CN114947740A - Method for predicting epileptic seizure based on quantum CNN-GRU - Google Patents

Method for predicting epileptic seizure based on quantum CNN-GRU Download PDF

Info

Publication number
CN114947740A
CN114947740A CN202210406450.5A CN202210406450A CN114947740A CN 114947740 A CN114947740 A CN 114947740A CN 202210406450 A CN202210406450 A CN 202210406450A CN 114947740 A CN114947740 A CN 114947740A
Authority
CN
China
Prior art keywords
quantum
epileptic
features
vqc
cnn
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210406450.5A
Other languages
Chinese (zh)
Inventor
张方言
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Turing Intelligent Computing Quantum Technology Co Ltd
Original Assignee
Shanghai Turing Intelligent Computing Quantum Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shanghai Turing Intelligent Computing Quantum Technology Co Ltd filed Critical Shanghai Turing Intelligent Computing Quantum Technology Co Ltd
Priority to CN202210406450.5A priority Critical patent/CN114947740A/en
Publication of CN114947740A publication Critical patent/CN114947740A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/40—Detecting, measuring or recording for evaluating the nervous system
    • A61B5/4076—Diagnosing or monitoring particular conditions of the nervous system
    • A61B5/4094—Diagnosing or monitoring seizure diseases, e.g. epilepsy
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7242—Details of waveform analysis using integration
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Psychiatry (AREA)
  • Physiology (AREA)
  • Artificial Intelligence (AREA)
  • Signal Processing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Neurology (AREA)
  • Neurosurgery (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Psychology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a quantum CNN-GRU-based epileptic seizure prediction method, and belongs to the technical field of quantum computing and neuroscience. Because the method constructs the quantum CNN-GRU by the linear module in the classic CNN-GRU through the quantum circuit so as to predict the possibility of epileptic seizure, the quantum CNN-GRU provided by the invention can save a large amount of computing resources, the prediction result is more accurate compared with the structure of the classic CNN-GRU, the model can be converged to a stable state more quickly in the training process, and meanwhile, the quantum chip and the electronic chip can cooperatively work to be used for the CNN-GRU model for epileptic prediction.

Description

Method for predicting epileptic seizure based on quantum CNN-GRU
Technical Field
The invention relates to the field of quantum computing and neuroscience, in particular to a method for predicting epileptic seizure based on quantum CNN-GRU.
Background
Epilepsy is a chronic non-infectious disease which affects all ages and is caused by paroxysmal abnormal hypersynchronous electricity of brain neurons, and one of the global common neurological diseases. The characteristics of epilepsy are recurrent attacks, unpredictability and difficult treatment of antiepileptic drugs, which impair the quality of life of people. Repeated seizures of epilepsy have a persistent negative impact on the mental cognitive function of a patient and are even life-threatening, so that the detection of a pre-seizure state before an epileptic seizure can save lives, and has important clinical significance for the study of epilepsy prediction.
The classic CNN-GRU algorithm is a research hotspot in the field of artificial intelligence and is applied to various application scenes such as biomedicine, materials, neuroscience and the like, but the operation of the models needs to consume a large amount of computing resources. In the past, the computational resources for algorithm operation were mainly provided by chips fabricated from electronic integrated circuits, and as the electron tunneling effect has constrained the fabrication process to the nanometer limit, the computational effort has been difficult to increase continuously. The quantum computing chip is a supplement to the computing mode of the electronic chip, however, the operation of the classic CNN-GRU algorithm on the quantum chip cannot be processed in the way that it is on the electronic chip.
Although the epilepsy prediction problem obtains good performance based on the CNN-GRU model, a certain module in the CNN-GRU algorithm has high parallelism and consumes a large amount of computing resources,
therefore, the invention provides a quantum CNN-GRU to predict the possibility of epileptic seizure.
Disclosure of Invention
The present invention has been made to solve the above problems, and an object of the present invention is to provide a method for predicting seizures based on quantum CNN-GRU.
The invention provides a method for predicting epileptic seizure based on quantum CNN-GRU, which is characterized by comprising the following steps: extracting epilepsia brain wave information and coding the epilepsia brain wave information into corresponding quantum states; and inputting the quantum state into a quantum CNN-GRU for evolution, and outputting the probability of epileptic seizure after measurement.
The method provided by the invention also has the following characteristics: wherein, the quantum CNN-GRU comprises quantum CNN and quantum GRU; the quantum CNN is used for extracting epileptic features in epileptic brain wave information; the quantum GRU is used for extracting time sequence characteristics in the epilepsia characteristics.
The method provided by the invention also has the following characteristics: the quantum CNN comprises a quantum convolution module and a quantum pooling module.
The method provided by the invention also has the following characteristics: the quantum convolution module and the quantum pooling module respectively comprise a Pouli rotating gate and a controlled gate; the pauli revolving door is used for providing optimizable parameters; controlled gates are used to achieve fusion of the quantum states.
The method provided by the invention also has the following characteristics: the quantum GRU comprises a first quantum wire, a second quantum wire and a third quantum wire; the first quantum circuit is used for screening the epileptic features in the previous hidden state and the epileptic features at the current time step; the second quantum circuit is used for screening epileptic features which can be preserved together with the current time step in a previous hidden state; the third quantum circuit is used for screening the epileptic features in the previous hidden state and the epileptic features at the current time step.
The method provided by the present invention has also been characterized in that the first quantum wire comprises two variational quantum wires VQC 1 、VQC 2 ，VQC 1 With the current time step x t Multiplication of epileptic features of (a); VQC 2 And h in the previous hidden state t-1 Multiplication of epileptic features of (a); and summing the multiplied products and screening the epilepsy characteristics by using a first activation function.
The method provided by the present invention has also been characterized in that the second quantum wire comprises two variational quantum wires VQC 3 、VQC 4 ，VQC 3 With the current time step x t Multiplication of epileptic features of (a); VQC 4 And h in the previous hidden state t-1 Multiplying the epilepsy features of the first quantum circuit screen by the epilepsy features of the first quantum circuit screen; and after summing the products obtained by multiplying the products, screening the epilepsy characteristics needing to be retained by using a second activation function.
The method provided by the present invention has also been characterized in that the third quantum wire comprises two variational quantum wires VQC 5 、VQC 6 ，VQC 5 With the current time step x t Multiplication of epileptic characteristics of (a); VQC 6 And h in the previous hidden state t-1 Multiplication of epileptic features of (a); and (4) after the products after respective multiplication are summed, screening the epilepsy characteristics needing to be retained by using the first activation function.
The method provided by the invention is also characterized by further comprising the following steps: the third quantum wire is connected to the previous hidden state h t-1 After multiplication, screening for features of epilepsy in the previous hidden state that need to be preserved.
The method provided by the invention is also characterized in that the quantum wire is utilized to screen the epilepsy characteristics under the previous hidden state to be reserved and output the hidden state h corresponding to the time step t t The characteristics of epilepsy below, and the probability of a post-outcome seizure.
Action and Effect of the invention
According to the method for predicting the epileptic seizure based on the quantum CNN-GRU, disclosed by the invention, because the method constructs the quantum CNN-GRU through the quantum circuit by using the linear module in the classic CNN-GRU so as to predict the possibility of the epileptic seizure, the quantum CNN-GRU provided by the invention can save a large amount of computing resources, the prediction result is more accurate than the structure of the classic CNN-GRU, the model can be quickly converged to a stable state in the training process, and meanwhile, the quantum chip and the electronic chip can cooperatively work to be used for a CNN-GRU model for epileptic prediction.
Drawings
Fig. 1 is a general flowchart of a method of predicting epileptic seizures based on quantum CNN-GRUs in an embodiment of the present invention.
Fig. 2 is a schematic diagram of a variational quantum wire.
