CN113743301B - Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network - Google Patents
Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network Download PDFInfo
- Publication number
- CN113743301B CN113743301B CN202111032628.6A CN202111032628A CN113743301B CN 113743301 B CN113743301 B CN 113743301B CN 202111032628 A CN202111032628 A CN 202111032628A CN 113743301 B CN113743301 B CN 113743301B
- Authority
- CN
- China
- Prior art keywords
- neural network
- signal
- residual
- noise
- encoder
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
The application provides a solid-state nanopore sequencing electric signal noise reduction processing method based on a residual error self-encoder convolutional neural network, which is characterized in that a self-encoder, a residual error bottleneck module, a shortcut connection, a convolutional layer, a batch regularization layer, an activation layer and other structures are introduced to construct a deep neural network model, a solid-state nanopore sequencing electric signal data set is utilized to train the model, so that the model accurately learns the characteristic mode of sequencing electric signal noise, a mapping from a noise signal to a clean signal is established, and finally the clean signal corresponding to the noise signal is predicted and estimated by using the learned mapping. The signal noise reduction method based on the residual error self-encoder convolutional neural network, disclosed by the application, enhances the recognition capability of the neural network on the noise part of the electric signal, establishes the accurate mapping from the noise electric signal to the clean signal, and realizes real-time denoising.
Description
Technical Field
The application belongs to the technical field of digital signal processing, and particularly relates to a solid-state nanopore sequencing electric signal noise reduction processing method based on a residual error self-encoder convolutional neural network, which can be applied to machine learning and digital signal processing technologies.
Background
In recent years, solid state nanopores have received widespread attention as emerging nanostructured materials. Compared with the existing gene sequencing system, the gene sequencing system developed based on the solid-state nano-pore is stable, has the advantages of high flux and low cost, and is widely focused by researchers. However, there are a number of issues still to be addressed in solid state nanopore sequencing system development, where noise in the solid state nanopore sequencing signal severely affects the accuracy of the final sequencing result. Further studies have shown that noise in solid state nanopore sequencing signals consists mainly of a flicker noise in the low frequency part and a thermal noise, a dielectric noise, a capacitive noise in the high frequency part, but essentially: the noise sequencing electrical signal Y consists of a superposition of a clean sequencing electrical signal X and an additive sequencing system noise N, which can be expressed as a mathematical expression: y=x+n. In order to improve the sequencing accuracy, noise reduction treatment is required to be carried out on the noise electric signals, superimposed noise is removed as much as possible, and the signals are restored to a clean signal state.
At present, the noise reduction processing method for the nanopore sequencing electric signal mainly utilizes various low-pass filters such as Bessel filtering and the like to remove a high-frequency noise part so as to improve the signal to noise ratio, but the low-pass filtering can only process the noise electric signal to the extent that signal analysis can be carried out, and the cost is the sacrifice of the high-frequency characteristic part of the sequencing signal. Meanwhile, the signal processing method cannot solve the influence of a low-frequency flicker part in the noise signal, the processing effect is not satisfactory, and great inconvenience is brought to subsequent sequencing signal analysis.
Disclosure of Invention
Aiming at the problems existing in the prior art, based on the needs of reality and production practice, the inventor invests a great deal of funds, and through long-term research, provides a solid-state nanopore sequencing electric signal noise reduction processing method based on a residual error self-encoder convolutional neural network, which enhances the recognition capability of the neural network on the electric signal noise part, establishes accurate mapping from a noise electric signal to a clean signal, and realizes real-time denoising of the solid-state nanopore electric signal.
According to the technical scheme of the application, a solid-state nanopore electric signal noise reduction processing method based on a residual error self-encoder convolutional neural network is provided, the method introduces a self-encoder structure, a residual error bottleneck module, a shortcut connection structure, a convolutional layer, a batch regularization layer, an activation layer and other structures to create a deep neural network model, and accurately learns the characteristics of noise in signals by combining the created solid-state nanopore sequencing electric signal training data set training model, so that the mapping from noise signals to clean signals is established, and the corresponding clean signals can be predicted and estimated according to the noise signals by using the learned mapping.
