process according to the invention, and vice versa.

Although the present invention has been described by specific embodiments such as specific components and the like, but the embodiments and the drawings are provided to assist in a more general understanding of the present invention, the present invention is not limited to the above embodiments. For those skilled in the art, various modifications and variations can be made from these descriptions.

Accordingly, the spirit of the present invention should not be limited to the above-described embodiments, and all of the equivalents or equivalents of the claims, as well as the appended claims, fall within the scope of the spirit of the present invention. I will say.

Claims (22)

In the computing method for determining a multi-arm, (a) When information of the k cancer diagnosis biomarker groups is input, the computing device may binaryly classify the information of the k cancer diagnosis biomarker groups to determine whether the information of the k cancer diagnosis biomarkers is cancer. Determining whether or not; And (b) if it is determined that the cancer information is the result of the binary classification of the k cancer diagnosis biomarker groups, the computing device analyzes the k cancer diagnosis biomarker group information by using a neural network and k k pieces. Determining which of the k cancers information of the cancer diagnosis biomarker group corresponds to; Computing method for multi-arm determination, comprising a. The method of claim 1, In the step (a), The computing device inputs the k cancer diagnosis biomarker group information to each of the k expert networks trained to infer whether the k cancer diagnosis biomarker group information is cancer, and the k cancer diagnosis biomarker Recalculate inference values of each of the k expert networks by using a gating function of a gating network whose contribution to each of the k expert networks is determined corresponding to group information, and using the recalculated result. Computing method for multi-cancer discrimination, characterized in that the cancer diagnosis whether the information of the biomarker group is cancer. The method of claim 2, The computing device may use a softmax algorithm to classify whether the information of the k cancer diagnosis biomarker groups is cancer using the recalculated result value. Way. The method of claim 2, Before step (a) above, (a01) The computing device inputs k training cancer diagnostic biomarker group information to each of the k expert networks before training to infer whether the training k cancer diagnostic biomarker group information is cancer. And recalculate inference values of each of the k expert networks using a gating function of a gating network whose contribution to each of the k expert networks is determined corresponding to the k cancer diagnostic biomarker group information for training. Using the computed result value, whether the information of the k cancer diagnostic biomarker group for training is cancer or not, and the parameters of the k expert networks before being trained with reference to the crashed result value, and Learning the parameters of the gating network before learning Computing method for determining a multi-arm. The method of claim 4, wherein The computing device includes a gradient ascent that maximizes log likelihood, EM algorithm, and quasi-Newton method in learning the k expert network before learning and the parameters of the gating network before learning. Computing method for multi-arm discrimination, characterized in that using any one. The method of claim 1, In the step (a), The binary replication includes decision trees, tree ensemble, random forests, Bayesian networks, support vector machines, neural networks. And a logistic regression method. The method of claim 1, In step (b), When the k cancer diagnosis biomarker group information determined as cancer in step (a) is input to an input layer, the computing device calculates the k cancer diagnosis biomarker information using the learned first connection strength. Input to a learned hidden layer composed of a plurality of neurons, calculate a value calculated using the first connection strength in the learned hidden layer using a hidden layer activation function, and each neuron of the learned hidden layer And applying the learned second connection strength to an operation value in the input layer, and clasping each of k arms in the output layer. The method of claim 7, wherein The hidden layer activation function may be any one of a sigmoid function (σ), a hyperbolic tangent function (tanh), and a rectified linear unit function (ReLU). The method of claim 7, wherein Computing for each of the k arms is a computing method for multi-arm determination, characterized in that using a softmax algorithm. The method of claim 7, wherein Before step (b), (b01) The k cancer diagnostic biomarker group for training using the first connection strength before training when the k cancer diagnostic biomarker group for training information is input to the input layer before the training is performed. To calculate the input to the hidden layer of the trained learning layer consisting of a plurality of neurons, using the hidden layer activation function to calculate a value calculated using the first connection strength before the training in the hidden layer before the training, Applying the second connection strength before learning to an operation value in each neuron of the hidden layer before learning, and inputting to the output layer before learning, for each of the k arms, and the recipe Corresponding to the result and the k cancer diagnostic biomarker group information for training. A loss value is obtained by comparing values, and the first connection strength before learning, the second connection strength before learning, parameters of the hidden layer before learning, and the learning through backpropagation using the loss value. Computing method for multi-arm discrimination, characterized in that