CN114217849A - Artificial intelligence model algorithm interpretation method and device and computer equipment - Google Patents

Artificial intelligence model algorithm interpretation method and device and computer equipment Download PDF

Info

Publication number
CN114217849A
CN114217849A CN202111313919.2A CN202111313919A CN114217849A CN 114217849 A CN114217849 A CN 114217849A CN 202111313919 A CN202111313919 A CN 202111313919A CN 114217849 A CN114217849 A CN 114217849A
Authority
CN
China
Prior art keywords
artificial intelligence
intelligence model
model algorithm
electromagnetic
diagram
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.)
Granted
Application number
CN202111313919.2A
Other languages
Chinese (zh)
Other versions
CN114217849B (en
Inventor
刘滨
邓克俭
黄创绵
夏姗姗
张鹏南
李乃鑫
李丹扬
赖岸
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Electronic Product Reliability and Environmental Testing Research Institute
Original Assignee
China Electronic Product Reliability and Environmental Testing Research Institute
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Electronic Product Reliability and Environmental Testing Research Institute filed Critical China Electronic Product Reliability and Environmental Testing Research Institute
Priority to CN202111313919.2A priority Critical patent/CN114217849B/en
Publication of CN114217849A publication Critical patent/CN114217849A/en
Application granted granted Critical
Publication of CN114217849B publication Critical patent/CN114217849B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F8/00—Arrangements for software engineering
    • G06F8/70—Software maintenance or management
    • G06F8/74—Reverse engineering; Extracting design information from source code
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R29/00—Arrangements for measuring or indicating electric quantities not covered by groups G01R19/00 - G01R27/00
    • G01R29/08—Measuring electromagnetic field characteristics
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R29/00—Arrangements for measuring or indicating electric quantities not covered by groups G01R19/00 - G01R27/00
    • G01R29/08—Measuring electromagnetic field characteristics
    • G01R29/0864—Measuring electromagnetic field characteristics characterised by constructional or functional features
    • G01R29/0871—Complete apparatus or systems; circuits, e.g. receivers or amplifiers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/3003—Monitoring arrangements specially adapted to the computing system or computing system component being monitored
    • G06F11/302—Monitoring arrangements specially adapted to the computing system or computing system component being monitored where the computing system component is a software system
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/30—Monitoring
    • G06F11/3058—Monitoring arrangements for monitoring environmental properties or parameters of the computing system or of the computing system component, e.g. monitoring of power, currents, temperature, humidity, position, vibrations

Abstract

The present application relates to a method, an apparatus, a computer device, a storage medium and a computer program product for interpreting an artificial intelligence model algorithm. The method comprises the following steps: the method comprises the steps of obtaining an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of a chip, obtaining an electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signal of the to-be-detected group and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic field evolution diagram and a Maxwell equation set, and obtaining an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip. Based on the obtained operation logic diagram of the artificial intelligence model algorithm, the operation logic of the artificial intelligence model algorithm can be clearly explained.

Description

Artificial intelligence model algorithm interpretation method and device and computer equipment
Technical Field
The present application relates to the field of artificial intelligence technology, and in particular, to an artificial intelligence model algorithm interpretation method, apparatus, computer device, storage medium, and computer program product.
Background
With the development of artificial intelligence technology, artificial intelligence model algorithms are becoming more and more complex, and the interpretability of artificial intelligence becomes an effective means for helping human beings to understand the working principle of artificial intelligence, and is the degree that the decision mechanism of an artificial intelligence system can be understood by human beings, and is one of the most difficult, most critical and most hot researches on artificial intelligence model algorithms.
At present, the interpretability of the artificial intelligence model algorithm is mainly researched by constructing an equivalent model or performing local interpretation from the perspective of the software model algorithm, however, the method for constructing the equivalent model or performing local interpretation has the problem of poor interpretability of the artificial intelligence model algorithm.
Disclosure of Invention
Based on this, it is necessary to provide an interpretation method, an interpretation apparatus, a computer device, a computer readable storage medium and a computer program product of an artificial intelligence model algorithm for clearly interpreting the operation logic of the artificial intelligence model algorithm in view of the technical problem that the conventional artificial intelligence model algorithm is poor in interpretability.
In a first aspect, the present application provides a method for interpreting an artificial intelligence model algorithm. The method comprises the following steps:
acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of the chip, wherein the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
obtaining an electromagnetic field evolution diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
obtaining a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm.
In one embodiment, obtaining the electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signal of the group to be measured and the electromagnetic signal of the comparison group comprises:
acquiring the electromagnetic field intensity which changes along with time and corresponds to the running of an artificial intelligence model algorithm according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group;
and obtaining an electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with the time and corresponds to the running of the artificial intelligence model algorithm.
In one embodiment, obtaining the electromagnetic field strength varying with time corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signal to be measured and the electromagnetic signal of the comparison group comprises:
obtaining the electromagnetic field intensity of the group to be detected along with the time change according to the electromagnetic signal of the group to be detected;
obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signals of the comparison group;
and subtracting the electromagnetic field intensity of the group to be measured changing along with the time from the electromagnetic field intensity of the comparison group changing along with the time to obtain the corresponding electromagnetic field intensity changing along with the time when the artificial intelligence model algorithm operates.
In one embodiment, obtaining a current signal change diagram corresponding to the operation of the artificial intelligence model algorithm through a maxwell equation system according to the electromagnetic field evolution diagram comprises:
obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram;
obtaining a current varying with time according to the electric field intensity varying with time;
and obtaining a corresponding current signal change diagram when the artificial intelligence model algorithm operates according to the current changing along with the time.
In one embodiment, the obtaining of the operational logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and the physical logic of the chip comprises:
obtaining the initial operation logic of an artificial intelligence model algorithm according to the current signal change diagram;
and obtaining an operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In one embodiment, the method for interpreting the artificial intelligence model algorithm further comprises:
analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm;
and obtaining an interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
In a second aspect, the present application further provides an apparatus for interpreting an artificial intelligence model algorithm. The device comprises:
the system comprises a signal acquisition module, a comparison module and a comparison module, wherein the signal acquisition module is used for acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of a chip, the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
the electromagnetic field evolution diagram acquisition module is used for acquiring a corresponding electromagnetic field evolution diagram when an artificial intelligence model algorithm operates according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
the current signal change diagram acquisition module is used for acquiring a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and the operation logic diagram acquisition module is used for acquiring an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:
the method comprises the steps of obtaining an electromagnetic signal of a group to be detected and an electromagnetic signal of a comparison group of a chip, wherein the electromagnetic signal of the group to be detected is an electromagnetic signal when the chip operates an artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, obtaining an electromagnetic field evolution diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic field evolution diagram through Maxwell equations, obtaining an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
the method comprises the steps of obtaining an electromagnetic signal of a group to be detected and an electromagnetic signal of a comparison group of a chip, wherein the electromagnetic signal of the group to be detected is an electromagnetic signal when the chip operates an artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, obtaining an electromagnetic field evolution diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic field evolution diagram through Maxwell equations, obtaining an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:
the method comprises the steps of obtaining an electromagnetic signal of a group to be detected and an electromagnetic signal of a comparison group of a chip, wherein the electromagnetic signal of the group to be detected is an electromagnetic signal when the chip operates an artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, obtaining an electromagnetic field evolution diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic field evolution diagram through Maxwell equations, obtaining an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
The interpretation method, the interpretation device, the computer equipment, the storage medium and the computer program product of the artificial intelligence model algorithm are characterized in that by acquiring the electromagnetic signals of the group to be detected and the electromagnetic signals of the contrast group of the chip, wherein the electromagnetic signal of the group to be detected is the electromagnetic signal when the chip operates the artificial intelligence model algorithm, the electromagnetic signal of the contrast group is the electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, obtaining an electromagnetic field evolution diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic field evolution diagram through a Maxwell equation set, and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm. According to the scheme, the electromagnetic signal of the chip is obtained, the corresponding electromagnetic field evolution diagram during the operation of the artificial intelligence model algorithm is obtained, the current signal variation diagram is obtained through the Maxwell equation set, the operation logic diagram of the artificial intelligence model algorithm is obtained by combining the physical logic of the chip, and the operation logic of the artificial intelligence model algorithm can be clearly explained.
Drawings
FIG. 1 is a diagram illustrating an example of an application environment of a method for interpreting an artificial intelligence model algorithm;
FIG. 2 is a schematic flow chart diagram illustrating a method for interpreting an artificial intelligence model algorithm, in accordance with an embodiment;
FIG. 3 is a flowchart illustrating steps of an artificial intelligence model algorithm in one embodiment;
FIG. 4 is a flowchart illustrating a method for interpreting an artificial intelligence model algorithm in another embodiment;
FIG. 5 is a block diagram showing an explanation apparatus of an artificial intelligence model algorithm in one embodiment;
FIG. 6 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The method for explaining the artificial intelligence model algorithm provided by the embodiment of the application can be applied to the application environment shown in fig. 1. Wherein terminal 102 communicates with chip 104 over a network. The data storage system may store data that the terminal 102 needs to process. The data storage system may be integrated on the terminal 102, or may be placed on the cloud or other network server. The terminal 102 obtains an electromagnetic signal of a to-be-tested group and an electromagnetic signal of a comparison group of the chip 104 by obtaining the electromagnetic signal of the to-be-tested group and the electromagnetic signal of the comparison group of the chip 104, wherein the electromagnetic signal of the to-be-tested group is an electromagnetic signal when the chip 104 runs the artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip 104 does not run the artificial intelligence model algorithm, a corresponding electromagnetic field evolution diagram during the running of the artificial intelligence model algorithm is obtained according to the electromagnetic field evolution diagram and a corresponding current signal variation diagram during the running of the artificial intelligence model algorithm is obtained through a Maxwell equation set, and a running logic diagram of the artificial intelligence model algorithm is obtained according to the current signal variation diagram and in combination with physical logic of the chip 104, and the running logic diagram is used for explaining the running logic of the artificial intelligence model algorithm. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices and portable wearable devices, and the internet of things devices may be smart speakers, smart televisions, smart air conditioners, smart car-mounted devices, and the like. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, and the like.
In one embodiment, as shown in fig. 2, there is provided an artificial intelligence model algorithm interpretation method, which is described by taking the method as an example applied to the terminal 102 in fig. 1, and includes the following steps:
s200: and acquiring the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group of the chip.
Artificial intelligence is a theory, method, technique and system that can simulate, extend and extend human intelligence. The development of artificial intelligence model algorithms such as neural network models is inspired by the progress of neuroscience, so that the research method of brain function mechanism can be referred to for the interpretability problem of the artificial intelligence model algorithms. In general, functional magnetic resonance imaging (fMRI), electroencephalogram (EEG), Magnetoencephalogram (MEG), and other instruments capable of recording blood flow changes during brain activities or brain electrical activity are used for studying brain function mechanisms, and when an experimental task is performed, the conditions of brain activities are recorded, and studies are performed by image presentation, and the like. For example, when studying the problem of "whether the face recognition process is affected by emotion", an experiment involving the task of recognizing faces of different expressions is designed. When a researcher asks to judge whether a face picture seen by the researcher belongs to a certain emotion, the researcher records the brain activity of the researcher by using an electroencephalogram, and finally, the functional area and the reaction speed of the brain, which are responsible for emotion judgment in face recognition, are determined by comparing the accuracy and time of the judgment of the researcher and the range of the brain activity. Referring to the human brain research method, the artificial intelligence model algorithm is equivalent to a virtual logic network of the human brain, a chip for carrying out operation calculation of the artificial intelligence model algorithm is equivalent to a physical entity of the human brain, and acquiring an electromagnetic signal of the chip is equivalent to recording the activity condition of the human brain.
Specifically, an electromagnetic signal of a group to be tested and an electromagnetic signal of a comparison group of the chip are obtained, wherein the electromagnetic signal of the group to be tested is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm, and the experimental conditions of the comparison group are completely the same as those of the group to be tested except that the artificial intelligence model algorithm is not run. Optionally, the chip in the scheme includes but is not limited to a common chip or an artificial intelligence chip, and in order to obtain the electromagnetic signal as accurately as possible, an electromagnetic shielding device may be disposed in a chip testing area to shield interference of the electromagnetic signal in the testing area.
S400: and obtaining an electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group.
The artificial intelligence model algorithm is divided into supervised learning, unsupervised learning, semi-supervised learning and reinforcement learning according to different model training modes, for example, the supervised learning model can comprise a Linear Classifier model (Linear Classifier), a Support Vector Machine (Support Vector Machine), a Naive Bayes model Classifier (Naive Bayes Classifier), a K-neighbor model (K-nearest neighbor), a Decision Tree model (Decision Tree), a Linear Regression model (Linear Regression), and a Regression Tree model (Regression Tree), and the unsupervised learning model mainly comprises a data clustering model (K-means), a data dimension reduction model (Primary Component Analysis), a changed magnetic field generates an electric field according to Maxwell equation electromagnetic field theory, the changed electric field generates a magnetic field, the electric field exists in the space around the electric charge, the electric field generates movement, and the movement electric charge generates a magnetic field, the current is a flow of electric charges, so that the current generates a magnetic field, a constant magnetic field is generated around the constant current, and an oscillating electric field is generated. And according to the obtained electromagnetic signal generated when the chip operates the artificial intelligence model algorithm and the electromagnetic signal generated when the chip does not operate the artificial intelligence model algorithm, obtaining a corresponding electromagnetic field evolution diagram when the artificial intelligence model algorithm operates, reflecting the operation rule of the artificial intelligence model algorithm, observing the activity degree of the change, and judging the use condition of the chip resource called by the artificial intelligence model algorithm.
Specifically, according to the electromagnetic signals of the group to be measured and the electromagnetic signals of the comparison group, the corresponding electromagnetic field evolution diagram during the operation of the artificial intelligence model algorithm is obtained.
S600: and obtaining a corresponding current signal change diagram during the operation of the artificial intelligence model algorithm according to the electromagnetic field evolution diagram and by a Maxwell equation set.
When both the electric field and the magnetic field change with time, the magnetic field excited by the changing electric field and the electric field excited by the changing magnetic field are collectively called a time-varying electromagnetic field, the time-varying electromagnetic field is excited to have a wave characteristic, and the time-varying electromagnetic field complies with the maxwell equation. The Maxwell equation set comprises a Gauss electric field law for describing static electricity, a Gauss magnetic field law for describing static electricity, a Faraday law for describing magnetism to generate electricity and an ampere-Maxwell law for describing electricity to generate magnetism.
Specifically, according to the electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm, the current signal variation diagram corresponding to the operation of the artificial intelligence model algorithm can be obtained through the Maxwell equation set.
S800: and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip.
The chip is a carrier of an integrated circuit, the integrated circuit is a large-scale and high-performance computing module consisting of a plurality of logic operation units, and the physical logic of the chip refers to each physical logic unit of the integrated circuit in the chip. According to different current signals, the chip shows 0 or 1 operation logic, and an operation logic diagram of the artificial intelligence model algorithm can be obtained by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
Specifically, an operating logic diagram of the artificial intelligence model algorithm is obtained according to the current signal change diagram and by combining the physical logic of the chip.
In the method for explaining the artificial intelligence model algorithm, the electromagnetic signal of the group to be measured and the electromagnetic signal of the comparison group of the chip are obtained, wherein the electromagnetic signal of the group to be measured is the electromagnetic signal when the chip operates the artificial intelligence model algorithm, the electromagnetic signal of the comparison group is the electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, the electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm is obtained according to the electromagnetic signal of the group to be measured and the electromagnetic signal of the comparison group, the current signal variation diagram corresponding to the operation of the artificial intelligence model algorithm is obtained according to the electromagnetic field evolution diagram through a Maxwell equation set, and the operation logic diagram of the artificial intelligence model algorithm is obtained according to the current signal variation diagram by combining the physical logic of the chip, so that the operation logic of the artificial intelligence model algorithm can be clearly explained.
In one embodiment, as shown in fig. 3, the step S400 of obtaining the electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be measured and the electromagnetic signals of the comparison group includes:
s420: and obtaining the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group.
S440: and obtaining an electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with the time and corresponds to the running of the artificial intelligence model algorithm.
In this embodiment, the electromagnetic signal is converted into a time-varying function of the electromagnetic field intensity according to the acquired electromagnetic signal of the chip to be tested and the acquired electromagnetic signal of the chip to be compared, and the time-varying electromagnetic field intensity that only reflects the operation of the artificial intelligence model algorithm is obtained according to the electromagnetic signal of the chip to be tested and the electromagnetic signal of the chip to be compared. And obtaining an electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with the time and corresponds to the running of the artificial intelligence model algorithm, so that the running rule of the artificial intelligence model algorithm can be reflected.
According to the scheme of the embodiment, the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm is obtained through the electromagnetic signals of the group to be tested and the electromagnetic signals of the comparison group, the electromagnetic field evolution diagram which corresponds to the running of the artificial intelligence model algorithm is obtained according to the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm, and the premise can be provided for supporting and explaining the running logic of the artificial intelligence model algorithm.
In another embodiment, obtaining the electromagnetic field strength varying with time corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signal to be measured and the electromagnetic signal of the comparison group comprises: obtaining the electromagnetic field intensity of the group to be detected along with the time change according to the electromagnetic signal of the group to be detected; obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signals of the comparison group; and subtracting the electromagnetic field intensity of the group to be measured changing along with the time from the electromagnetic field intensity of the comparison group changing along with the time to obtain the corresponding electromagnetic field intensity changing along with the time when the artificial intelligence model algorithm operates.
In the embodiment, the electromagnetic field intensity of the group to be detected, which changes along with time, is obtained according to the electromagnetic signal of the group to be detected; the method comprises the steps of obtaining the electromagnetic field intensity of a contrast group changing along with time according to the electromagnetic signals of the contrast group, specifically, converting the electromagnetic signals of a group to be detected into a change function of the electromagnetic field intensity along with time to obtain the electromagnetic field intensity of the group to be detected changing along with time, converting the electromagnetic signals of the contrast group into the change function of the electromagnetic field intensity along with time to obtain the electromagnetic field intensity of the contrast group changing along with time, subtracting the electromagnetic field intensity along with time when a chip runs an artificial intelligence model algorithm from the electromagnetic field intensity along with time when the chip does not run the artificial intelligence model algorithm, thereby eliminating the influence of the electromagnetic signals generated when the chip runs and obtaining the electromagnetic field intensity along with time corresponding to the time when the artificial intelligence model algorithm runs.
According to the scheme of the embodiment, the electromagnetic signal is converted into the time-varying function of the electromagnetic field intensity, the time-varying electromagnetic field intensity only reflecting the corresponding time variation of the artificial intelligence model algorithm in operation is obtained according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, and the premise can be provided for supporting and explaining the operation logic of the artificial intelligence model algorithm.
In another embodiment, as shown in fig. 4, the step S600 of obtaining a current signal variation graph corresponding to the operation of the artificial intelligence model algorithm by using maxwell equations according to the electromagnetic field evolution diagram includes:
s620: and obtaining the electric field intensity changing along with time through a Maxwell equation system according to the electromagnetic field evolution diagram.
S640: the time-varying current is obtained from the time-varying electric field strength.
S660: and obtaining a corresponding current signal change diagram when the artificial intelligence model algorithm operates according to the current changing along with the time.
In this embodiment, according to maxwell electromagnetic field theory, an induced electric field is excited by a changing magnetic field, a vortex field is induced when the electric field is induced, a corresponding electromagnetic field evolution diagram is obtained when an artificial intelligence model algorithm operates, an electric field strength changing with time is obtained through maxwell equations, positive charges move towards electric field lines and negative charges move towards opposite directions of the electric field according to the electric field strength changing with time, the directional movement of the charges forms a current, a current changing with time is obtained according to the electric field strength changing with time, and further, a current signal change diagram is obtained according to the current changing with time, and the current signal change diagram is a corresponding current signal change diagram when the artificial intelligence model algorithm operates.
According to the scheme of the embodiment, the electric field intensity changing along with time is obtained according to the electromagnetic field evolution diagram through the Maxwell equation set, the current changing along with time is obtained according to the electric field intensity changing along with time, the current signal change diagram corresponding to the operation of the artificial intelligence model algorithm is obtained according to the current changing along with time, and the premise can be provided for supporting and explaining the operation logic of the artificial intelligence model algorithm.
In another embodiment, the step S800 of obtaining the operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and the physical logic of the chip includes: obtaining the initial operation logic of an artificial intelligence model algorithm according to the current signal change diagram; and obtaining an operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In this embodiment, the initial operation logic of the artificial intelligence model algorithm is obtained according to the current signal change diagram corresponding to the operation of the artificial intelligence model algorithm, for example, the current signal is 0, which indicates that a certain PN junction in the chip is not conducted, the artificial intelligence model algorithm is not executed at this position, the current signal is 1, which indicates that the PN junction is conducted, and the artificial intelligence model algorithm is executed at this position. And obtaining an operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic of the artificial intelligence model algorithm and by combining the physical logic of the chip.
According to the scheme of the embodiment, the initial operation logic of the artificial intelligence model algorithm is obtained according to the current signal change diagram, the operation logic diagram of the artificial intelligence model algorithm is obtained according to the initial operation logic and by combining the physical logic of the chip, and the explanation of the operation logic of the artificial intelligence model algorithm can be supported.
In another embodiment, the interpretation of the artificial intelligence model algorithm further comprises: and analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm, and obtaining the interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
In this embodiment, the output result and the operation logic of the artificial intelligence model algorithm are analyzed according to the obtained operation logic diagram of the artificial intelligence model algorithm, and the learning inference decision process of the algorithm is evaluated to form the interpretable result of the artificial intelligence model algorithm.
According to the scheme of the embodiment, the operation logic of the artificial intelligence model algorithm is obtained by analyzing the operation logic diagram, the interpretable result of the artificial intelligence model algorithm is obtained according to the operation logic of the artificial intelligence model algorithm, and the effect of supporting and explaining the operation logic of the artificial intelligence model algorithm can be achieved.
To explain the explanation method and effect of the artificial intelligence model algorithm in the present solution in detail, the following description is made with a most detailed embodiment:
the electromagnetic signal of the chip is collected by adopting a square electromagnetic probe of 50 microns by 50 microns, the probe array is parallelly covered on the running chip, the height of the probe array from the surface of the chip is kept unchanged, an electromagnetic shielding device is arranged in a chip area corresponding to the electromagnetic probe, and a group to be tested and a comparison group are respectively arranged, wherein the electromagnetic signal when the test chip of the group to be tested runs the artificial intelligent model algorithm, the electromagnetic signal when the test chip of the comparison group does not run the artificial intelligent model algorithm, and other conditions are consistent with those of the group to be tested except that the comparison group does not run the artificial intelligent model algorithm. The method comprises the steps of obtaining an electromagnetic signal of a group to be detected and an electromagnetic signal of a comparison group of a chip, wherein the electromagnetic signal of the group to be detected is an electromagnetic signal when the chip operates an artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, obtaining an electromagnetic field evolution diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group, obtaining a current signal change diagram corresponding to the artificial intelligence model algorithm during operation according to the electromagnetic field evolution diagram through Maxwell equations, obtaining an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm. The electromagnetic field strength of the to-be-measured group changing along with time is obtained according to the electromagnetic signals of the to-be-measured group, the electromagnetic field strength of the comparison group changing along with time is obtained according to the electromagnetic signals of the comparison group, the electromagnetic field strength of the to-be-measured group changing along with time is subtracted from the electromagnetic field strength of the comparison group changing along with time, the electromagnetic field strength corresponding to the time when the artificial intelligence model algorithm operates is obtained, and the electromagnetic field evolution diagram corresponding to the time when the artificial intelligence model algorithm operates is obtained according to the electromagnetic field strength corresponding to the time when the artificial intelligence model algorithm operates. Obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram, obtaining the current changing along with time according to the electric field intensity changing along with time, obtaining a current signal change diagram corresponding to the operation of the artificial intelligence model algorithm according to the current changing along with time, obtaining the initial operation logic of the artificial intelligence model algorithm according to the current signal change diagram, and obtaining the operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of a chip. And analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm, and obtaining the interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
According to the scheme of the embodiment, the electromagnetic signal of the to-be-detected group and the electromagnetic signal of the comparison group of the chip are obtained, wherein the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs the artificial intelligence model algorithm, the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm, the electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm is obtained according to the electromagnetic signal of the to-be-detected group and the electromagnetic signal of the comparison group, the current signal variation diagram corresponding to the running of the artificial intelligence model algorithm is obtained according to the electromagnetic field evolution diagram through a Maxwell equation set, and the running logic diagram of the artificial intelligence model algorithm is obtained according to the current signal variation diagram by combining the physical logic of the chip, so that the running logic of the artificial intelligence model algorithm can be clearly explained.
It should be understood that, although the steps in the flowcharts related to the embodiments as described above are sequentially displayed as indicated by arrows, the steps are not necessarily performed sequentially as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.
Based on the same inventive concept, the embodiment of the application also provides an interpretation device of the artificial intelligence model algorithm, which is used for realizing the interpretation method of the artificial intelligence model algorithm. The solution scheme provided by the apparatus is similar to the solution scheme described in the above method, so the specific limitations in the following explanation apparatus embodiments of one or more artificial intelligence model algorithms may refer to the limitations in the above explanation method of artificial intelligence model algorithms, and are not described herein again.
In one embodiment, as shown in FIG. 5, there is provided an interpretation apparatus 500 of an artificial intelligence model algorithm, comprising: a signal acquisition module 520, an electromagnetic field evolution diagram acquisition module 540, a current signal variation diagram acquisition module 560 and an operation logic diagram acquisition module 580, wherein:
the signal acquisition module 520 is configured to acquire an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of the chip, where the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
the electromagnetic field evolution diagram obtaining module 540 is configured to obtain an electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be measured and the electromagnetic signals of the comparison group;
the current signal change diagram acquisition module 560 is used for acquiring a corresponding current signal change diagram during the operation of the artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and the operation logic diagram obtaining module 580 is configured to obtain an operation logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining physical logic of the chip, where the operation logic diagram is used to explain operation logic of the artificial intelligence model algorithm.
In the explaining device of the artificial intelligence model algorithm, the electromagnetic signal of the group to be tested and the electromagnetic signal of the comparison group of the chip are obtained, wherein the electromagnetic signal of the group to be tested is the electromagnetic signal when the chip operates the artificial intelligence model algorithm, the electromagnetic signal of the comparison group is the electromagnetic signal when the chip does not operate the artificial intelligence model algorithm, the electromagnetic field evolution diagram corresponding to the operation of the artificial intelligence model algorithm is obtained according to the electromagnetic signal of the group to be tested and the electromagnetic signal of the comparison group, the current signal variation diagram corresponding to the operation of the artificial intelligence model algorithm is obtained according to the electromagnetic field evolution diagram through a Maxwell equation set, and the operation logic diagram of the artificial intelligence model algorithm is obtained according to the current signal variation diagram by combining the physical logic of the chip, so that the operation logic of the artificial intelligence model algorithm can be clearly explained.
In one embodiment, the electromagnetic field evolution diagram obtaining module 540 is further configured to obtain, according to the electromagnetic signal of the group to be measured and the electromagnetic signal of the comparison group, a corresponding electromagnetic field intensity varying with time when the artificial intelligence model algorithm operates; and obtaining an electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with the time and corresponds to the running of the artificial intelligence model algorithm.
In one embodiment, the electromagnetic field evolution diagram obtaining module 540 is further configured to obtain an electromagnetic field intensity of the group to be measured, which varies with time, according to the electromagnetic signal of the group to be measured; obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signals of the comparison group; and subtracting the electromagnetic field intensity of the group to be measured changing along with the time from the electromagnetic field intensity of the comparison group changing along with the time to obtain the corresponding electromagnetic field intensity changing along with the time when the artificial intelligence model algorithm operates.
In one embodiment, the current signal variation graph obtaining module 560 is further configured to obtain the electric field strength varying with time through maxwell equations according to the electromagnetic field variation graph; obtaining a current varying with time according to the electric field intensity varying with time; and obtaining a corresponding current signal change diagram when the artificial intelligence model algorithm operates according to the current changing along with the time.
In one embodiment, the operation logic diagram obtaining module 580 is further configured to obtain an initial operation logic of the artificial intelligence model algorithm according to the current signal variation diagram; and obtaining an operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In one embodiment, the interpreting means 500 of the artificial intelligence model algorithm is further configured to analyze the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm; and obtaining an interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
The respective modules in the interpretation apparatus of the artificial intelligence model algorithm described above may be wholly or partially implemented by software, hardware, and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 6. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities.
The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used for storing an electromagnetic field evolution diagram, a current signal variation diagram and an operation logic diagram which correspond to the operation of the artificial intelligence model algorithm. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of interpreting an artificial intelligence model algorithm.
Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of the chip, wherein the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
obtaining an electromagnetic field evolution diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
obtaining a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
and obtaining the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group, and obtaining the electromagnetic field evolution diagram which corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
obtaining the electromagnetic field intensity of the to-be-measured group changing along with time according to the electromagnetic signal of the to-be-measured group, obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signal of the comparison group, subtracting the electromagnetic field intensity of the to-be-measured group changing along with time from the electromagnetic field intensity of the comparison group changing along with time, and obtaining the corresponding electromagnetic field intensity changing along with time when the artificial intelligence model algorithm operates.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram, obtaining the current changing along with time according to the electric field intensity changing along with time, and obtaining the corresponding current signal changing diagram when an artificial intelligence model algorithm operates according to the current changing along with time.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
and obtaining the initial operation logic of the artificial intelligence model algorithm according to the current signal change diagram, and obtaining the operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
and analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm, and obtaining the interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of the chip, wherein the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
obtaining an electromagnetic field evolution diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
obtaining a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and obtaining the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group, and obtaining the electromagnetic field evolution diagram which corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining the electromagnetic field intensity of the to-be-measured group changing along with time according to the electromagnetic signal of the to-be-measured group, obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signal of the comparison group, subtracting the electromagnetic field intensity of the to-be-measured group changing along with time from the electromagnetic field intensity of the comparison group changing along with time, and obtaining the corresponding electromagnetic field intensity changing along with time when the artificial intelligence model algorithm operates.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram, obtaining the current changing along with time according to the electric field intensity changing along with time, and obtaining the corresponding current signal changing diagram when an artificial intelligence model algorithm operates according to the current changing along with time.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and obtaining the initial operation logic of the artificial intelligence model algorithm according to the current signal change diagram, and obtaining the operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm, and obtaining the interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, performs the steps of:
acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of the chip, wherein the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
obtaining an electromagnetic field evolution diagram corresponding to the operation of an artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
obtaining a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and obtaining the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group, and obtaining the electromagnetic field evolution diagram which corresponds to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with time and corresponds to the running of the artificial intelligence model algorithm.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining the electromagnetic field intensity of the to-be-measured group changing along with time according to the electromagnetic signal of the to-be-measured group, obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signal of the comparison group, subtracting the electromagnetic field intensity of the to-be-measured group changing along with time from the electromagnetic field intensity of the comparison group changing along with time, and obtaining the corresponding electromagnetic field intensity changing along with time when the artificial intelligence model algorithm operates.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram, obtaining the current changing along with time according to the electric field intensity changing along with time, and obtaining the corresponding current signal changing diagram when an artificial intelligence model algorithm operates according to the current changing along with time.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and obtaining the initial operation logic of the artificial intelligence model algorithm according to the current signal change diagram, and obtaining the operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
In one embodiment, the computer program when executed by the processor further performs the steps of:
and analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm, and obtaining the interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
It should be noted that, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), Magnetic Random Access Memory (MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing based data processing logic devices, etc., without limitation.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims (10)

1. A method for interpreting an artificial intelligence model algorithm, the method comprising:
acquiring an electromagnetic signal of a group to be detected and an electromagnetic signal of a comparison group of a chip, wherein the electromagnetic signal of the group to be detected is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
obtaining an electromagnetic field evolution diagram corresponding to the running of an artificial intelligence model algorithm according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group;
obtaining a corresponding current signal change diagram when an artificial intelligence model algorithm operates through a Maxwell equation set according to the electromagnetic field evolution diagram;
and acquiring an operating logic diagram of the artificial intelligence model algorithm according to the current signal change diagram and by combining the physical logic of the chip, wherein the operating logic diagram is used for explaining the operating logic of the artificial intelligence model algorithm.
2. The method for interpreting an artificial intelligence model algorithm according to claim 1, wherein obtaining the electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic signals of the group to be measured and the electromagnetic signals of the comparison group comprises:
acquiring the electromagnetic field intensity which changes along with time and corresponds to the running of an artificial intelligence model algorithm according to the electromagnetic signal of the group to be detected and the electromagnetic signal of the comparison group;
and obtaining an electromagnetic field evolution diagram corresponding to the running of the artificial intelligence model algorithm according to the electromagnetic field intensity which changes along with the time and corresponds to the running of the artificial intelligence model algorithm.
3. The method for interpreting an artificial intelligence model algorithm according to claim 2, wherein obtaining a time-varying electromagnetic field strength corresponding to an artificial intelligence model algorithm in operation from the electromagnetic signals of the set to be measured and the electromagnetic signals of the comparison set comprises:
obtaining the electromagnetic field intensity of the group to be detected along with the time change according to the electromagnetic signal of the group to be detected;
obtaining the electromagnetic field intensity of the comparison group changing along with time according to the electromagnetic signals of the comparison group;
and subtracting the electromagnetic field intensity of the group to be measured changing along with the time from the electromagnetic field intensity of the comparison group changing along with the time to obtain the corresponding electromagnetic field intensity changing along with the time when the artificial intelligence model algorithm operates.
4. The method for interpreting an artificial intelligence model algorithm according to claim 1, wherein obtaining a current signal variation graph corresponding to an artificial intelligence model algorithm in operation through maxwell's equations according to the electromagnetic field evolution graph comprises:
obtaining the electric field intensity changing along with time through a Maxwell equation set according to the electromagnetic field evolution diagram;
obtaining a current varying with time according to the electric field intensity varying with time;
and obtaining a corresponding current signal change chart when an artificial intelligence model algorithm operates according to the current changing along with the time.
5. The method for interpreting an artificial intelligence model algorithm according to claim 4, wherein obtaining an operational logic diagram of the artificial intelligence model algorithm from the current signal variation diagram in combination with the physical logic of the chip comprises:
obtaining the initial operation logic of an artificial intelligence model algorithm according to the current signal change diagram;
and obtaining an operation logic diagram of the artificial intelligence model algorithm according to the initial operation logic and by combining the physical logic of the chip.
6. The method of interpreting an artificial intelligence model algorithm according to claim 1, further comprising:
analyzing the operation logic diagram to obtain the operation logic of the artificial intelligence model algorithm;
and obtaining an interpretable result of the artificial intelligence model algorithm according to the operation logic of the artificial intelligence model algorithm.
7. An apparatus for interpreting an artificial intelligence model algorithm, the apparatus comprising:
the system comprises a signal acquisition module, a comparison module and a comparison module, wherein the signal acquisition module is used for acquiring an electromagnetic signal of a to-be-detected group and an electromagnetic signal of a comparison group of a chip, the electromagnetic signal of the to-be-detected group is an electromagnetic signal when the chip runs an artificial intelligence model algorithm, and the electromagnetic signal of the comparison group is an electromagnetic signal when the chip does not run the artificial intelligence model algorithm;
the electromagnetic field evolution diagram acquisition module is used for acquiring a corresponding electromagnetic field evolution diagram when an artificial intelligence model algorithm runs according to the electromagnetic signals of the group to be detected and the electromagnetic signals of the comparison group;
the current signal change diagram acquisition module is used for acquiring a corresponding current signal change diagram during the operation of an artificial intelligence model algorithm through a Maxwell equation set according to the electromagnetic field evolution diagram;
and the operation logic diagram acquisition module is used for acquiring an operation logic diagram of the artificial intelligence model algorithm by combining the physical logic of the chip according to the current signal change diagram, and the operation logic diagram is used for explaining the operation logic of the artificial intelligence model algorithm.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 6 when executed by a processor.
CN202111313919.2A 2021-11-08 2021-11-08 Interpretation method and device for artificial intelligent model algorithm and computer equipment Active CN114217849B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111313919.2A CN114217849B (en) 2021-11-08 2021-11-08 Interpretation method and device for artificial intelligent model algorithm and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111313919.2A CN114217849B (en) 2021-11-08 2021-11-08 Interpretation method and device for artificial intelligent model algorithm and computer equipment

Publications (2)

Publication Number Publication Date
CN114217849A true CN114217849A (en) 2022-03-22
CN114217849B CN114217849B (en) 2023-11-07

Family

ID=80696637

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111313919.2A Active CN114217849B (en) 2021-11-08 2021-11-08 Interpretation method and device for artificial intelligent model algorithm and computer equipment

Country Status (1)

Country Link
CN (1) CN114217849B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050246114A1 (en) * 2004-04-29 2005-11-03 Rannow Randy K In-line field sensor
US20200272435A1 (en) * 2019-02-25 2020-08-27 Atos Syntel, Inc. Systems and methods for virtual programming by artificial intelligence
CN112347679A (en) * 2020-11-16 2021-02-09 北京环境特性研究所 Metamaterial simulation method and device with electromagnetic dispersion
CN112527281A (en) * 2020-12-17 2021-03-19 北京百度网讯科技有限公司 Operator upgrading method and device based on artificial intelligence, electronic equipment and medium

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050246114A1 (en) * 2004-04-29 2005-11-03 Rannow Randy K In-line field sensor
US20200272435A1 (en) * 2019-02-25 2020-08-27 Atos Syntel, Inc. Systems and methods for virtual programming by artificial intelligence
CN112347679A (en) * 2020-11-16 2021-02-09 北京环境特性研究所 Metamaterial simulation method and device with electromagnetic dispersion
CN112527281A (en) * 2020-12-17 2021-03-19 北京百度网讯科技有限公司 Operator upgrading method and device based on artificial intelligence, electronic equipment and medium

Also Published As

Publication number Publication date
CN114217849B (en) 2023-11-07

Similar Documents

Publication Publication Date Title
Rasmussen et al. Visualization of nonlinear kernel models in neuroimaging by sensitivity maps
Kumar et al. BrainIAK: the brain imaging analysis kit
Tian Functional data analysis in brain imaging studies
CN114052668B (en) Brain function analysis method based on magnetoencephalogram data
Peng et al. Self-weighted semi-supervised classification for joint EEG-based emotion recognition and affective activation patterns mining
Jozwik et al. Face dissimilarity judgments are predicted by representational distance in morphable and image-computable models
Kimm et al. Performance comparision of tpu, gpu, cpu on google colaboratory over distributed deep learning
Chen et al. DNNBrain: a unifying toolbox for mapping deep neural networks and brains
Ni et al. Neurally-inspired hyperdimensional classification for efficient and robust biosignal processing
Yue et al. Intention recognition from spatio-temporal representation of EEG signals
Nastase et al. Neural responses to naturalistic clips of behaving animals in two different task contexts
Sharaev et al. Causal interactions within the default mode network as revealed by low-frequency brain fluctuations and information transfer entropy
Xie et al. Wavelet kernel principal component analysis in noisy multiscale data classification
CN114217849B (en) Interpretation method and device for artificial intelligent model algorithm and computer equipment
US10945627B2 (en) Systems and methods for super-resolving electromagnetic localization and causality
Bahrami et al. Brain network analysis: A review on multivariate analytical methods
Bi et al. FIGCONs: exploiting FIne-Grained CONstructs of facial expressions for efficient and accurate estimation of in-vehicle drivers’ statistics
He et al. An extended application ‘Brain Q’processing EEG and MEG data of finger stimulation extended from ‘Zeffiro’based on machine learning and signal processing
Pearson et al. A biomimetic haptic sensor
Mohammad et al. Tri-model classifiers for EEG based mental task classification: hybrid optimization assisted framework
Hafsa et al. A genetic algorithm for image reconstruction in electrical impedance tomography for gesture recognition
Zak et al. Fourier Transform Layer: A proof of work in different training scenarios
Jiao et al. MMDF-ESI: Multi-Modal Deep Fusion of EEG and MEG for Brain Source Imaging
Batta et al. SVR-based Multimodal Active Subspace Analysis for the Brain using Neuroimaging Data
Parmar et al. Visualizing temporal brain-state changes for fMRI using t-distributed stochastic neighbor embedding

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