Further, the solid-state nanopore electrical signal noise reduction processing method based on the residual self-encoder convolutional neural network comprises the following steps of:
step S1, constructing a residual error self-encoder convolutional neural network model;
s2, selecting and creating a training data set, and setting training parameters of the residual error self-encoder convolutional neural network model;
step S3, training the residual error self-encoder convolutional neural network model by taking a minimized loss function as a target according to set model training parameters, and completing construction of a neural network model for noise reduction processing of the solid-state nanopore sequencing electric signal;
and S4, inputting the noise electric signal to be processed into a solid-state nanopore sequencing electric signal noise reduction processing neural network model, and outputting a noise reduction processed signal.
Preferably, the residual self-encoder convolutional neural network model in step S1 comprises an encoder portion and a decoder portion, the encoder portion comprising a plurality of residual coding modules, the residual coding modules comprising a plurality of convolutional layers, a plurality of active layers, and a plurality of batch regularization layers; the decoder section is composed of a plurality of deconvolution decoding modules, a convolution layer and a sigmoid activation layer, and the deconvolution decoding modules are composed of a plurality of deconvolution layers, a plurality of activation layers and a plurality of batch regularization layers.
Preferably, in step S2, the training data set in step S2 consists of clean sequencing electrical signals and corresponding noise signals. Further, collecting noise of a blank solid-state nanopore sequencing system, constructing a simulated via signal data set according to a DNA base signal standard library, overlapping the simulated via signal data set with the collected noise of the solid-state nanopore sequencing system to create a training data set, and setting training parameters of a residual self-encoder convolutional neural network model.
More preferably, in step S3, the loss function in step S3 is a mean square error function:
wherein X is i 、Y i And respectively creating a noise signal and a clean signal in the training set, wherein θ is a weight, n represents the number of data points in the signal, and F (-) is a mapping from the noise signal obtained after training to the clean signal.
In the step S3, the initial value of the weight of the neural network model of the noise reduction processing of the solid-state nanopore sequencing electric signal is generated by a Gaussian random function, and the weight parameter of the last batch of regularization layers of the residual error module is initially set to be zero.
Compared with the prior art, the application has the beneficial effects that: the application forms a signal noise reduction processing neural network model under the corresponding noise variance aiming at a plurality of different noise variance training residual error self-encoder convolutional neural network models, and carries out noise reduction processing on the base via hole signal to be processed through the signal noise reduction processing neural network model under the noise variance corresponding to the image to be processed, thereby having high processing speed.
Drawings
FIG. 1 is a flow chart of a solid state nanopore sequencing electrical signal noise reduction processing method based on a residual self-encoder convolutional neural network in accordance with the present application;
FIG. 2 is a schematic diagram of the internal architecture of a residual self-encoder convolutional neural network model in accordance with the present application;
fig. 3 is a schematic diagram of an internal construction of a residual coding module according to the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the inventor, are within the scope of the application.
According to the solid-state nanopore sequencing electrical signal noise reduction processing method based on the residual self-encoder convolutional neural network, a self-encoder structure, a residual bottleneck module, a shortcut connection, a convolutional layer, a batch regularization layer, an activation layer and other structures are introduced to create a deep neural network model, the characteristics of noise in signals are accurately learned by combining the created solid-state nanopore sequencing electrical signal training data set training model, and a mapping from noise signals to clean signals is established, so that the clean signals corresponding to the noise signals can be predicted and estimated according to the noise signals by using the learned mapping.