learning the parameters of the output layer before being. The method of claim 7, wherein The hidden layer is a computing method for multi-arm determination, characterized in that formed in a single layer or multiple layers. A computing device for determining a multi-arm, Communication unit for obtaining information of the k cancer diagnostic biomarker group; And (i) a process of determining binary information of the k cancer diagnostic biomarker groups to determine whether the information of the k cancer diagnostic biomarker groups is cancer or cancer; and (ii) the k cancer diagnostic biomarkers. If it is determined that the information of the marker group is cancer, the k cancer diagnosis biomarker group information is analyzed using a neural network, and the information of the k cancer diagnosis biomarker group is determined from any of the k cancers. A processor that performs a process of determining whether a cancer corresponds; Computing device for multi-arm determination, comprising a. The method of claim 12, The processor, In the process (i), the k cancer diagnosis biomarker group information is input to each of the k expert networks trained to infer whether the k cancer diagnosis biomarker group information is cancer, and the k cancer diagnosis is performed. Recalculate inference values of each of the k expert networks by using a gating function of a gating network whose contribution to each of the k expert networks is determined corresponding to biomarker group information, and using the recalculated result Computing device for multi-cancer discrimination, characterized in that it is determined whether the information of the k cancer diagnostic biomarker group is cancer. The method of claim 13, The processor may use a softmax algorithm to determine whether the information of the k cancer diagnosis biomarker groups is cancer using the recalculated result, and uses a softmax algorithm. . The method of claim 13, The processor, Prior to the process (i), the k cancer diagnostic biomarker group information for training is input to each of the k expert networks before training to infer whether the k cancer diagnostic biomarker group information for training is cancer. And recalculate inference values of each of the k expert networks using a gating function of a gating network whose contribution to each of the k expert networks is determined corresponding to the k cancer diagnostic biomarker group information for training. Using the computed result value, whether the information of the k cancer diagnostic biomarker group for training is cancer or not, and the parameters of the k expert networks before being trained with reference to the crashed result value, and The process of learning the parameters of the gating network before learning Computing device for multi-arm determination, characterized in that further performing. The method of claim 15, The processor may be configured to learn the parameters of the k expert networks before learning and the parameters of the gating network before learning, such as gradient ascent, EM algorithm, and quasi-Newton method to maximize log likelihood. Computing device for multi-arm determination, characterized in that using one. The method of claim 12, The processor, In the process (i), the binary replication includes a decision tree, a tree based ensemble, a random forest, a Bayesian network, a support vector machine. ), A neural network, and a logistic regression method. The method of claim 12, In the above (ii) process, The processor, When the k cancer diagnosis biomarker group information determined to be cancer in the process (i) is input to an input layer, the k cancer diagnosis biomarker group information is calculated by using the learned first connection intensity to a plurality of neurons. Input to the learned hidden layer, and calculate a value calculated using the first connection strength in the learned hidden layer using a hidden layer activation function, and calculate a value in each neuron of the learned hidden layer And inputting to the output layer by applying the learned second connection strength to each other, and for each of the k arms in the output layer, the computing device for multi-arm determination. The method of claim 18, The hidden layer activation function is any one of a sigmoid function (σ), a hyperbolic tangent function (tanh), and a rectified linear unit function (ReLU). The method of claim 18, Computing for each of the k arms is a computing device for multi-arm determination, characterized in that using a softmax algorithm. The method of claim 18, The processor, Prior to the (ii) process, if k cancer diagnostic biomarker group training information is inputted to the input layer before learning, k training cancer cancer biomarker group information for training using the first connection strength before learning. To calculate the input to the hidden layer of the trained learning layer consisting of a plurality of neurons, using the hidden layer activation function to calculate a value calculated using the first connection strength before the training in the hidden layer before the training, Applying the second connection strength before learning to an operation value in each neuron of the hidden layer before learning, and inputting to the output layer before learning, for each of the k arms, and the recipe Corresponding to the calculated result and the k cancer diagnostic biomarker group information for training. Comparing the quasi values to obtain a loss value, and through the back propagation using the loss value, the first connection strength before the learning, the second connection strength before the learning, parameters of the hidden layer before the learning, and the learning Computing device for multi-arm determination, further comprising the process of learning the parameters of the output layer before it is. The method of claim 18, The hidden layer is a computing device for multi-arm determination, characterized in that formed in a single layer or multiple layers.

Images