Fig. 3 is a schematic diagram of a process of encoding quantum states in an embodiment of the invention.
Fig. 4 is a schematic structural diagram of a quantum convolution module in an embodiment of the present invention.
Fig. 5 is a schematic diagram of a structure of a quantum pooling module in an embodiment of the invention.
Fig. 6 is a schematic diagram of a classical cycle neural network.
Fig. 7 is a schematic structural diagram of a quantum GRU in an embodiment of the present invention.
Fig. 8a is a schematic structural diagram of a first variational quantum wire in an embodiment of the present invention.
Fig. 8b is a schematic structural diagram of a second variational quantum wire in an embodiment of the present invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the method for predicting the epileptic seizure based on the quantum CNN-GRU is specifically described below with reference to the embodiment and the accompanying drawings.
< example >
As shown in fig. 1, the general flow of the method for predicting epileptic seizure based on quantum CNN-GRU provided in this embodiment includes the following steps: preprocessing EEG epilepsia brain wave data to obtain corresponding quantum states, inputting the quantum states into QCNN to perform feature extraction to obtain epilepsia features, extracting time sequence features in the epilepsia features by using QGRU, measuring probability of outputting epilepsia seizures, and finally optimizing output through a loss function to predict possibility of epilepsia seizures.
The EEG epileptic brain wave dataset in this embodiment is epileptic brain wave data CHB-MIT, which is collected from boston's hospital, including electroencephalographic recordings of pediatric patients with refractory epileptic seizures. The scalp electroencephalographic data set was recorded by the boston hospital for children in cooperation with the massachusetts institute of technology, and published on pytion et. The 23 electrode recording dataset was placed on the scalp of an epileptic patient. The data set was collected from 22 subjects, including 17 women and 5 men, women aged 1.5 to 19 years, and men aged 3 to 22 years.
The quantum CNN and the quantum GRU in this embodiment use a variational quantum line as shown in fig. 2. A Variational Quantum Circuit (VQC) is a quantum circuit that has adjustable parameters that can be iteratively optimized by an optimization algorithm. Where the u (x) module is the encoding stage, which is responsible for encoding classical input data into quantum state x. The V (θ) module is a variational stage having a parameter θ that can be adjusted by an optimization algorithm. Finally, we obtain classical information output by measuring all or part of the qubits.
Existing research proves that the circuit has the advantage of resisting quantum noise, and is suitable for a noisy short-term medium-scale quantum computer. VQC has been successfully applied to function approximation, classification, network generation, etc. In addition, there is evidence that VQC is more expressive than classical neural networks. Expressive power means that finite parameters are used to fit a function or distribution, so this embodiment enables CNNs and GRUs to predict the probability of seizures more quickly and accurately by means of VQC.
As shown in fig. 3, the process of encoding the quantum state in this embodiment is specifically as follows:
s1, segmenting the epilepsia electroencephalogram data with the length of n to obtain epilepsia data with the length of m of each segment;
S2，convert epilepsy data of length m to m × m (where m is qubit and nqubit is 2) n Number of bits of qutorch quantum wires) of the characteristic matrix U 1 ；
S3, combining the matrix U 1 Is multiplied by U 1 (i.e. the
Figure BDA0003602333900000061
) Obtaining a Gram semi-positive definite matrix of m multiplied by m;
and S4, substituting the Gram semi-positive definite matrix into an encoding function (encoding can convert the Gram semi-positive definite matrix of m multiplied by m into a quantum state density matrix of m multiplied by m) to obtain quantum state data corresponding to the epileptic sequence.
Fig. 4-5 are a quantum convolution module and a quantum pooling module in quantum CNN, respectively, and specifically, the QCNN partial quantum circuit includes parameterized pauli rotary gates as learnable parameters of a neural network, and performs quantum state error correction through controlled gates, and then performs feature extraction through reduced density matrix operation. The parameterized quantum circuit part can adopt quantum convolution kernels and quantum pooling kernels shown in fig. 3 and 4, the kernels are arranged according to a certain rule, a classical CNN neural network is simulated, features of input epileptic brain wave data are extracted to obtain epileptic features, and the epileptic features are processed to obtain a corresponding epileptic feature sequence (hereinafter referred to as the sequence).
FIG. 6 is a classical recurrent neural network algorithm (RNN), a machine learning model that efficiently processes sequential data, modeling time dependencies by remembering earlier input data. Timing data may be processed and input into the RNN, and parameters of the model may be trained to achieve this capability. For example, a model can generate control signals, with inputs being continuous state data of the physical system. The goal of the training is to find the parameter values, i.e. weights, that minimize the loss function at each moment.
The reason why the RNN can capture the time dependency is that the model not only checks one input value at each time, but also checks the output (hidden state) at the previous time at the same time, which enables the model to save the information input at the previous time, thereby making a correct predictionAnd (6) measuring. The hidden state is the key to the model having "memory". FIG. 6 shows a time series { x } 0 ,., xn as input to the model, the output of the model is also a sequence. The information flow can be clearly seen if we repeatedly plot the input and output of the model at each moment, and any one RNN unit at different moments is the same neural network.
As shown in fig. 7, a quantum GRU (gated recurrent neural network) 51 provided in the present embodiment includes a first quantum wire 52, a second quantum wire 53, and a third quantum wire 54.
The first quantum wire 52 is used for screening the sequence for epileptic features in a previous hidden state and epileptic features at a current time step, and the first quantum wire 52 comprises a first variational quantum wire 521 and a second variational quantum wire 522.
Wherein the expression of the first quantum wire 52 is as follows:
r t ＝σ(VQC 1 (x t )+VQC 2 (h t-1 ))
σ is the activation function, VQC 1 Is a first variational quantum wire 521, x t Is the epileptic feature at time t in the sequence, h t-1 For epileptic features in the hidden state at the time t-1 of the sequence, VQC 2 A second variational quantum wire 522.
Fig. 8 is a schematic structural diagram of a first quantum wire in an embodiment of the present invention.
As shown in fig. 8, the first quantum wire 52 of the present embodiment includes a first variational quantum wire 521 and a second variational quantum wire 522.
Fig. 8a is a schematic structural diagram of a first variational quantum wire.
As shown in fig. 8a, a first variational quantum wire 521 is used to extract the epileptic feature x in the sequence t And compresses it to obtain a variation vector corresponding to the epileptic feature, the first variation quantum circuit 521 includes a parameterized rotation module 5211 and an entanglement module 5212.
The parameterized rotation module 5211 is used to provide parameters that the first variational quantum wire 521 can learn and encode each element in the input vector corresponding to an epileptic feature as a quantum superposition state.
In this embodiment, the input vector corresponding to the epileptic feature is encoded into the superimposed quantum state by the following formula, which is specifically as follows:
Figure BDA0003602333900000081
where the summation index i is a decimal number representing a bit string of the corresponding ground state. n-dimensional reduced input vector x ═ x 1 ,...,x n ) Corresponding to each epileptic feature in the sequence, and each element in an n-dimensional input vector x will be used to generate two rotation angles, e.g., arctan (x) 1 )，
Figure BDA0003602333900000082
The first angle of rotation is generated by rotating the y-axis (parameterized Pauli rotation gate R) Y ) The second angle of rotation is obtained by rotating the z-axis (parameterized Pauli rotation gate R) Z ) Thus obtaining the product. By applying two rotation operations to each quantum bit, the input vector x, which is an equal superposition initial state, is transformed into a corresponding quantum superposition state.
The entanglement module 5212 includes a predetermined number of control gates 5213 and single bit rotary gates (U) 5214.
The control not gate 5213 acts on each pair of adjacent qubits, or qubits spaced by one, in the quantum superposition state, thereby creating an entangled multi-bit quantum state.
The single bit rotary gate 5214 has a plurality of rotation angles, such as α, β, γ, and the rotation angles in the single bit rotary gate 5214 are calculated by using the existing optimization method, and the first n qubits are measured, and the output quantum state is 2 n A dimension vector. The use of the first variational quantum wire 521 for each position in the predetermined sequence is repeated, and the output shape is (predetermined sequence length, 2) n ) Of the matrix of (a).
In the present embodiment, by measuring the quantum measurement layer of the first variation quantum wire 521, therebyThe outcome of the variational quantum line output, i.e. the probability of a seizure, is output. In this embodiment, E is 4, corresponding to four qubits q in a transmutation quantum wire 0 、q 1 、q 2 、q 3 In other embodiments, E may correspond to a number of qubits greater than 4. All or part of the qubits (determined by the number of output features) are specifically measured. A single measurement will return a classical bit string, e.g. 00.. 1, corresponding to the ground state |1>. To estimate the modulus of all amplitudes of the quantum states, the equal quantum superposition states are repeatedly measured multiple times, and the frequency of the measurement is used to predict the probability of epileptic seizure (modulus of amplitude).
Here, the present embodiment resets and updates the rotation angle in the first variational quantum wire 521 using the existing loss function.
Fig. 8b is a schematic structural diagram of a second variational quantum wire.
As shown in fig. 8b, the structure of the second variational quantum wire 522 is the same as that of the first variational quantum wire 521, except that the second variational quantum wire 522 is used to extract and compress the epilepsy feature x input by the first variational quantum wire 521 t Corresponding hidden state h t-1 And for the hidden state h t-1 The epileptic features under are compressed, thereby outputting a hidden state h t-1 The corresponding variation vectors are not described herein.
As shown in fig. 7, the second quantum wire 53 includes a third variational quantum wire 531 and a fourth variational quantum wire 532.
The second quantum wire 53 is used to determine the number of epileptic features that need to be retained in the previous hidden state and the number of epileptic features that need to be retained at the current time step, respectively.
In the present embodiment, the expression of the second quantum wire 53 is as follows:
n t ＝tanh(VQC 3 (x t )+r t *VQC 4 (h t-1 ))
tan h is the activation function, VQC 3 Is a third variational quantum wire 531, r t Is the first quantum wire 52, VQC 4 Is a fourth variational quantum wire 532.
In the present embodiment, the third variational quantum wire 531 and the first variational quantum wire 521 have the same structure, and the fourth variational quantum wire 532 and the second variational quantum wire 522 have the same structure, and are not described herein again.
The third quantum wire 54 is used to screen the number of epileptic features that need to be preserved in the previously hidden state, and the expression of the third quantum wire 54 is as follows:
z t ＝σ(VQC 5 (x t )+VQC 6 (h t-1 ))
VQC 5 is a fifth variational quantum line 541, VQC 6 In the sixth variational quantum wire 542, the fifth variational quantum wire 541 has the same structure as the first variational quantum wire 521, and the sixth variational quantum wire 542 has the same structure as the second variational quantum wire 522, and thus the description thereof is omitted.
In this example, the expression of all epileptic features processed in the previous hidden state is as follows:
h t ＝(1-z t )*n t +z t *h t-1
z t is a third quantum wire 54, n t Is the second quantum wire 53.
Further, the probability of the epileptic seizure is obtained based on the result measured after QCNN, the predetermined loss function is used for optimization based on the probability, and finally the probability of the epileptic seizure is output.
Effects and effects of the embodiments
According to the method for predicting the epileptic seizure based on the quantum CNN-GRU, disclosed by the embodiment, because the method constructs the quantum CNN-GRU through the quantum circuit by using the linear module in the classic CNN-GRU so as to predict the possibility of the epileptic seizure, the quantum CNN-GRU provided by the invention can save a large amount of computing resources, the prediction result is more accurate than that of the classic CNN-GRU structure, the model can be quickly converged to a stable state in the training process, and meanwhile, the quantum chip and the electronic chip can cooperatively work to be used for the CNN-GRU model for epileptic prediction.
Further, in the embodiment, the possibility of epileptic seizure is predicted based on quantum CNN-GRU, and the method of the embodiment may also be used for predicting disease prediction of other neuroscience, or related prediction problems in other fields such as biomedical field.
The above embodiments are preferred examples of the present invention, and are not intended to limit the scope of the present invention.

Claims (10)

1. A method for predicting epileptic seizure based on quantum CNN-GRU, comprising:
extracting epilepsia brain wave information and coding the epilepsia brain wave information into corresponding quantum states;
and inputting the quantum state into a quantum CNN-GRU for evolution, and outputting the probability of epileptic seizure after measurement.
2. The method of claim 1, wherein:
the quantum CNN-GRU comprises quantum CNN and quantum GRU;
the quantum CNN is used for extracting epileptic features in the epileptic brain wave information;
the quantum GRU is used for extracting time sequence features in the epilepsia features.
3. The method of claim 2, wherein:
the quantum CNN comprises a quantum convolution module and a quantum pooling module.
4. The method of claim 3, wherein:
the quantum convolution module and the quantum pooling module respectively comprise a Poilli rotating gate and a controlled gate;
the pauli rotary door is used for providing optimizable parameters;
the controlled gate is used to achieve fusion of the quantum states.
5. The method of claim 2, wherein:
wherein the quantum GRU includes a first quantum wire, a second quantum wire, and a third quantum wire;
the first quantum circuit is used for screening epileptic features in a previous hidden state and epileptic features of a current time step;
the second quantum circuit is used for screening epileptic features which can be preserved together with the current time step in a previous hidden state;
the third quantum wire is used for screening the epileptic features in the previous hidden state and the epileptic features at the current time step.
6. The method of claim 5,
wherein the first quantum wire comprises two variational quantum wires VQC 1 、VQC 2 ，
VQC 1 With the current time step x t Multiplication of epileptic features of (a);
VQC 2 and h in the previous hidden state t-1 Multiplication of epileptic features of (a);
and summing the multiplied products and screening the epilepsy characteristics by using a first activation function.
7. The method of claim 5,
wherein the second quantum wire comprises two variational quantum wires VQC 3 、VQC 4 ，
VQC 3 With the current time step x t Multiplication of epileptic features of (a);
VQC 4 and h in the previous hidden state t-1 Multiplying the epilepsy feature of (a) by the epilepsy feature of the first quantum wire screening;
and after summing the products obtained by multiplying the products, screening the epilepsy characteristics needing to be retained by using a second activation function.
8. The method of claim 5,
wherein the third quantum wire comprises two variational quantum wires VQC 5 、VQC 6 ，
VQC 5 With the current time step x t Multiplication of epileptic features of (a);
VQC 6 and h in the previous hidden state t-1 Multiplication of epileptic features of (a);
and after the products after respective multiplication are summed, screening the epilepsy characteristics needing to be retained by using the first activation function.
9. The method of claim 8, further comprising:
the third quantum wire is connected with the previous hidden state h t-1 After multiplication, screening for features of epilepsy in the previous hidden state that need to be preserved.
10. The method according to claim 5 or 9,
using the quantum circuit and the screening to output the hidden state h corresponding to the time step t of the epileptic feature in the previous hidden state to be preserved t The characteristics of epilepsy below, and the probability of a post-outcome seizure.
CN202210406450.5A 2022-04-18 2022-04-18 Method for predicting epileptic seizure based on quantum CNN-GRU Pending CN114947740A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210406450.5A CN114947740A (en) 2022-04-18 2022-04-18 Method for predicting epileptic seizure based on quantum CNN-GRU

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210406450.5A CN114947740A (en) 2022-04-18 2022-04-18 Method for predicting epileptic seizure based on quantum CNN-GRU

Publications (1)

Publication Number Publication Date
CN114947740A true CN114947740A (en) 2022-08-30

Family

ID=82978066

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210406450.5A Pending CN114947740A (en) 2022-04-18 2022-04-18 Method for predicting epileptic seizure based on quantum CNN-GRU

Country Status (1)

Country Link
CN (1) CN114947740A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114461069A (en) * 2022-02-07 2022-05-10 上海图灵智算量子科技有限公司 Quantum CNN-LSTM-based emotion recognition method

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114461069A (en) * 2022-02-07 2022-05-10 上海图灵智算量子科技有限公司 Quantum CNN-LSTM-based emotion recognition method

Similar Documents

Publication Publication Date Title
Choi et al. EmbraceNet: A robust deep learning architecture for multimodal classification
CN113627518B (en) Method for realizing neural network brain electricity emotion recognition model by utilizing transfer learning
Abdelhameed et al. A deep learning approach for automatic seizure detection in children with epilepsy
CN112667080B (en) Intelligent control method for electroencephalogram signal unmanned platform based on deep convolution countermeasure network
CN111631688B (en) Algorithm for automatic sleep staging
Gupta et al. A novel approach for classification of mental tasks using multiview ensemble learning (MEL)
Manjunath et al. A low-power lstm processor for multi-channel brain eeg artifact detection
Dong et al. Subject sensitive EEG discrimination with fast reconstructable CNN driven by reinforcement learning: A case study of ASD evaluation
Mohdiwale et al. Statistical wavelets with harmony search-based optimal feature selection of EEG signals for motor imagery classification
Malviya et al. A novel technique for stress detection from EEG signal using hybrid deep learning model
Xie et al. Physics-constrained deep learning for robust inverse ecg modeling
Da et al. Brain CT image classification with deep neural networks
Dutta et al. MED-NET: a novel approach to ECG anomaly detection using LSTM auto-encoders
CN114947740A (en) Method for predicting epileptic seizure based on quantum CNN-GRU
Sridevi et al. Quanvolution neural network to recognize arrhythmia from 2D scaleogram features of ECG signals
Nakra et al. Deep neural network with harmony search based optimal feature selection of EEG signals for motor imagery classification
Rana et al. A novel spiking neural network for ecg signal classification
CN114550907A (en) Epilepsy detection system
Rana et al. Comparison of artificial neural networks for low-power ECG-classification system
CN113995417A (en) Electrocardiosignal abnormity prediction method and system based on LSTM self-encoder
CN114580614A (en) Model for generating data features
CN115329929A (en) Hypergraph representation method of brain function network
CN114967911A (en) Sleep segmentation method based on quantum CNN-LSTM
CN114461069A (en) Quantum CNN-LSTM-based emotion recognition method
Micheli-Tzanakou et al. Comparison of neural network algorithms for face recognition

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination