Pulsar search network training method, pulsar search method, device, and equipment

Abstract

The present invention provides a pulsar search network training method, including: based on a training sample set and the pulse profile image of the candidate, a first training is performed on the backbone network of the pulsar search network to determine the initial network parameters of the backbone network; based on the training sample set and the pulsar label value, a second training is performed on the backbone network of the pulsar search network to obtain the target parameters of the backbone network of the pulsar search network, thereby being able to stably improve the accuracy of pulsar search network training, reduce overfitting of the pulsar search network, and enhance the generalization ability of the pulsar search network model under the premise of reducing the total amount of training data and without the need for re-supervised training. The present invention also provides a pulsar search method, device, electronic device, computer program product, and storage medium. The embodiments of the present invention can be applied to various scenarios such as cloud technology, artificial intelligence, smart transportation, and assisted driving.

Description

Technical Field

[0001] The present invention relates to image processing technology in the field of artificial intelligence, and in particular to a pulsar search network training method, a pulsar search method, a device, an electronic device, a computer program product and a storage medium. Background Art

[0002] Classification recognition based on deep learning has always been an important tool for solving large amounts of data classification in various application scenarios. For example, in application scenarios such as image and natural language processing, large-scale classification and recognition of large amounts of data can quickly and accurately obtain relevant classification prediction results and accelerate the functional realization of the application scenario.

[0003] In the classification prediction of images, different tasks can be performed according to the different application scenarios deployed. When searching for pulsars, a large number of candidate files are usually generated in the pulsar search task. For example, FAST will have an annual increase of PB and is expected to generate tens of millions of candidates. Therefore, in the search for pulsar candidates, both the accuracy of the search model and the inference speed of the search model will be crucial. Summary of the invention

[0004] In view of this, the embodiments of the present invention provide a pulsar search network training method, a pulsar search method device, an electronic device, a computer program product and a storage medium, which can realize the training of the pulsar search network, and can stably improve the accuracy of the pulsar search network training, reduce the overfitting of the pulsar search network, and enhance the generalization ability of the pulsar search network model, so as to facilitate the deployment of the trained pulsar search network model in the network terminal in time, and realize the large-scale application of the pulsar search network model.

[0005] The technical solution of the embodiment of the present invention is achieved as follows:

[0006] An embodiment of the present invention provides a pulsar search network training method, comprising:

[0007] Obtaining a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal;

[0008] Combining the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network;

[0009] Acquire a pulse profile image and a pulsar label value of the candidate body corresponding to the pulsar search candidate body signal;

[0010] Based on the training sample set and the pulse profile image of the candidate, performing a first training on the backbone network of the pulsar search network to determine initial network parameters of the backbone network;

[0011] Based on the training sample set and the pulsar label value, a second training is performed on the backbone network of the pulsar search network to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network.

[0012] The embodiment of the present invention also provides a pulsar search method, including:

[0013] Receive unlabeled pulsar search candidate signals;

[0014] Acquire a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal;

[0015] Processing the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result;

[0016] The image processing result is processed by the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal.

[0017] The embodiment of the present invention further provides a pulsar search network training device, the device comprising:

[0018] An information transmission module is used to obtain a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal;

[0019] An information processing module, used for combining the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network;

[0020] The information processing module is used to obtain the pulse profile image and pulsar label value of the candidate body corresponding to the pulsar search candidate body signal;

[0021] The information processing module is used to perform a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate, and determine the initial network parameters of the backbone network;

[0022] The information processing module is used to perform a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value, so as to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network.

[0023] In the above scheme,

[0024] The information processing module is used to determine the noise values ​​of the time phase distribution image and the frequency phase distribution image, and perform noise removal processing on the determined time phase distribution image and the frequency phase distribution image respectively;

[0025] The information processing module is used to perform standardization processing on the time phase distribution image and the frequency phase distribution image after the noise removal processing, so as to obtain a time phase distribution image and a frequency phase distribution image that meet the single-channel grayscale image standard;

[0026] The information processing module is used to combine the time phase distribution image and the frequency phase distribution image that meet the single-channel grayscale image standard to form a training sample set for the pulsar search network.

[0027] In the above scheme,

[0028] The information processing module is used to determine a dynamic noise threshold that matches the use environment of the pulsar search network according to the position of the target area corresponding to the pulsar search candidate signal;

[0029] The information processing module is used to perform noise reduction processing on the time phase distribution image and the frequency phase distribution image according to the dynamic noise threshold to form a time phase distribution image and a frequency phase distribution image that match the dynamic noise threshold.

[0030] In the above scheme,

[0031] The information processing module is used to determine a fixed noise threshold that matches the use environment of the pulsar search network according to the type of radio telescope corresponding to the pulsar search candidate signal;

[0032] The information processing module is used to perform noise reduction processing on the time phase distribution image and the frequency phase distribution image according to the fixed noise threshold to form a time phase distribution image and a frequency phase distribution image that match the fixed noise threshold.

[0033] In the above scheme,

[0034] The information processing module is used to obtain the original pulse profile image of the candidate body corresponding to the pulsar search candidate body signal;

[0035] The information processing module is used to perform standardization processing on the sequence value of the original pulse profile image to obtain the standard sequence value of the original pulse profile image;

[0036] The information processing module is used to adjust the standard sequence value of the original pulse profile image to obtain the pulse profile image corresponding to the pulsar search candidate signal to achieve matching with the accuracy of the pulsar search network.

[0037] In the above scheme,

[0038] The information processing module is used to determine the label value of the pulse profile image of the candidate body based on the pulse profile image of the candidate body;

[0039] The information processing module is used to determine a first loss function corresponding to the first training based on the label value of the pulse profile image through a pre-trained fully connected layer network corresponding to the backbone network;

[0040] The information processing module is used to bring the training samples in the training sample set into the first loss function;

[0041] The information processing module is used to determine the initial network parameters of the backbone network when the first loss function meets the first convergence condition.

[0042] In the above scheme,

[0043] The information processing module is used to determine a second loss function corresponding to the second training based on the pulsar label value through the classification task fully connected layer network corresponding to the backbone network;

[0044] The information processing module is used to bring the training samples in the training sample set into the second loss function;

[0045] The information processing module is used to determine the update parameters of the backbone network when the second loss function satisfies the second convergence condition;

[0046] The information processing module is used to adjust the initial parameters of the backbone network of the pulsar search network based on the updated parameters of the backbone network to obtain the target parameters of the backbone network of the pulsar search network.

[0047] The embodiment of the present invention further provides a pulsar search device, comprising:

[0048] A signal transmission module, used for receiving unlabeled pulsar search candidate signals;

[0049] A signal processing module, used to obtain a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal;

[0050] The signal processing module is used to process the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result;

[0051] The signal processing module is used to process the image processing result through the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal.

[0052] An embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0053] A memory for storing executable instructions;

[0054] The processor is used to implement the aforementioned pulsar search network training method when running the executable instructions stored in the memory.

[0055] An embodiment of the present invention further provides a computer-readable storage medium storing executable instructions, wherein the executable instructions, when executed by a processor, implement the aforementioned pulsar search network training method.

[0056] The embodiments of the present invention have the following beneficial effects:

[0057] The present invention obtains a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal; combines the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network; obtains a pulse profile image and a pulsar label value of the candidate corresponding to the obtained pulsar search candidate signal; performs a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate to determine the initial network parameters of the backbone network; performs a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network, thereby achieving the training of the pulsar search network, and stably improving the accuracy of the pulsar search network training, reducing the overfitting of the pulsar search network, and enhancing the generalization ability of the pulsar search network model under the premise of reducing the total amount of training data and without the need for re-supervised training, so as to facilitate the deployment of the timely trained pulsar search network model in the network terminal and realize the large-scale application of the pulsar search network model. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the use environment of the pulsar search network training method provided by an embodiment of the present invention;

[0059] Figure 2 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention;

[0060] Figure 3 An optional flowchart of a pulsar search network training method provided in an embodiment of the present invention;

[0061] Figure 4A A schematic diagram of a sample set of a pulsar search network in an embodiment of the present invention;

[0062] Figure 4B A schematic diagram of a sample set of a pulsar search network in an embodiment of the present invention;

[0063] Figure 5 An optional flowchart of a pulsar search network training method provided in an embodiment of the present invention;

[0064] Figure 6 A schematic diagram of the network structure of a pulsar search network in an embodiment of the present invention;

[0065] Figure 7 An optional flowchart of a pulsar search method provided in an embodiment of the present invention;

[0066] Figure 8 Schematic diagram of searching for candidate pulsar signals in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.

[0068] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it can be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0069] Before further describing the embodiments of the present invention in detail, the nouns and terms involved in the embodiments of the present invention are described. The nouns and terms involved in the embodiments of the present invention are subject to the following interpretations.

[0070] 1) In response, it is used to indicate the conditions or states on which the executed operations depend. When the dependent conditions or states are met, one or more operations executed may be in real time or have a set delay. Unless otherwise specified, there is no restriction on the order in which the multiple operations executed are executed.

[0071] 2) Downsampling: for a sample sequence, a sample is sampled once every several sample values, so that the new sequence is the downsampling of the original sequence. For example, for an image I with a size of M*N, it is downsampled s times, that is, a resolution image with a size of (M / s)*(N / s) is obtained, where s should be the common divisor of M and N.

[0072] 3) Model training, multi-classification learning of image datasets. The model can be built using deep learning frameworks such as TensorFlow and torch, and uses multiple layers of neural network layers such as CNN to form a multi-classification model. The input of the model is a three-channel or original channel matrix formed by reading the image through tools such as openCV. The model output is a multi-classification probability, and the image category is finally output through algorithms such as softmax. During training, the model approaches the correct trend through objective functions such as cross entropy.

[0073] 4) Noisy recognition: Training of image recognition tasks based on noisy samples. Noisy samples include samples with wrong category labels and samples with inaccurate category labels. For example, the image does not completely correspond to the category label, the concepts of the two category labels partially overlap, and the image has the attributes of the above two category labels, but only has one category label.

[0074] 5) Contrastive loss: Contrastive loss function can learn a mapping relationship. This mapping relationship can make the points of the same category but far away in high-dimensional space closer after being mapped to low-dimensional space by the function, and the points of different categories but close distances become farther in low-dimensional space after mapping. The result is that in low-dimensional space, points of the same category will produce a clustering effect, and means of different categories will be separated. It is similar to Fisher dimensionality reduction, but Fisher dimensionality reduction does not have the effect of out-of-sample extension and cannot act on new samples.

[0075] 6) Client, a carrier that implements specific functions in a terminal, for example, a mobile client (APP) is a carrier of specific functions in a mobile terminal, such as a program that performs the function of user gesture recognition.

[0076] 7) Soft max: Normalized exponential function, which is a generalization of the logic function. It can "compress" a K-dimensional vector containing any real number into another K-dimensional real vector, so that each element is between [0, 1] and the sum of all elements is 1.

[0077] Figure 1 A schematic diagram of a use scenario of the pulsar search network training method provided by an embodiment of the present invention, see Figure 1, the terminal (including the terminal 10-1 and the terminal 10-2) is provided with a corresponding client capable of executing different functions, wherein the server 200 stores the pulse signal received by the radio telescope 400, and the client is the terminal (including the terminal 10-1 and the terminal 10-2) which obtains different pulsar candidate signals from the corresponding server 200 through the network 300 for browsing, or obtains the identification result of the corresponding pulsar candidate signal, and the terminal is connected to the server 200 through the network 300, and the network 300 can be a wide area network or a local area network, or a combination of the two, and a wireless link is used to realize data transmission, wherein , the corresponding pulsar candidate signals obtained by the terminal (including the terminal 10-1 and the terminal 10-2) from the corresponding server 200 through the network 300 may be the same or different. For example, the terminal (including the terminal 10-1 and the terminal 10-2) may obtain the pulsar signal with labeled information corresponding to a known pulsar from the corresponding server 200 through the network 300, or may obtain the pulsar candidate signal without labeled information from the corresponding server 200 through the network 300, or may obtain the final recognition result of the pulsar candidate signal without labeled information (for unknown pulsars). The server 200 may store the pulsar candidate signals without labeled information corresponding to different target objects and the corresponding recognition results. In some embodiments of the present invention, the pulsar candidate signals without labeled information stored in the server 200 are collected by the radio telescope 400. At the same time, the server 200 can also store pulse signals of known pulsars, for example: (1) HTRU: High Time Resolution Universe Survey, the data set contains 1196 known pulsars and 89996 non-pulsars, most of which are relatively strong pulsar signals, and the number of negative samples is far greater than the positive samples, which is an extremely imbalanced classification task. (2) PMPS, PMPS-26k, P309: Parkes Multi-beam Pulsar Survey, PMPS has a total of 4.25T, most of which are preliminary screening samples, PMPS-26k contains 2000 pulsar samples, 20000 radio interferences, 2000 other signals, and 2000 unlabeled samples, totaling 26k. P309 contains 2698 pulsars and 1656 radio interferences. In general, the dataset contains many types of samples, and the positive samples include both samples with strong signals and samples with weak signals.Taking the new generation 500m aperture spherical radio telescope (FAST Five Hundred Meter Aperture Spherical Telescope) as an example, the data stored in the server contains 1163 pulsars (837 available training sets) and 14319 non-pulse signals (998 available training sets).

[0078] The pulsar search network trained by the pulsar search network training method provided in the present application can realize the search for pulsars using unlabeled pulsar search candidate signals. Of course, before use, the pulsar search network needs to be trained first, specifically including:

[0079] Acquire a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal; combine the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network; acquire a pulse contour image and a pulsar label value of the candidate corresponding to the acquired pulsar search candidate signal; perform a first training on a backbone network of the pulsar search network based on the training sample set and the pulse contour image of the candidate to determine initial network parameters of the backbone network; perform a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label values ​​to adjust the initial parameters of the backbone network of the pulsar search network and obtain target parameters of the backbone network of the pulsar search network.

[0080] In order to better understand the method provided in the embodiments of the present application, artificial intelligence, its various branches, and the application fields, cloud technology, and artificial intelligence cloud services involved in the method provided in the embodiments of the present application are first explained.

[0081] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0082] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. Basic artificial intelligence technologies generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning. The following is a description of each direction.

[0083] Natural language processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, that is, the language people use in daily life, so it is closely related to the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question answering, knowledge graph and other technologies.

[0084] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning and other technologies.

[0085] Cloud technology refers to a hosting technology that unifies hardware, software, network and other resources in a wide area network or local area network to achieve data computing, storage, processing and sharing. Cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model. It can form a resource pool that can be used on demand and is flexible and convenient. Cloud computing technology will become an important support. The background services of technical network systems require a large amount of computing and storage resources, such as video websites, picture websites and more portal websites. With the high development and application of the Internet industry, in the future, each item may have its own identification mark, and all need to be transmitted to the background system for logical processing. Data of different levels will be processed separately. All kinds of industry data require strong system backing support, which can only be achieved through cloud computing.

[0086] The so-called artificial intelligence cloud service, also generally referred to as AI as a Service (AIaaS), is a mainstream service mode of artificial intelligence platform. Specifically, the AIaaS platform will split several common AI services and provide independent or packaged services in the cloud. This service model is similar to opening an AI theme mall: all developers can access and use one or more artificial intelligence services provided by the platform through API interfaces. Some senior developers can also use the AI ​​framework and AI infrastructure provided by the platform to deploy and operate their own cloud artificial intelligence services.

[0087] The solutions provided in the embodiments of the present application involve technologies such as natural language processing, machine learning, and artificial intelligence cloud services of artificial intelligence, which are specifically illustrated by the following embodiments.

[0088] The pulsar search network training method provided in the embodiment of the present application will be explained in combination with the exemplary application and implementation of the terminal provided in the embodiment of the present application.

[0089] The structure of the electronic device of the embodiment of the present invention is described in detail below. The electronic device can be implemented in various forms, such as a dedicated terminal with an endoscopic image processing function, or an electronic device or a cloud server with an endoscopic image processing function, such as the aforementioned Figure 1 Server 200 in. Figure 2 The schematic diagram of the structure of the electronic device provided by the embodiment of the present invention can be understood as follows: Figure 2 Only exemplary structures of electronic devices are shown, not all structures, and may be implemented as needed. Figure 2 Partial or complete structure shown.

[0090] The electronic device provided in the embodiment of the present invention includes: at least one processor 201, a memory 202, a user interface 203 and at least one network interface 204. The various components in the electronic device are coupled together through a bus system 205. It can be understood that the bus system 205 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 205 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, the bus system 205 is not used in the following examples. Figure 2 Various buses are labeled as bus system 205 .

[0091] The user interface 203 may include a display, a keyboard, a mouse, a trackball, a click wheel, a key, a button, a touch pad or a touch screen, etc. The terminals in the embodiments of the present invention include but are not limited to mobile phones, computers, intelligent voice interaction devices, smart home appliances, vehicle-mounted terminals, etc. The embodiments of the present invention can be applied to various scenarios, including but not limited to cloud technology, artificial intelligence, smart transportation, assisted driving, etc. When the pulsar search network training method provided by the present invention is executed by different terminals, the specific usage scenarios are not limited by the present invention.

[0092] It is understood that the memory 202 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. The memory 202 in the embodiment of the present invention can store data to support the operation of the terminal (such as 10-1). Examples of these data include: any computer program for operating on the terminal (such as 10-1), such as an operating system and an application. Among them, the operating system includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., which are used to implement various basic services and process hardware-based tasks. The application can include various applications.

[0093] In some embodiments, the pulsar search network training device provided in the embodiment of the present invention can be implemented in a combination of software and hardware. As an example, the pulsar search network training device provided in the embodiment of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the pulsar search network training method provided in the embodiment of the present invention. For example, the processor in the form of a hardware decoding processor can adopt one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), or other electronic components.

[0094] As an example of a pulsar search network training device provided in an embodiment of the present invention being implemented by a combination of software and hardware, the pulsar search network training device provided in an embodiment of the present invention can be directly embodied as a combination of software modules executed by a processor 201, and the software module can be located in a storage medium, and the storage medium is located in a memory 202. The processor 201 reads the executable instructions included in the software module in the memory 202, and combines with necessary hardware (for example, including the processor 201 and other components connected to the bus 205) to complete the pulsar search network training method provided in an embodiment of the present invention.

[0095] As an example, the processor 201 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0096] As an example of hardware implementation of the pulsar search network training device provided in an embodiment of the present invention, the device provided in an embodiment of the present invention can be directly executed by a processor 201 in the form of a hardware decoding processor, for example, one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), or other electronic components to implement the pulsar search network training method provided in an embodiment of the present invention.

[0097] The memory 202 in the embodiment of the present invention is used to store various types of data to support the operation of the electronic device. Examples of such data include: any executable instructions for operating on the electronic device, such as executable instructions, and the program for implementing the method for searching for network training from pulsars in the embodiment of the present invention can be included in the executable instructions.

[0098] In other embodiments, the pulsar search network training device provided by the embodiment of the present invention can be implemented in software. Figure 2 The pulsar search network training device 2020 stored in the memory 202 is shown, which may be software in the form of a program and a plug-in, and includes a series of modules. As an example of a program stored in the memory 202, the pulsar search network training device 2020 may be included. The pulsar search network training device 2020 includes the following software modules:

[0099] The information transmission module 2081 is used to obtain the time phase distribution image and the frequency phase distribution image corresponding to the pulsar search candidate signal.

[0100] The information processing module 2082 is used to combine the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network.

[0101] The information processing module 2082 is used to obtain the pulse profile image and pulsar label value of the candidate body corresponding to the pulsar search candidate body signal.

[0102] The information processing module 2082 is used to perform a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate, and determine the initial network parameters of the backbone network.

[0103] The information processing module 2082 is used to perform a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value, so as to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network.

[0104] It should be noted that after the pulsar search network training is completed, it can also be deployed in an electronic device (such as a star search server or a data processing server of a radio telescope) to process the received pulsar search candidate signals. Therefore, the electronic device can also include:

[0105] The signal transmission module is used to receive unlabeled pulsar search candidate signals.

[0106] The signal processing module is used to obtain a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal.

[0107] The signal processing module is used to process the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result.

[0108] The signal processing module is used to process the image processing result through the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal.

[0109] according to Figure 2 The electronic device shown, in one aspect of the present application, the present application also provides a computer program product or a computer program, the computer program product or the computer program includes a computer instruction, and the computer instruction is stored in a computer-readable storage medium. The processor of the computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the method provided in various optional implementations of the pulsar search network training method provided in the present application.

[0110] Combination Figure 2 The pulsar search network training device shown in the figure illustrates the pulsar search network training method provided by the embodiment of the present invention, see Figure 3 , Figure 3 An optional flow chart of the pulsar search network training method provided by an embodiment of the present invention, it can be understood that: Figure 3 The steps shown can be performed by various electronic devices running the pulsar search network training device, such as a dedicated terminal, server or server cluster with a pulsar search function. Figure 3 The steps shown are explained.

[0111] Step 301: The pulsar search network training device obtains a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal.

[0112] Step 302: The pulsar search network training device combines the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network.

[0113] Among them, reference Figure 4A , Figure 4A Schematic diagram of a sample set of a pulsar search network in an embodiment of the present invention, wherein the time phase distribution image and the frequency phase distribution image are combined to form a training sample set of the pulsar search network, which can be achieved in the following manner:

[0114] Determine the noise value of the time phase distribution image and the frequency phase distribution image, perform denoising on the determined time phase distribution image and the frequency phase distribution image respectively; perform standardization on the denoised time phase distribution image and the frequency phase distribution image respectively to obtain a time phase distribution image and a frequency phase distribution image that meet the single-channel grayscale image standard; combine the time phase distribution image and the frequency phase distribution image that meet the single-channel grayscale image standard to form a training sample set for the pulsar search network. Figure 1 Taking the radio telescope shown in the figure as an example, the time-phase distribution image resolution of the fid format file obtained by receiving the pulsar signal is 256*64 after feature reading, and the frequency-phase distribution image resolution is 96*64; while the time-phase distribution image resolution of the phcx format file obtained by other radio telescopes is generally (16~24)*64 after feature extraction, and the frequency-phase distribution image resolution is 16*64. The sizes of the above images are not equal. In addition, the feature background distinction between the phase distribution image and the frequency phase distribution image is also different. For example, the pulse feature in the phcx format file image is black and the background is white, while the pulse feature in the fid format file image is white and the background is black. Therefore, denoising and standardization are required to facilitate the training and use of the pulsar search network.

[0115] In some embodiments of the present invention, reference Figure 4B , Figure 4B The sample set diagram of the pulsar search network in the embodiment of the present invention is shown in Figure 1. The time-phase distribution diagram and the frequency-phase distribution diagram can be subjected to data standardization operations respectively to obtain the corresponding 64x64 single-channel grayscale image, and then the two grayscale images are stacked in the channel dimension into a 2-channel 2-dimensional signal as the input feature vector of the backbone network. In the pre-training task and classification task, the output result of the backbone network is used to adjust the output of the backbone network to an output dimension that can calculate the task loss through different fully connected layers.

[0116] In some embodiments of the present invention, determining the noise values ​​of the time phase distribution image and the frequency phase distribution image, and performing noise removal processing on the time phase distribution image and the frequency phase distribution image respectively, can be achieved in the following manner:

[0117] According to the position of the target area corresponding to the signal of the pulsar search candidate, a dynamic noise threshold matching the use environment of the pulsar search network is determined; according to the dynamic noise threshold, the time phase distribution image and the frequency phase distribution image are subjected to noise reduction processing to form a time phase distribution image and a frequency phase distribution image matching the dynamic noise threshold. Wherein, when detecting pulsars in the low-frequency radio wave band (for example, below 350MHz), since the radiation spectrum presents a power-law spectrum above a relatively low frequency lower limit, it is expected to have a higher flow density than the high frequency, so it is easier to detect. However, due to the different galaxies in which pulsars appear, and at the same time, since the interstellar medium has different refractive indices (i.e., dispersion) for radio waves of different frequencies, the pulsar pulses of different frequencies aligned at the signal emission site will be staggered after a long-distance propagation, and the periodic pulse profile that is very obvious in the time domain will become almost impossible to directly identify and detect after the dispersion effect. Therefore, detecting pulsars in the low-frequency radio band requires a dynamic noise threshold to perform noise reduction processing on the time phase distribution image and the frequency phase distribution image to form a time phase distribution image and a frequency phase distribution image that match the dynamic noise threshold to ensure the accuracy of the pulsar search network.

[0118] In some embodiments of the present invention, determining the noise values ​​of the time phase distribution image and the frequency phase distribution image, and performing noise removal processing on the time phase distribution image and the frequency phase distribution image respectively, can be achieved in the following manner:

[0119] According to the type of radio telescope corresponding to the pulsar search candidate signal, a fixed noise threshold matching the use environment of the pulsar search network is determined; the time phase distribution image and the frequency phase distribution image are subjected to noise reduction processing according to the fixed noise threshold to form a time phase distribution image and a frequency phase distribution image matching the fixed noise threshold. Among them, since the number of pulsars that have been detected and can be used to adjust the detection algorithm parameters is very limited, and unknown pulsars cannot be used as a correction training sample set before they are actually searched. Therefore, for a brand new radio telescope, or an existing radio telescope that has never detected a pulsar, the time phase distribution image and the frequency phase distribution image can be subjected to noise reduction processing according to the dynamic noise threshold to form a time phase distribution image and a frequency phase distribution image matching the dynamic noise threshold, and a noise threshold matching the radio telescope can be determined.

[0120] Furthermore, when the pulsar search network is fixed in the terminal device of the radio telescope, the time phase distribution image and the frequency phase distribution image are denoised according to a fixed noise threshold to form a time phase distribution image and a frequency phase distribution image that match the fixed noise threshold, which can continue to reduce the training time of the pulsar search network and improve the training efficiency of the pulsar search network.

[0121] Step 303: The pulsar search network training device obtains the pulse profile image and pulsar label value of the candidate body corresponding to the pulsar search candidate body signal.

[0122] In some embodiments of the present invention, obtaining a pulse profile image of a candidate body corresponding to a pulsar search candidate body signal may be achieved by:

[0123] The original pulse profile image of the candidate body corresponding to the pulsar search candidate body signal is obtained; the sequence value of the original pulse profile image is standardized to obtain the standard sequence value of the original pulse profile image; the standard sequence value of the original pulse profile image is adjusted to obtain the pulse profile image corresponding to the pulsar search candidate body signal to match the accuracy of the pulsar search network. In order to increase the difficulty of the pre-training task, the original pulse profile sequence can be standardized, and then the 64-dimensional sequence can be linearly interpolated to a 100-dimensional sequence as the label value of the pre-training task. When the training accuracy increases, the linear interpolation of sequences of different dimensions can be flexibly adjusted to meet the different needs of the pulsar search network.

[0124] Step 304: The pulsar search network training device performs a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate, and determines the initial network parameters of the backbone network.

[0125] Step 305: The pulsar search network training device performs a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value, so as to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network.

[0126] Continue to combine Figure 2 The pulsar search network training device shown in the figure illustrates the pulsar search network training method provided by the embodiment of the present invention, see Figure 5 , Figure 5 An optional flow chart of the pulsar search network training method provided by an embodiment of the present invention, it can be understood that: Figure 5 The steps shown can be performed by various electronic devices running the pulsar search network training device, such as a dedicated terminal, server or server cluster with a pulsar search function. Figure 5 The steps shown are explained.

[0127] Step 501: The pulsar search network training device determines the label value of the pulse profile image of the candidate based on the pulse profile image of the candidate.

[0128] Step 502: The pulsar search network training device determines a first loss function corresponding to the first training based on the label value of the pulse profile image through the pre-trained fully connected layer network corresponding to the backbone network.

[0129] Step 503: The pulsar search network training device brings the training samples in the training sample set into the first loss function.

[0130] Step 504: When the first loss function satisfies the first convergence condition, the pulsar search network training device determines the initial network parameters of the backbone network.

[0131] refer to Figure 6 , Figure 6The network structure diagram of the pulsar search network in the embodiment of the present invention is shown in FIG. 5 , wherein, through the processing of steps 501 to 504, the time-phase distribution diagram and the frequency-phase distribution diagram of the candidate signal are used as the input features of the pulsar search network, the pulse contour diagram of the candidate is used as the label value, and the root mean square error is used as the loss function to calculate the initial network parameters of the pulsar search network. Through the backbone network, the time-phase distribution diagram and the frequency-phase distribution diagram can be summed according to the column dimension, and then the two summed sequences are summed again to obtain the 64-dimensional original pulse contour sequence, wherein, Figure 6 The backbone network in the network structure shown may be a ResNet18 network, and the network connected to the backbone network may include multiple fully connected layers. It is worth noting that in order to balance the space occupied by image features and image quantization coding, the dimension of image features may be set smaller than the dimension of image quantization coding, otherwise the dimension of image features is too large, which will occupy too much storage space.

[0132] The process of training the pulsar search network mainly involves the following steps.

[0133] 1) Parameter initialization. For the initial feature extraction network, the weight parameters can be initialized by training with open source image sets (such as ImageNet image sets, Open Image image sets, etc.). For other networks, such as the maximum pooling network, the embedded representation network, and the quantized coding network, the weight parameters can be randomly initialized, for example, using a Gaussian distribution with a variance of 0.01 and a mean of 0.

[0134] 2) Determine the weight parameters that need to be learned (i.e., need to be updated). This can be divided into two training stages. In the first training stage, the weight parameters of the initial feature extraction network, the maximum pooling network, and the embedding representation network are updated; in the second training stage, the weight parameters of the quantization encoding network are updated.

[0135] 3) Determine the learning rate. In the embodiment of the present application, each network in the pulsar search network can use the same learning rate, for example, the learning rate is 0.005. After every 10 rounds of iteration, the learning rate can be updated to 0.1 times the original value.

[0136] 4) Learning process (training process). Here, the pulsar search network can be iterated by epoch rounds. Epoch is an integer greater than 1 and can be set according to the actual application scenario. In each round of iteration, all samples are processed once. For example, all similar image pairs are divided according to Batch Size (which can be set according to the actual application scenario) to obtain Nb batches. For each batch (Batch), the following processing is performed:

[0137] ① Forward propagation processing. Here, each training image in the batch is forward propagated through the pulsar search network to obtain image features and image quantization coding.

[0138] ② Loss value calculation: Here, the feature extraction network and the quantization encoding network correspond to different loss values.

[0139] ③ Weight parameter update. For example, the stochastic gradient descent algorithm or other gradient descent algorithms can be used to back-propagate the loss value, and in the process of back-propagation, the weight parameters of the network to be trained are updated along the gradient descent direction.

[0140] It should be noted that the specific structure in the above embodiment is only exemplary, and other model structures can be selected as needed. For example, the backbone network can also use resnet50, inceptionv4, resnet18, etc., which is not limited in this application.

[0141] After completing the pre-training task through steps 501 to 504 , proceed to step 505 .

[0142] Step 505: The pulsar search network training device determines a second loss function corresponding to the second training based on the pulsar label value through the classification task fully connected layer network corresponding to the backbone network.

[0143] Step 506: The pulsar search network training device brings the training samples in the training sample set into the second loss function.

[0144] Step 507: When the second loss function satisfies the second convergence condition, the pulsar search network training device determines the update parameters of the backbone network.

[0145] Step 508: The pulsar search network training device adjusts the initial parameters of the backbone network of the pulsar search network based on the updated parameters of the backbone network to obtain the target parameters of the backbone network of the pulsar search network.

[0146] Among them, through steps 505 to 508, the feature dimension is mapped to the classification dimension through the fully connected layer, and then the predicted value and the pulsar label value are used to calculate the cross entropy loss. In order to prevent the overfitting phenomenon of the pre-trained backbone network during the optimization process of the classification task, the learning rate of the backbone network parameters can be 0.25 times the learning rate of the fully connected parameters of the classification task.

[0147] After the pulsar search network is trained through the previous embodiment, it can be deployed in the corresponding data processing device to predict the acquired pulsar search candidate signals, see Figure 7 , Figure 7 An optional flow chart of the pulsar search method provided in an embodiment of the present invention is provided. It can be understood that: Figure 7 The steps shown can be performed by various electronic devices running a pulsar search device, such as a dedicated terminal with a pulsar search function, a data server of a radio telescope, or a server cluster. Figure 7 The steps shown are explained.

[0148] Step 701: A pulsar search device receives an unlabeled pulsar search candidate signal.

[0149] Step 702: The pulsar search device obtains a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal.

[0150] Among them, reference Figure 8 , Figure 8 The schematic diagram of the pulsar search candidate signal in an embodiment of the present invention is shown in (1), wherein the time phase distribution image corresponding to the pulsar search candidate signal is shown in (2), and the frequency phase distribution image is shown in (1). When non-numeric values ​​(NaN Not a Number) appear in the time phase distribution image and the frequency phase distribution image, all NaN values ​​can be replaced, for example, by using linear interpolation to complete and replace the non-numeric values ​​in the time phase distribution image and the frequency phase distribution image. If data is missing, data filtering can also be performed to ensure the integrity of the non-numeric values ​​in the time phase distribution image and the frequency phase distribution image.

[0151] Step 703: The pulsar search device processes the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result.

[0152] refer to Figure 8Taking the new generation 500m aperture spherical radio telescope (FAST Five Hundred Meter Aperture Spherical Telescope) as an example, as the pulsar search proceeds, a large number of pulsar candidate signals without labeled information will be generated. However, there are inherent connections between different modal signals of pulsars, for example, Figure 8 The integral contour diagram of the pulsar shown in (3) is obtained by integrating the time-phase distribution diagram over time or integrating the frequency-phase distribution diagram over frequency. The inherent relationship between different signals in the pulsar search candidate signal can be determined by a trained pulsar search network. Among them, the backbone network of the pulsar search network can adopt a network structure such as resnet50, inceptionv4, resnet18, etc., which is not limited in this application.

[0153] Step 704: The pulsar search device processes the image processing result through the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal.

[0154] Among them, Figure 8 As shown in (4) of China, the image processing results are processed through the classification task fully connected layer network of the pulsar search network to obtain the image processing results of the time phase distribution image and the frequency phase distribution image, which can determine the discovery of new pulsars.

[0155] Beneficial technical effects:

[0156] The present invention obtains a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal; combines the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network; obtains a pulse profile image and a pulsar label value of the candidate corresponding to the obtained pulsar search candidate signal; performs a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate to determine the initial network parameters of the backbone network; performs a second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network, thereby achieving the training of the pulsar search network, and stably improving the accuracy of the pulsar search network training, reducing the overfitting of the pulsar search network, and enhancing the generalization ability of the pulsar search network model under the premise of reducing the total amount of training data and without the need for re-supervised training, so as to facilitate the deployment of the timely trained pulsar search network model in the network terminal and realize the large-scale application of the pulsar search network model.

[0157] The above descriptions are merely embodiments of the present invention and are not intended to limit the protection scope of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A pulsar search network training method, characterized in that: The method comprises: Obtaining a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal; Combining the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network; Acquire an original pulse profile image of a candidate object corresponding to the pulsar search candidate object signal; Standardizing the sequence values ​​of the original pulse profile image to obtain standard sequence values ​​of the original pulse profile image; Adjusting the standard sequence value of the original pulse profile image to obtain a pulse profile image of a candidate body corresponding to the pulsar search candidate body signal; Acquire pulsar label values, and perform first training on a backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate, and determine initial network parameters of the backbone network; Based on the training sample set and the pulsar label value, a second training is performed on the backbone network of the pulsar search network to obtain target parameters of the backbone network of the pulsar search network.

2. The method according to claim 1, characterized in that The combining of the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network includes: Determine the noise values ​​of the time phase distribution image and the frequency phase distribution image, and perform noise removal processing on the determined time phase distribution image and frequency phase distribution image respectively; The time phase distribution image and the frequency phase distribution image that have been subjected to the denoising process are respectively subjected to standardization processing to obtain a time phase distribution image and a frequency phase distribution image that meet the single-channel grayscale image standard; The time phase distribution image and the frequency phase distribution image that meet the single-channel grayscale image standard are combined to form a training sample set of the pulsar search network.

3. The method according to claim 2, characterized in that The step of determining the noise values ​​of the time phase distribution image and the frequency phase distribution image, and performing noise removal processing on the determined time phase distribution image and the frequency phase distribution image respectively, comprises: Determining a dynamic noise threshold that matches the use environment of the pulsar search network according to the position of the target area corresponding to the pulsar search candidate signal; The time phase distribution image and the frequency phase distribution image are subjected to noise reduction processing according to the dynamic noise threshold, so as to form a time phase distribution image and a frequency phase distribution image matching the dynamic noise threshold.

4. The method according to claim 2, characterized in that: The step of determining the noise values ​​of the time phase distribution image and the frequency phase distribution image, and performing noise removal processing on the determined time phase distribution image and the frequency phase distribution image respectively, comprises: Determining a fixed noise threshold that matches the use environment of the pulsar search network according to the type of radio telescope corresponding to the pulsar search candidate signal; The time phase distribution image and the frequency phase distribution image are subjected to noise reduction processing according to the fixed noise threshold, so as to form a time phase distribution image and a frequency phase distribution image matching the fixed noise threshold.

5. The method according to claim 1, characterized in that The first training of the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate to determine the initial network parameters of the backbone network includes: Determining a label value of the pulse profile image of the candidate body based on the pulse profile image of the candidate body; Determine, by means of a pre-trained fully connected layer network corresponding to the backbone network, a first loss function corresponding to the first training based on the label value of the pulse profile image; Substituting the training samples in the training sample set into the first loss function; When the first loss function satisfies a first convergence condition, initial network parameters of the backbone network are determined.

6. The method according to claim 1, characterized in that The second training of the backbone network of the pulsar search network based on the training sample set and the pulsar label value to adjust the initial parameters of the backbone network of the pulsar search network and obtain the target parameters of the backbone network of the pulsar search network includes: Determining a second loss function corresponding to the second training based on the pulsar label value through a classification task fully connected layer network corresponding to the backbone network; Substituting the training samples in the training sample set into the second loss function; When the second loss function satisfies a second convergence condition, determining an update parameter of the backbone network; Based on the updated parameters of the backbone network, the initial parameters of the backbone network of the pulsar search network are adjusted to obtain the target parameters of the backbone network of the pulsar search network.

7. A pulsar search method, characterized in that: The method comprises: Receive unlabeled pulsar search candidate signals; Acquire a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal; Processing the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result; Processing the image processing result through the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal; Wherein, the pulsar search network is trained by the method described in any one of claims 1 to 6.

8. A pulsar search network training device, characterized in that: The device comprises: An information transmission module is used to obtain a time phase distribution image and a frequency phase distribution image corresponding to a pulsar search candidate signal; An information processing module, used for combining the time phase distribution image and the frequency phase distribution image to form a training sample set of the pulsar search network; The information processing module is used to obtain the original pulse profile image of the candidate body corresponding to the pulsar search candidate body signal; standardize the sequence value of the original pulse profile image to obtain the standard sequence value of the original pulse profile image; adjust the standard sequence value of the original pulse profile image to obtain the pulse profile image of the candidate body corresponding to the pulsar search candidate body signal; and obtain the pulsar label value; The information processing module is used to perform a first training on the backbone network of the pulsar search network based on the training sample set and the pulse profile image of the candidate, and determine the initial network parameters of the backbone network; The information processing module is used to perform second training on the backbone network of the pulsar search network based on the training sample set and the pulsar label value to obtain target parameters of the backbone network of the pulsar search network.

9. A pulsar search device, characterized in that: The device comprises: A signal transmission module, used for receiving unlabeled pulsar search candidate signals; A signal processing module, used to obtain a time phase distribution image and a frequency phase distribution image corresponding to the pulsar search candidate signal; The signal processing module is used to process the time phase distribution image and the frequency phase distribution image through the backbone network of the pulsar search network to determine the image processing result; The signal processing module is used to process the image processing result through the classification task fully connected layer network of the pulsar search network to obtain the prediction result of the pulsar search candidate signal; Wherein, the pulsar search network is trained by the method described in any one of claims 1 to 6.

10. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable instructions; The processor is used to implement the pulsar search network training method described in any one of claims 1 to 6 when running the executable instructions stored in the memory, or to implement the pulsar search method described in claim 7.

11. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the pulsar search network training method described in any one of claims 1 to 6 is implemented, or the pulsar search method described in claim 7 is implemented.

12. A computer-readable storage medium storing executable instructions, characterized in that: When the executable instructions are executed by the processor, the pulsar search network training method described in any one of claims 1 to 6 is implemented, or the pulsar search method described in claim 7 is implemented.

Images