Specifically, the solid-state nanopore sequencing electric signal noise reduction processing method based on the residual self-encoder convolutional neural network mainly comprises the following steps of: building a residual self-encoder convolutional neural network model, wherein the residual self-encoder convolutional neural network model comprises two main parts of an encoder and a decoder, and each part comprises a plurality of convolutional layers and an activating layer behind each convolutional layer; creating a solid-state nanopore electric signal training set, and setting training parameters of the residual error self-encoder convolutional neural network model; training the residual self-encoder neural network model with a minimum loss function as a target according to the residual self-encoder neural network model and training parameters thereof to form a solid-state nanopore electric signal noise reduction processing model; and inputting the solid-state nanopore electric signal to be processed into the solid-state nanopore electric signal noise reduction processing model, and outputting the electric signal after noise reduction processing.
The application will be further described with reference to the following drawings in conjunction with the preferred embodiments. As shown in fig. 1, a solid-state nanopore electrical signal noise reduction processing method based on a residual self-encoder convolutional neural network comprises the following steps:
step S1, constructing a residual error self-encoder convolutional neural network model;
s2, selecting and creating a training data set and setting initial parameters of a model;
step S3, training the residual error self-encoder convolutional neural network model by taking a minimized loss function as a target according to set model training parameters, and completing construction of a solid nanopore sequencing electric signal noise reduction processing neural network model;
and S4, inputting the noise electric signal to be processed into a solid-state nanopore sequencing electric signal noise reduction processing neural network model, and outputting a noise reduction processed signal.
Step S1, constructing a residual self-encoder neural network model, wherein the residual self-encoder neural network model comprises an encoder part and a decoder part; the encoder portion includes a plurality of residual coding modules including a plurality of convolutional layers, a plurality of active layers, and a plurality of batch regularization layers; the decoder portion includes a plurality of deconvolution layers, a plurality of activation layers, and a plurality of batch regularization layers. Still further, the decoder section is comprised of a plurality of deconvolution decoding modules, a convolutional layer and a sigmoid active layer, the deconvolution decoding modules being comprised of a plurality of deconvolution layers, a plurality of active layers and a plurality of batch regularization layers.
The residual coding module of the residual neural network model coder part comprises a plurality of convolution layers with convolution kernels larger than 1×1 and a plurality of convolution layers with convolution kernels of 1×1. Still further, the residual coding module comprises a convolution layer with a convolution kernel size of 3×3 and two convolution layers with convolution kernels of 1×1; and the first layer and the third layer of the residual coding module have the convolution kernel size of 1 multiplied by 1, the second layer of the residual coding module has the convolution kernel size of 3 multiplied by 3, and the constructed residual coding module is of a bottleneck neural network structure. In a further embodiment, the residual coding module comprises a shortcut connection for directly mapping identities of inputs and outputs bypassing a bottleneck convolutional network structure.
And step S2, selecting and creating a training data set and setting initial parameters of a model, wherein the training data set in the step S2 consists of clean sequencing electric signals and corresponding noise signals. And further, acquiring noise of a blank solid-state nanopore sequencing system, constructing a simulated via signal data set according to a DNA base signal standard library, overlapping the simulated via signal data set with the acquired noise of the solid-state nanopore sequencing system to create a training data set, and setting training parameters of the residual self-encoder convolutional neural network model.
Step S3, training the residual self-encoder convolutional neural network model to form a signal noise reduction processing neural network model by taking a minimized loss function as a target according to the residual self-encoder convolutional neural network model and training parameters thereof; the loss function in step S3 is a mean square error function:
wherein X is i 、Y i And respectively creating a noise signal and a clean signal in the training set, wherein θ is a weight, n represents the number of data points in the signal, and F (-) is a mapping from the noise signal obtained after training to the clean signal. Furthermore, in step S3, an initial value of the weight of the neural network model for the noise reduction processing of the solid-state nanopore sequencing electrical signal is generated by a gaussian random function, and a weight parameter of the regularization layer of the last batch of residual modules is initially set to zero.
Step S4, the noise electric signal to be processed is further preprocessed and then input into the electric signal noise reduction processing neural network model, and the noise reduction processed signal is output.
As shown in fig. 2, the residual self-encoder convolutional neural network model constructed in step S1 is further described herein. In a preferred embodiment of the present application, the encoder portion of the residual self-encoder convolutional neural network consists of eight residual coding modules, each comprising a three-layer bottleneck convolutional network structure and a shortcut connection. And the decoder part is composed of four deconvolution decoding modules, one convolution layer and one sigmoid activation layer, wherein each deconvolution module is composed of one deconvolution layer, one activation layer and one batch regularization layer.
As shown in fig. 3, a residual coding module in the residual self-encoder convolutional neural network model is further described. The residual coding module comprises three convolution layers in total, wherein the convolution kernel of the first convolution layer is 1×1, and the number of channels is the number of channels N of the module C The method comprises the steps of carrying out a first treatment on the surface of the The convolution kernel of the second convolution layer is 3×3, and the number of channels is the number of module channels N C The method comprises the steps of carrying out a first treatment on the surface of the The convolution kernel of the third convolution layer is 1×1, and the number of channels is the number of module channels N C Four times, i.e. 4N C . The residual error coding module constructed in this way belongs to a bottleneck neural network structure, and functions correspondingly realized by the three convolution layers are dimension compression, feature extraction and dimension recovery respectively. Whereas the shortcut connection is used to directly map identities of inputs and outputs bypassing the bottleneck convolutional network structure.
The residual coding module can thus be expressed as a function of:
wherein X and Y represent input and output matrices, respectively, W is a weight parameter variable in a bottleneck neural network in the residual coding module,representing the mapping transformation learned by the bottleneck neural network structure, I (·) represents the identity mapping of the shortcut connection. Bottleneck neural network structure in residual error coding module>Part of the learning is a mapping of the input to the difference between the input and the output.
Still further, the output after each convolution layer process may go through the ReLU activation layer and regularization layer processes before being input into the next convolution layer computation. The ReLU activation layer in the network structure can remove neurons with the value lower than 0 after convolution calculation so as to screen more meaningful characteristics, improve the stability of the network and effectively avoid the problem of gradient explosion.
In the method of the application, the data of the method is derived from the occlusion current signal generated when DNA collected by the solid state nanopore sequencing system passes through the solid state nanopore. The training data set is constructed by carrying out solid-state nanopore sequencing experiment on the designed DNA standard sequence to acquire standard sequence via signal data. Alternatively, a standard signal database may be constructed based on the standard sequence via signal data, and then a large amount of analog signal data may be generated from the standard signal database for use as training data. Meanwhile, the analog signal data can be used for constructing signal data sets with different system noise influences by considering superposition of different degrees of system noise so as to train a neural network model capable of processing different noise influences. And the regularization layer uses Batch Normalization method to normalize the calculation result of the convolution layer to make it follow the same distribution. The introduction of Batch Normalization can make the function of the whole network representation smoother, which is beneficial to the training and subsequent optimization of the stable network.
Step S2 of the embodiment of the present application further includes: and resampling and dimensionality converting the noise base sequence electric signals and the clean base sequence electric signals into two-dimensional signal data of 20 multiplied by 10. The method is characterized in that through the combined action of dimension conversion and convolution calculation, the model is more beneficial to mining relevant characteristics in the time dimension of the current data, the training speed is accelerated to a certain extent, and the noise processing performance is improved. And since the input data is only current amplitude data, the number of input data channels is 1. Then, the number of channels of the first convolution layer is set to 64, so that the dimension of the input data is increased, and the learning ability of the neural network model on the data characteristics is improved. Number of module channels N of eight residual coding modules C Which are sequentially arranged as 64, 32, 16, 8 for mining and extracting features in the signal. Such processing can effectively improve the accuracy of model learning. Correspondingly, the channel numbers of the eight decoding modules are sequentially set to 32, 64, 128 and 256 for signal reconstruction. The rate of neural network learning is set to 0.001 (in other embodiments, any value from 0.1 to 0.001 may be set, and optimization is performed according to the actual situation).
In this embodiment, the initial value of the weight of the neural network model is generated by a gaussian random function, but the weight parameter of the last batch regularization layer of each residual module is initially set to zero, so as to ensure that the initial generation result of the residual branch of the residual module is zero, and the residual coding module will be initially represented as an identity mapping from input to output.
In step S3, according to the residual self-encoder convolutional neural network model and the training parameters thereof, training the residual self-encoder convolutional neural network model with the minimum loss function as a target to complete the construction of the signal noise reduction processing neural network model.
The optimization device for neural network training selects an Adam optimization device, and when the Adam optimization device optimizes network parameters, the Adam optimization device takes first-order momentum (index moving average of gradient) and second-order momentum (index moving average of gradient square) of gradients into account in calculation when calculating iteration step sizes, so that when each time step is subjected to iterative optimization, self-adaptive adjustment can be carried out on the first-order momentum and the second-order momentum of the gradients, convergence of a neural network model to a local optimal solution in optimization is effectively avoided, and meanwhile overall efficiency of optimization is accelerated.
In step S4, the noise electric signal to be processed is preprocessed and then input into the electric signal noise reduction neural network model, and the noise reduction processed signal is output.
In step S2, solid state nanopore sequencing electrical signal datasets that create different noise effects may be selected, and a residual self-encoder neural network model of the corresponding noise is processed targeted according to the training in step S3. In step S4, the noise electrical signal to be processed is input into the neural network model corresponding to the noise influence, so that a corresponding clean signal can be predicted, and a sequencing electrical signal after noise processing is output.
According to the signal noise reduction processing method, the processing speed of the noise signal is extremely high through the trained residual self-encoder neural network model, the noise reduction processing of the signal can be completed in the time lower than 0.1 second under most hardware conditions, and the signal noise reduction processing method has great advantages in practical application, and particularly, the situation that the real-time noise reduction processing is needed is realized.
Compared with the prior art, the solid-state nanopore sequencing electric signal noise reduction processing method combines the self-encoder and the convolutional neural network structure, a bottleneck module in a residual error network is adopted in the encoder part to construct the residual error self-encoder neural network, deep learning is carried out on the characteristics of the sequencing electric signal noise by means of the learning capacity of a convolutional layer and the screening capacity of a ReLU activation layer, the most representative noise characteristics are learned from training set, and then accurate mapping from noise signals to clean signals is established by means of conversion of an inverse convolutional layer and a sigmoid activation layer, so that real-time denoising can be realized.
In a further scheme, the application can also have the following beneficial effects: the bottleneck design of the residual bottleneck module can effectively improve the depth of the model on the premise of not sacrificing the time complexity of the model. And the shortcut connection of the identity mapping can effectively avoid the problem of gradient disappearance or gradient explosion caused by the increase of the number of layers of the network structure without adding additional parameters, and the accuracy improvement caused by the deep network is enjoyed while the stability of the model is ensured. Meanwhile, the residual bottleneck module selects convolution kernels with proper sizes, so that an excellent noise processing effect can be achieved under the condition that no additional pooling layer is introduced, and the problems of insufficient model accuracy and poor noise processing effect caused by the fact that the pooling layer is introduced to reduce model parameters are effectively avoided.
The foregoing is only a preferred embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the technical scope of the present application disclosed in the embodiments of the present application should be covered by the scope of the present application. Therefore, the protection scope of the patent of the application should be subject to the protection scope of the claims.
Claims (5)
1. A solid-state nanopore electrical signal noise reduction processing method based on a residual self-encoder convolutional neural network is characterized by comprising the following steps of: the method comprises the steps of introducing a self-encoder, a residual bottleneck module, a shortcut connection, a convolution layer, a batch regularization layer and an activation layer structure to construct a deep neural network model, training the model by utilizing a solid-state nanopore sequencing electric signal data set, accurately learning a characteristic mode of sequencing electric signal noise, establishing a mapping from a noise signal to a clean signal, and finally predicting and estimating the clean signal corresponding to the noise signal by utilizing the learned mapping;
which comprises the following steps:
step S1, constructing a residual error self-encoder convolutional neural network model;
s2, selecting and creating a training data set, and setting training parameters of the residual error self-encoder convolutional neural network model;
step S3, training the residual error self-encoder convolutional neural network model by taking a minimized loss function as a target according to set model training parameters, and completing construction of a neural network model for noise reduction processing of the solid-state nanopore sequencing electric signal;
s4, inputting the noise electric signal to be processed into a solid-state nanopore sequencing electric signal noise reduction processing neural network model, and outputting a noise reduction processed signal;
the residual self-encoder convolutional neural network model in the step S1 comprises an encoder part and a decoder part, wherein the encoder part comprises a plurality of residual coding modules, and the residual coding modules comprise a plurality of convolutional layers, a plurality of activating layers and a plurality of batch regularization layers; the decoder part consists of a plurality of deconvolution decoding modules, a convolution layer and a sigmoid activation layer, wherein the deconvolution decoding modules consist of a plurality of deconvolution layers, a plurality of activation layers and a plurality of batch regularization layers;
wherein the residual coding module comprises a plurality of convolution layers with convolution kernels larger than 1×1 and a plurality of convolution layers with convolution kernels of 1×1;
wherein, the residual coding module comprises a convolution layer with a convolution kernel size of 3×3 and two convolution layers with convolution kernels of 1×1; the convolution kernel sizes of the first convolution layer and the third convolution layer of the residual error coding module are 1 multiplied by 1, the convolution kernel size of the second convolution layer is 3 multiplied by 3, and the constructed residual error coding module is of a bottleneck neural network structure;
wherein, the loss function in step S3 is a mean square error function:
wherein X is i 、Y i And respectively creating a noise signal and a clean signal in the training set, wherein θ is a weight, n represents the number of data points in the signal, and F (-) is a mapping from the noise signal to the clean signal, which is obtained after training.
2. The method for noise reduction of solid state nanopore electrical signals based on residual self encoder convolutional neural network of claim 1, wherein the residual encoding module comprises a shortcut connection for directly mapping identities of inputs and outputs bypassing a bottleneck convolutional network structure.
3. The method for noise reduction processing of solid state nanopore electrical signals based on residual self encoder convolutional neural network of claim 1, wherein the deconvolution decoding module of the decoder portion consists of an deconvolution layer, an activation layer and a batch regularization layer.
4. The residual self-encoder convolutional neural network-based solid state nanopore electrical signal noise reduction processing method of claim 1, wherein the training dataset in step S2 consists of clean sequencing electrical signals and corresponding noise signals.
5. The solid state nanopore electrical signal noise reduction processing method based on the residual self encoder convolutional neural network according to claim 1, wherein in step S3, an initial value of a weight of a neural network model of the solid state nanopore sequencing electrical signal noise reduction processing is generated by a gaussian random function, and a weight parameter of a last batch regularization layer of a residual module is initially set to zero.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111032628.6A CN113743301B (en) | 2021-09-03 | 2021-09-03 | Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111032628.6A CN113743301B (en) | 2021-09-03 | 2021-09-03 | Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113743301A CN113743301A (en) | 2021-12-03 |
| CN113743301B true CN113743301B (en) | 2023-09-26 |
Family
ID=78735415
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111032628.6A Active CN113743301B (en) | 2021-09-03 | 2021-09-03 | Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113743301B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114781445B (en) * | 2022-04-11 | 2022-11-18 | 山东省人工智能研究院 | Deep neural network electrocardiosignal noise reduction method based on interpretability |
| CN116904569B (en) * | 2023-09-13 | 2023-12-15 | 北京齐碳科技有限公司 | Signal processing method, device, electronic equipment, medium and product |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204467A (en) * | 2016-06-27 | 2016-12-07 | 深圳市未来媒体技术研究院 | A kind of image de-noising method based on cascade residual error neutral net |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11486873B2 (en) * | 2016-03-31 | 2022-11-01 | Ontera Inc. | Multipore determination of fractional abundance of polynucleotide sequences in a sample |
| US10883962B2 (en) * | 2017-01-19 | 2021-01-05 | University Of Kentucky Research Foundation | Electrical double layer in nanopores for detection and identification of molecules and submolecular units |
-
2021
- 2021-09-03 CN CN202111032628.6A patent/CN113743301B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204467A (en) * | 2016-06-27 | 2016-12-07 | 深圳市未来媒体技术研究院 | A kind of image de-noising method based on cascade residual error neutral net |
Non-Patent Citations (2)
| Title |
|---|
| 固态纳米孔上端粒序列的共价修饰与 Ｇ－四联体折叠监测;王森 等;《微纳电子技术》;第57卷(第6期);第483-491页 * |
| 基于ＬｉＣｌ盐浓度梯度的固态纳米孔ＤＮＡ分子检测;张庞 等;《微纳电子技术》;第58卷(第1期);第72-79页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113743301A (en) | 2021-12-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109035142B (en) | Satellite image super-resolution method combining countermeasure network with aerial image prior | |
| CN109816593B (en) | Super-resolution image reconstruction method for generating countermeasure network based on attention mechanism | |
| CN113743301B (en) | Solid-state nanopore sequencing electric signal noise reduction processing method based on residual self-encoder convolutional neural network | |
| CN108416755A (en) | A kind of image de-noising method and system based on deep learning | |
| CN111932461B (en) | Self-learning image super-resolution reconstruction method and system based on convolutional neural network | |
| CN111815516B (en) | Super-resolution reconstruction method for weak supervision infrared remote sensing image | |
| CN110533591B (en) | Super-resolution image reconstruction method based on codec structure | |
| CN108734675A (en) | Image recovery method based on mixing sparse prior model | |
| CN111598804B (en) | Deep learning-based image multi-level denoising method | |
| CN115187610A (en) | Neuron morphological analysis method and device based on graph neural network and storage medium | |
| CN113421187B (en) | Super-resolution reconstruction method, system, storage medium and equipment | |
| CN111861886A (en) | Image super-resolution reconstruction method based on multi-scale feedback network | |
| CN114882524A (en) | Monocular three-dimensional gesture estimation method based on full convolution neural network | |
| CN111461978A (en) | Attention mechanism-based resolution-by-resolution enhanced image super-resolution restoration method | |
| CN109063834B (en) | Neural network pruning method based on convolution characteristic response graph | |
| CN113194493B (en) | Wireless network data missing attribute recovery method and device based on graph neural network | |
| CN114494047A (en) | Biological image denoising method based on dual-enhancement residual error network | |
| CN109272450B (en) | Image super-resolution method based on convolutional neural network | |
| CN114648048A (en) | Electrocardiosignal noise reduction method based on variational self-coding and PixelCNN model | |
| CN113450364B (en) | Tree-shaped structure center line extraction method based on three-dimensional flux model | |
| CN115909077A (en) | Hyperspectral image change detection method based on unsupervised spectrum unmixing neural network | |
| CN115375537A (en) | Nonlinear sensing multi-scale super-resolution image generation system and method | |
| CN113935908B (en) | Remote sensing image cloud removing method based on double-branch channel and feature strengthening mechanism | |
| CN115131414A (en) | Unmanned aerial vehicle image alignment method based on deep learning, electronic equipment and storage medium | |
| CN114998137A (en) | Ground penetrating radar image clutter suppression method based on generation countermeasure network |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |