Continuous learning method and device based on memory consolidation, medium and equipment

Abstract

The invention discloses a continuous learning method and device based on memory consolidation, a medium and equipment. In the method, a sensory cortex module can sequentially learn training sample sets of different tasks in each round of training, and parameters, obtained through learning, of the sensory cortex module are transferred to a hippocampus module. And then the weight of the hippocampus module is loaded to the prefrontal cortex module, so that the weight of the hippocampus module is superposed on the prefrontal cortex module, and memory playback of the hippocampus is realized. And finally, a slow wave rhythm signal is input into the prefrontal cortex module, so that the slow wave rhythm signal can be adjusted in combination with the weight of the hippocampus module, the problem of disastrous forgetting of the neural network model in continuous learning can be effectively relieved, and the accuracy is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, medium and equipment for continuous learning based on memory consolidation. Background Art

[0002] With the development of machine learning technology, neural network models have made remarkable achievements in processing complex tasks. In order to improve the processing ability of neural network models for complex tasks, it is usually possible to enable the neural network model to learn new tasks or new data in a constantly changing environment, so that the neural network model can continuously improve its performance in a dynamic and unpredictable environment.

[0003] However, when learning multiple tasks continuously through a neural network model, there is usually a problem of "catastrophic forgetting", that is, each time the neural network model learns a new task, its processing performance of the old task will drop sharply. This "catastrophic forgetting" problem will cause the neural network model to perform poor accuracy in processing old tasks, which in turn limits the further development of artificial intelligence technology. Summary of the invention

[0004] This specification provides a continuous learning method, device, medium and equipment based on memory consolidation to partially solve the above-mentioned problems existing in the prior art.

[0005] This manual adopts the following technical solutions: The present specification provides a continuous learning method based on memory consolidation, the method is applied to a multi-task continuous learning system, the multi-task continuous learning system includes: a prefrontal cortex module, a hippocampus module, and a sensory cortex module, the prefrontal cortex module, the hippocampus module, and the sensory cortex module include various network layers, and for each network layer, the network layer includes various neuron nodes, the method includes: Acquire training sample sets for different tasks, wherein each training sample set includes: at least one of image data, text data, and audio data; According to each training sample set, multiple rounds of parameter adjustment are performed on the module parameters of the prefrontal cortex module to obtain a target prefrontal cortex module, and the task is executed through the target prefrontal cortex module; wherein, For each round of parameter adjustment, a target training sample set used in the current round of parameter adjustment is selected from each training sample set, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; The sensory cortex module is trained according to the target training sample set to obtain a trained sensory cortex module, the module parameters of the trained sensory cortex module are transferred to the hippocampus module, and the connection weights between the neuronal nodes contained in the hippocampus module are extracted as the target weights, the target weights are loaded into the prefrontal cortex module to be adjusted, and the preset slow wave rhythm signal is input into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuronal nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuronal nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

[0006] Optionally, there is a one-to-one correspondence between each network layer included in the hippocampus module and each network layer included in the sensory cortex module. For each network layer included in the hippocampus module, there is a one-to-one correspondence between each neuron node included in the network layer and each neuron node included in the network layer corresponding to the network layer in the sensory cortex module.

[0007] Optionally, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the slow wave stage of the slow wave cycle in which the neuron node is currently located, and if the membrane potential corresponding to the neuron node exceeds a preset pulse threshold, a pulse is issued to the neuron node of the next network layer that has a connection relationship with the neuron node; the slow wave stage of the current slow wave cycle includes: a rising stage and a falling stage; wherein, if the slow wave stage of the current slow wave cycle is the rising stage, the update amplitude of the membrane potential corresponding to the neuron node each time is greater than the update amplitude of the membrane potential corresponding to the neuron node when it is currently in the falling stage; the electrical signal includes: a slow wave rhythmic signal, and at least one of the pulses output by the neuron node of the previous network layer that has a connection relationship with the neuron node.

[0008] Optionally, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the slow wave stage of the slow wave cycle in which the neuron node is currently located, specifically including: For each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node based on the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the scaling factor corresponding to the slow wave stage of the slow wave cycle in which the neuron node is currently located.

[0009] Optionally, the prefrontal cortex module and the hippocampus module are composed of a spiking neural network model, and the sensory cortex module is composed of a spiking neural network model or an artificial neural network model.

[0010] This specification provides a task execution method, including: Acquire task data of the task to be executed, wherein the task data is at least one of image data, text data, and audio data; The task data is input into a pre-trained target model so that the task data is processed by the target model to execute the task to be executed and obtain an execution result. The target model is obtained based on the target prefrontal cortex module trained by the above-mentioned continuous learning method based on memory consolidation.

[0011] This specification provides a continuous learning device based on memory consolidation, including: An acquisition module, used to acquire each training sample set for different tasks, wherein each training sample set includes: at least one of image data, text data, and audio data; A training module is used to perform multiple rounds of parameter adjustment on the module parameters of the prefrontal cortex module according to the training sample sets to obtain a target prefrontal cortex module, and perform tasks through the target prefrontal cortex module; wherein, for each round of parameter adjustment, a target training sample set used in the current round of parameter adjustment is selected from the training sample sets, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; the sensory cortex module is trained according to the target training sample set to obtain a trained sensory cortex module, and the module parameters of the trained sensory cortex module are transferred to the sensory cortex module after the training. A hippocampus module is prepared, and the connection weights between the neuron nodes contained in the hippocampus module are extracted as the target weights, the target weights are loaded into the prefrontal cortex module to be adjusted, and the preset slow wave rhythm signal is input into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuron nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

[0012] This specification provides a task execution device, including: A task data acquisition module, used to acquire task data of the task to be executed, wherein the task data is at least one of image data, text data, and audio data; A task execution module is used to input the task data into a pre-trained target model so as to process the task data through the target model to execute the task to be executed and obtain an execution result. The target model is obtained based on the target prefrontal cortex module trained by the above-mentioned continuous learning method based on memory consolidation.

[0013] This specification provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned continuous learning and task execution method based on memory consolidation.

[0014] This specification provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the above-mentioned continuous learning and task execution method based on memory consolidation is implemented.

[0015] At least one of the above technical solutions adopted in this specification can achieve the following beneficial effects: In the continuous learning method based on memory consolidation provided in the present specification, each training sample set for different tasks is obtained, where each training sample set includes: at least one of image data, text data, and audio data. According to each training sample set, multiple rounds of parameter adjustment are performed on the module parameters of the prefrontal cortex module to obtain a target prefrontal cortex module, and the task is executed through the target prefrontal cortex module; wherein, for each round of parameter adjustment, the target training sample set used in the current round of parameter adjustment is selected from each training sample set, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; the sensory cortex module is trained according to the target training sample set , obtain the trained sensory cortex module, transfer the module parameters of the trained sensory cortex module to the hippocampus module, extract the connection weights between the neuron nodes contained in the hippocampus module as the target weight, load the target weight into the prefrontal cortex module to be adjusted, and input the preset slow wave rhythm signal into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuron nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, and obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

[0016] It can be seen from the above method that each time the server needs to use the prefrontal cortex module to learn new task content, the sensory cortex module can be trained through the training sample set corresponding to the new task, and the module parameters of the sensory cortex module after learning can be transferred to the hippocampus module, and then the weight of the hippocampus module can be extracted and loaded into the prefrontal cortex module, so that the prefrontal cortex module superimposes the weight of the hippocampus module, and then the slow wave rhythm signal can be input into the prefrontal cortex module so that each neuron node contained in the prefrontal cortex module emits pulses under the action of the slow wave rhythm signal, and then the neuron nodes used to represent the new task content and the old task content and the connection mode between the neuron nodes can be adjusted according to the order in which each neuron node contained in the prefrontal cortex module emits pulses under the action of the slow wave rhythm signal, so that the prefrontal cortex module can retain the content of the learned new task and the content of the old task, so as to alleviate the "catastrophic forgetting" problem in the continuous learning process of the neural network model, and then the accuracy of the neural network model in executing tasks can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification. In the drawings: Figure 1 A flowchart of a continuous learning method based on memory consolidation provided in this specification; Figure 2 A schematic diagram of the first round of parameter adjustment process provided in this specification; Figure 3 A schematic diagram of the second round of parameter adjustment process provided in this specification; Figure 4 A flowchart of a task execution method provided in this specification; Figure 5 A schematic diagram of a continuous learning device based on memory consolidation provided in this specification; Figure 6 A schematic diagram of a task execution device provided in this specification; Figure 7 A method corresponding to the Figure 1 Schematic diagram of electronic equipment. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0019] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0020] Figure 1 A flowchart of a continuous learning method based on memory consolidation provided in this specification includes the following steps: S101: Acquire training sample sets for different tasks, wherein each training sample set includes at least one of image data, text data, and audio data.

[0021] In the biological brain, memory is one of the foundations of intelligence, and the hippocampus in the biological brain is a key area for the brain to form long-term memory. During sleep, the biological brain transforms short-term learning content into long-term memory through memory consolidation mechanisms. During slow-wave sleep, coordinated activities between the cerebral cortex and the hippocampus support this process.

[0022] Specifically, when an organism is in a learning state, the hippocampus in the organism's brain first converts the received information into short-term memory, which is then processed and integrated in the hippocampus to gradually stabilize it, thereby avoiding interference from other information and rapid fading. Furthermore, when an organism is asleep, the hippocampus in the organism's brain can transmit the stored short-term memory to the cerebral cortex in the form of memory playback, so that the cerebral cortex can regulate the global situation by generating slow waves based on the memory played back by the hippocampus, thereby enabling the cerebral cortex to adjust the connection strength and structure between synapses in the form of long-term potentiation (LTP) and long-term depression (LTD) to achieve memory consolidation.

[0023] Based on this, the present specification provides a continuous learning method based on memory consolidation, by transferring the module parameters of the perceptual cortex module trained by different training sample sets to the hippocampus module, and simulating the activities of the hippocampus and cerebral cortex of the biological brain during sleep through the hippocampus module and the prefrontal cortex module, so as to adjust the connection pattern between each neuronal node in the prefrontal cortex module, thereby achieving a memory consolidation function with a bionic effect by combining the synergistic characteristics of the slow wave and its influence on the plasticity of the network synapses during the slow wave sleep simulated by the hippocampus module and the prefrontal cortex module, thereby alleviating the problem of "catastrophic forgetting" in the continuous learning process of the neural network model, so as to improve the accuracy of performing tasks through the neural network model.

[0024] In this specification, the execution entity used to implement the continuous learning method based on memory consolidation can be a designated device such as a server set up on the business platform. Of course, it can also be a terminal device such as a laptop computer or a desktop computer. For the sake of ease of description, this specification only takes the server as the execution entity as an example to illustrate a continuous learning method based on memory consolidation provided in this specification.

[0025] Among them, the above-mentioned consolidation-based continuous learning method can be applied to a multi-task continuous learning system, and the multi-task continuous learning system here includes: a prefrontal cortex module, a hippocampus module, and a sensory cortex module.

[0026] The above-mentioned prefrontal cortex module and hippocampus module include various network layers. For each network layer, the network layer includes various neuron nodes. The prefrontal cortex module and hippocampus module here can be composed of a spiking neural network model (Spiking Neural Network, SNN).

[0027] The sensory cortex module mentioned above includes various network layers. For each network layer, the network layer includes various neuron nodes. The sensory cortex module here can be composed of artificial neural network models (ANNs) or pulse neural network models.

[0028] Furthermore, the server may obtain various training sample sets required for continuous learning of the prefrontal cortex module.

[0029] In the above content, each training sample set corresponds to a different task, and the data contained in each training sample set may be at least one of image data, text data, and audio data. The above tasks may be determined according to actual conditions.

[0030] For example: the above-mentioned different tasks can be recognition tasks for different image data, such as: a recognition task for image data containing the number "1" and image data containing the number "2", a recognition task for image data containing the number "7" and image data containing the number "8", etc.

[0031] Of course, the above-mentioned different tasks can also refer to image classification tasks for data sets such as CIFAR and ImageNet.

[0032] S102: According to each training sample set, performing multiple rounds of parameter adjustment on the module parameters of the prefrontal cortex module to obtain a target prefrontal cortex module, and performing the task through the target prefrontal cortex module. Among them, for each round of parameter adjustment, a target training sample set used in this round of parameter adjustment is selected from the training sample sets, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in this round of parameter adjustment; the sensory cortex module is trained according to the target training sample set to obtain the trained sensory cortex module, the module parameters of the trained sensory cortex module are transferred to the hippocampus module, and the connection weights between the neuronal nodes contained in the hippocampus module are extracted as the target weights, and the target weights are loaded into the prefrontal cortex module to be adjusted, and the preset slow wave rhythm signal is input into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuronal nodes in the prefrontal cortex module to be adjusted according to the pulse emission order between the neuronal nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

[0033] In this specification, the server can perform multiple rounds of parameter adjustment on the prefrontal cortex module. The training sample sets used in each round of parameter adjustment are different. In this way, the server can obtain a prefrontal cortex module after continuous learning as a target prefrontal cortex module. The target prefrontal cortex module can not only process new tasks corresponding to new training sample sets in multiple rounds of parameter adjustment, but also process old tasks corresponding to old training sample sets at the same time.

[0034] Specifically, for each round of parameter adjustment in the above-mentioned multiple rounds of parameter adjustment, the server can determine the training sample set used in this round of parameter adjustment from the training sample sets of different tasks as the target training sample set, and use the prefrontal cortex module after the previous round of parameter adjustment as the prefrontal cortex module to be adjusted in this round of parameter adjustment.

[0035] Furthermore, the server may input the target training sample set into the sensory cortex module used in the current round of parameter adjustment, so as to train the sensory cortex module used in the current round of parameter adjustment using the target training sample set to obtain a trained sensory cortex module. The module parameters of the trained sensory cortex module may then be transferred to the hippocampus module used in the current round of parameter adjustment.

[0036] The target training sample set used in each round of parameter adjustment is different, and the sensory cortex module and hippocampus module used in each round of parameter adjustment are also different. It can be understood that a new training sample set is used in each round to train a new sensory cortex module, and the module parameters of the trained sensory cortex module are transferred to the new hippocampus module.

[0037] In actual application scenarios, the above-mentioned sensory cortex modules can be of two types. The following describes these two types of sensory cortex modules in detail.

[0038] The first type of sensory cortex module may be composed of an artificial neural network model. At this time, the method by which the server transfers the module parameters of the trained sensory cortex module to the hippocampus module may be: according to the module parameters of the trained sensory cortex module, determine the number of network layers included in the sensory cortex module, the number of neuron nodes included in each network layer in the sensory cortex module, the connection relationship between the neuron nodes in the sensory cortex module, and the connection weight between the neuron nodes in the sensory cortex module.

[0039] Furthermore, the server can adjust the module parameters of the hippocampus module according to the number of network layers included in the sensory cortex module, the number of neuron nodes included in each network layer in the sensory cortex module, the connection relationship between the neuron nodes in the sensory cortex module, and the connection weight between the neuron nodes in the sensory cortex module, so as to transfer the module parameters of the sensory cortex module after this round of training to the hippocampus module.

[0040] Specifically, the server can construct each neuron node in the pulse neural network model that constitutes the hippocampus module through a leaky integrate and fire (LIF) type neuron node. Each neuron node in the hippocampus module corresponds one-to-one to each computing node in the sensory cortex module. The relationship between the membrane potential of each neuron node and time can be expressed as:

[0041] in, is the initial membrane potential, For the moment The corresponding membrane potential, For the moment The current input corresponding to the moment, To preset time parameters, For preset resistance parameters, in this manual, , .

[0042] It can be understood that the network layers included in the hippocampus module in the above content correspond one-to-one to the network layers included in the sensory cortex module, and for each network layer included in the hippocampus module, each neuron node included in the network layer corresponds one-to-one to each neuron node in the network layer corresponding to the network layer in the sensory cortex module. For any two neuron nodes included in the hippocampus module, the connection relationship between the two neuron nodes is the same as the connection relationship between the two neuron nodes corresponding to the two neuron nodes in the sensory cortex module.

[0043] In other words, for each connection existing in the artificial neural network model constituting the sensory cortex module, there is also a corresponding connection in the spiking neural network model constituting the hippocampus module. The server can assign initial membrane potential values ​​and pulse thresholds to all neuron nodes in the spiking neural network model constituting the hippocampus module. Preferably, the server can assign initial membrane potential values ​​and pulse thresholds to all neuron nodes in the spiking neural network model constituting the hippocampus module. and pulse threshold Set to = -75 mV, = -55mV.

[0044] The second type of sensory cortex module may be composed of a spiking neural network model. At this time, the server may directly use the spiking neural network model that constitutes the trained sensory cortex module as the spiking neural network model that constitutes the hippocampus module, so as to transfer the module parameters of the sensory cortex module after this round of training to the hippocampus module.

[0045] It should be noted that the server can determine the pulse time series of each neuron node input into the input layer of the pulse neural network model constituting the hippocampus module according to each training sample data in the target training sample set used in this round of training through Poisson distribution.

[0046] The pulse value corresponding to each time unit in the above-mentioned pulse time sequence can be expressed as 1 or 0, that is, 1 represents the issuance of a pulse or 0 represents the non-issuance of a pulse, so that the above-mentioned pulse time sequence can be input into the input layer of the hippocampal module, so that the input layer of the hippocampal module releases pulse signals to the neuron nodes connected to the input layer contained in other network layers of the hippocampal module according to the above-mentioned pulse time sequence, so as to initialize the hippocampal module.

[0047] Furthermore, the server can extract the connection weights between the neuron nodes contained in the pulse neural network model constituting the hippocampus module as the target weights, and then load the target weights into the pulse neural network model constituting the prefrontal cortex module to load the connection weights of the hippocampus module into the prefrontal cortex module, and initialize the prefrontal cortex module using the above-mentioned pulse sequence.

[0048] It should be noted that in the first round of parameter adjustment, the prefrontal cortex module to be adjusted is also obtained by transferring the module parameters of the sensory cortex module, and the prefrontal cortex module to be adjusted used in subsequent rounds of parameter adjustment is the prefrontal cortex module after the previous round of parameter adjustment.

[0049] Furthermore, the server may input a preset slow wave rhythm signal into the prefrontal cortex module to be adjusted, so that each neuron node in the prefrontal cortex module to be adjusted emits pulses under the action of the slow wave rhythm signal.

[0050] Among them, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the slow wave stage of the slow wave cycle that the neuron node is currently in. And after updating the membrane potential, if the membrane potential corresponding to the neuron node exceeds the preset pulse threshold, a pulse is issued to the neuron node of the next network layer that has a connection relationship with the neuron node.

[0051] The electrical signal mentioned above includes at least one of a slow-wave rhythm signal and a pulse output by a neuron node in an upper network layer that is connected to the neuron node.

[0052] In the above content, the slow wave phases of the slow wave cycle in which the prefrontal cortex module is currently located include: rising phase and falling phase. Among them, the rising phase and the falling phase here are alternately rotated, such as: the prefrontal cortex module to be adjusted can be in the rising phase in the first unit time, and in the falling phase in the second unit time, and so on.

[0053] If the slow wave phase of the slow wave cycle currently in which the neuron node is located is a rising phase, the updating amplitude of the membrane potential corresponding to the neuron node each time is greater than the updating amplitude of the membrane potential corresponding to the neuron node when it is currently in a falling phase.

[0054] It should be noted that, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node can update the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the scaling factor corresponding to the slow wave stage of the slow wave cycle in which the neuron node is currently located. For details, refer to the following formula:

[0055] in, is the current input of the slow-wave rhythm signal integrated by the neuron node at time t, is the pulse time sequence corresponding to the slow wave rhythm signal (0 / 1), is the connection weight between each neuron node at time t, is the preset scaling factor for Scaling is performed so that the neuron node can receive all the information contained in the slow wave rhythm signal.

[0056] Specifically, for each neuron node in the prefrontal cortex module to be adjusted, the neuron node can update the membrane potential corresponding to the neuron node according to the current input integrated by the received slow wave rhythm signal. If the membrane potential corresponding to the neuron node exceeds the preset pulse threshold (-55 mV), a pulse is sent to the next layer of neuron nodes connected to the neuron node, and the initial membrane potential is reset to -75 mV. Otherwise, no pulse is sent, and the next input slow wave rhythm signal is received. The membrane potential corresponding to the neuron node is updated again according to the next received slow wave rhythm signal. For details, please refer to the following formula:

[0057] in, is the current input of the slow-wave rhythm signal integrated by each neuron node at time t, is the pulse time series corresponding to the slow wave rhythm signal, is the connection weight between each neuron node at time t, is the scaling factor, [sign(z)] + is a rectification function, when When it is greater than 0, the return value is 1, and in other cases, the return value is 0, thereby determining the slow wave stage of the slow wave cycle of the prefrontal cortex module to be adjusted at different unit times t. 1 / T represents the slow wave frequency corresponding to slow wave sleep, which determines the time length of a single slow wave cycle and can be set to 1 Hz in this manual. q represents the noise term, which obeys Uniformly distributed from -b to b, b is the noise parameter.

[0058] From the above content, it can be seen that in the rising stage, the server can gradually increase the membrane potential of each neuron node of the prefrontal cortex module to be adjusted until it reaches a threshold, at which time the neuron node begins to emit pulses. In this process, the excitability of the neuron nodes increases, so that more neuron nodes can be activated, so that the prefrontal cortex module to be adjusted can integrate information, which helps to strengthen the connection between the neuron nodes related to the current task in the pulse neural network. In the falling stage, the membrane potential of each neuron node decreases and enters a relatively inhibited state. In this process, the excitability of the neuron nodes decreases, and the pulse emission decreases or stops, so that the pulse neural network weakens the connection weights corresponding to the connections between those neuron nodes that are not frequently used, thereby reducing noise interference.

[0059] As can be seen from the above content, in this specification, the server can temporarily store the module parameters learned by the sensory cortex module in each round of training in each hippocampal module, that is, for each round of training, the service can transfer the module parameters of the sensory cortex module after that round of training to the hippocampal module, so that in each round of parameter adjustment, the information learned by the sensory cortex module in training can be short-term stored through the hippocampal module, and then the connection weights of the hippocampal module can be loaded through the prefrontal cortex module to be adjusted, and adjusted under the action of the slow wave rhythm signal, and then the prefrontal cortex module to be adjusted can be consolidated according to the information stored in the hippocampal module to reduce the influence of the information stored in the hippocampal module learned by the adjusted prefrontal cortex module on the coverage of the new information learned in the subsequent training process.

[0060] It should be noted that the method for the server to load the target connection weight of the hippocampus module into the prefrontal cortex module to be adjusted can be, for each target connection weight of the hippocampus module, determining the connection weight corresponding to the target connection weight from the prefrontal cortex module to be adjusted, and then the target connection weight and the connection weight corresponding to the target connection weight can be fused to obtain the fused connection weight, and the fused connection weight is updated to the prefrontal cortex module to be adjusted, thereby obtaining the prefrontal cortex module to be adjusted after the target connection weight of the hippocampus module is loaded.

[0061] Among them, the target connection weight and the connection weight corresponding to the target connection weight are the connection weights between the same two neuronal nodes. In other words, for the connection weights between every two neuronal nodes contained in the prefrontal cortex module, if it is determined that these two neuronal nodes are the same as the two neuronal nodes corresponding to the target connection weight, then the connection weight between the two neuronal nodes can be determined to be the connection weight corresponding to the target connection weight.

[0062] Furthermore, the server can adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission order between the neuron nodes connected in the prefrontal cortex module to be adjusted, and obtain the prefrontal cortex module after this round of adjustment. In the subsequent rounds of parameter adjustment, the prefrontal cortex module after this round of adjustment can be used as the prefrontal cortex module used in the next round of parameter adjustment.

[0063] In addition, the server can also convert the pulse neural network model constituting the prefrontal cortex module after this round of adjustment into an artificial neural network model to perform periodic testing on the converted artificial neural network model.

[0064] Among them, the above-mentioned testing method can be that for each round of parameter adjustment, the server can also obtain the test set corresponding to the target training set used in this round of parameter adjustment, and use the test set to test the artificial neural network model obtained by converting the prefrontal cortex module after this round of adjustment to obtain the test result of the prefrontal cortex module after this round of adjustment.

[0065] If the test results of the prefrontal cortex module after this round of adjustment determine that the prefrontal cortex module after this round of adjustment does not meet the preset conditions, the parameters of the prefrontal cortex module can be re-adjusted by the above method. In this process, when it is determined that the preset termination conditions are met, the prefrontal cortex module after this round of adjustment can be used as the target prefrontal cortex module, and when it is determined that the preset termination conditions are not met, the prefrontal cortex module after this round of adjustment can be used as the prefrontal cortex module used in the next round of parameter adjustment.

[0066] Among them, the above test results can be used to characterize the accuracy of the task execution results output by the above-mentioned artificial neural network model for each test sample included in the test set, and whether the degree of improvement in the accuracy of the task execution results output by the sensory cortex module for each training sample included in the target sample set used in this round of parameter adjustment is higher than the preset improvement threshold.

[0067] In the above content, the method for the server to adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission order between the connected neuron nodes in the prefrontal cortex module to be adjusted can be that for each neuron node in the prefrontal cortex module to be adjusted, if the neuron node in the next network layer that has a connection relationship with the neuron node emits a pulse after the neuron node, then the connection weight between the neuron node and the next neuron node is increased.

[0068] If a neuron node in the next network layer that is connected to the neuron node emits a pulse before the neuron node, the connection weight between the neuron node and the neuron node in the next network layer is reduced.

[0069] The weight adjustment value in the process of increasing or decreasing the connection weight between each neuron node can be based on the current connection weight and the difference between the corresponding pulse emission time of the neuron node and the corresponding pulse emission time of the neuron node in the next network layer ( ) is determined, and the specific formula can be referred to as follows:

[0070]

[0071]

[0072] In the above formula, is a neuron node with a connection relationship and neuron nodes The weight adjustment value of the connection weight between them, ε (w) represents the relationship that the degree of weight change also depends on the weight size itself. In this specification, =0.0103, =0.0051, =14 ms, =34 ms.

[0073] Furthermore, after each time the server monitors the pulse emission sequence corresponding to each neuron node, it can adjust the connection weights between the neuron nodes of the prefrontal cortex module to be adjusted based on the weight adjustment value determined by the above formula to obtain the adjusted connection weights, and determine the adjusted prefrontal cortex module based on the adjusted connection weights.

[0074] For ease of understanding, the following takes the above-mentioned training sample sets for different tasks as two training sample sets as an example to explain the above-mentioned continuous learning method based on memory consolidation in detail. Figure 2 , Figure 3 shown.

[0075] Figure 2 This is a schematic diagram of the first round of parameter adjustment process provided in this specification.

[0076] Combination Figure 2 It can be seen that the server can use the first training sample set of task 1 (i.e., Figure 2 The sensory cortex module is adjusted by adjusting parameters of the sensory cortex module based on the image data containing the number “0” and the image data containing the number “1” to obtain the sensory cortex module after the first round of parameter adjustment. Then, the module parameters of the sensory cortex module after the first round of parameter adjustment can be transferred to the hippocampus module and the prefrontal cortex module to be adjusted.

[0077] Among them, the hippocampus module is used to simulate the hippocampus in the biological brain, and the prefrontal cortex module is used to simulate the cortex in the biological brain. Figure 2 Each dot in the hippocampus module is used to represent each neuron node in the hippocampus module. Similarly, each dot in the prefrontal cortex module is used to represent each neuron node in the prefrontal cortex module to be adjusted.

[0078] Furthermore, the server can load the connection weights of the hippocampus module into the prefrontal cortex module to be adjusted, and input the preset slow wave rhythm signal into the prefrontal cortex module to be adjusted, so as to replay the interaction relationship between each neuron node to the prefrontal cortex module when the task corresponding to the first training sample set used in the first round of parameter adjustment is performed through the sensory cortex module.

[0079] Furthermore, each neuron node in the prefrontal cortex module emits pulses under the action of the slow-wave rhythmic signal, thereby consolidating the interaction relationship between each neuron node that plays a key role in obtaining the task results.

[0080] Combination Figure 2 It can be seen that after the slow wave rhythm signal is input into the prefrontal cortex module to be adjusted and passes through the slow wave sleep stage, the adjusted prefrontal cortex module is obtained. Figure 2 The bold black lines between the neuronal nodes in the adjusted prefrontal cortex module represent the strengthening of the interaction relationship between the neuronal nodes that play a key role in obtaining the task results.

[0081] Furthermore, the server can test the prefrontal cortex module that has undergone the first round of parameter adjustment through the test set corresponding to the first training sample set to obtain the test results, and determine, based on the test results, whether to use the prefrontal cortex module that has undergone the first round of parameter adjustment as the prefrontal cortex module in the second round of parameter adjustment.

[0082] Figure 3 This is a schematic diagram of the second round of parameter adjustment process provided in this specification.

[0083] Combination Figure 3 It can be seen that the server can select a new sensory cortex module for the second round of training, and in the second round of training, the second training sample set of task 2 (i.e., Figure 2 The sensory cortex module in the second round of parameter adjustment is trained with the image data containing the number "7" and the image data containing the number "8" in the second round of parameter adjustment to obtain the sensory cortex module after the second round of parameter adjustment, and then the module parameters of the sensory cortex module after the second round of parameter adjustment can be transferred to the hippocampus module used in the second round of parameter adjustment. Note that the hippocampus module in the second round is completely different from the hippocampus module in the first round, which corresponds to the fact that the hippocampus only retains short-term memory.

[0084] Furthermore, the server can load the connection weights of the hippocampus module used in the second round of parameter adjustment into the adjusted prefrontal cortex module obtained by the previous round of parameter adjustment (i.e., the prefrontal cortex module to be adjusted in the second round of parameter adjustment), and input the preset slow wave rhythm signal into the prefrontal cortex module to be adjusted in the second round of parameter adjustment, thereby strengthening the interaction relationship between the neuronal nodes in the prefrontal cortex module to be adjusted that play a key role in obtaining the task results of Task 2.

[0085] It should be noted that, combined with Figure 2 and Figure 3 It can be seen that the content of the task corresponding to the training sample set used in each round of parameter adjustment is encoded by different neurons and connection patterns. Then, according to the order in which the pulses of the neuron nodes contained in the prefrontal cortex module are emitted under the action of the slow wave rhythm signal, the neuron nodes used to represent the content of the new task and the content of the old task and the connection pattern between the neuron nodes can be adjusted, so that the prefrontal cortex module can retain the content of the learned new task and the content of the old task.

[0086] In this specification, when the server determines that all the training sample sets obtained for different tasks are used, it can be regarded as satisfying the above-mentioned termination condition, and then the adjusted prefrontal cortex module after the last round of parameter adjustment can be used as the target prefrontal cortex module, and the task is executed through the target prefrontal cortex module.

[0087] Among them, the above tasks can be determined according to actual application scenarios, such as: image recognition, natural language processing, information recommendation and other tasks.

[0088] For example, the above-mentioned training sample sets may be training sample sets containing different types of images, where training sample set 1 contains images of various bird animals, and training sample set 2 contains images of various cats, etc. Thus, in the first round of parameter adjustment, training sample set 1 can be input into the sensory cortex module used in the first round of parameter adjustment, and then the prefrontal cortex module after parameter adjustment can be obtained by the above method. In the second round of parameter adjustment, training sample set 2 can be input into the sensory cortex module used in the second round of parameter adjustment, and then the prefrontal cortex module after the second round of parameter adjustment can be obtained by the above method as the target prefrontal cortex module.

[0089] In practical applications, after receiving an image of a bird or a cat input by a user, the image of a bird or a cat input by the user may be identified through the target model.

[0090] In the above content, the target model can be a pulse neural network model constituting a target prefrontal cortex module, or it can be an artificial neural network model obtained by converting the pulse neural network model constituting the target prefrontal cortex module.

[0091] From the above content, it can be seen that each time the server needs to use the prefrontal cortex module to learn new task content, the sensory cortex module can be trained through the training sample set corresponding to the new task, and the module parameters of the sensory cortex module after learning can be transferred to the hippocampus module, and then the weight of the hippocampus module can be extracted and loaded into the prefrontal cortex module, so that the prefrontal cortex module superimposes the weight of the hippocampus module, and then the slow wave rhythm signal can be input into the prefrontal cortex module so that each neuron node contained in the prefrontal cortex module emits pulses under the action of the slow wave rhythm signal, and then the neuron nodes used to represent the new task content and the old task content and the connection mode between the neuron nodes can be adjusted according to the order in which each neuron node contained in the prefrontal cortex module emits pulses under the action of the slow wave rhythm signal, so that the prefrontal cortex module can retain the content of the learned new task and the content of the old task, so as to alleviate the "catastrophic forgetting" problem in the continuous learning process of the neural network model, and then the accuracy of executing tasks through the neural network model can be improved.

[0092] For ease of understanding, the following describes in detail the process of executing tasks on the target model trained by the above method. Figure 4 shown.

[0093] Figure 4 A flowchart of a task execution method provided in this specification includes the following steps: S401: Acquire task data of a task to be executed, wherein the task data is at least one of image data, text data, and audio data; S402: Input the task data into a pre-trained target model so as to process the task data through the target model to execute the task to be executed and obtain an execution result. The target model is obtained based on the target prefrontal cortex module trained by the above-mentioned continuous learning method based on memory consolidation.

[0094] The above is one or more implementation model deployment methods of this specification. Based on the same idea, this specification also provides a corresponding continuous learning device based on memory consolidation, such as Figure 5 , Figure 6 shown.

[0095] Figure 5 A schematic diagram of a continuous learning device based on memory consolidation provided in this specification includes: An acquisition module 501 is used to acquire training sample sets for different tasks, wherein each training sample set includes at least one of image data, text data, and audio data; The training module 502 is used to perform multiple rounds of parameter adjustment on the module parameters of the prefrontal cortex module according to the training sample sets to obtain a target prefrontal cortex module, and perform tasks through the target prefrontal cortex module; wherein, for each round of parameter adjustment, a target training sample set used in the current round of parameter adjustment is selected from the training sample sets, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; the sensory cortex module is trained according to the target training sample set to obtain a trained sensory cortex module, and the module parameters of the trained sensory cortex module are transferred to the sensory cortex module after the training. The hippocampus module is selected, and the connection weights between the neuron nodes contained in the hippocampus module are extracted as the target weights, the target weights are loaded into the prefrontal cortex module to be adjusted, and the preset slow wave rhythm signal is input into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuron nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

[0096] Optionally, there is a one-to-one correspondence between each network layer included in the hippocampus module and each network layer included in the sensory cortex module. For each network layer included in the hippocampus module, there is a one-to-one correspondence between each neuron node included in the network layer and each neuron node included in the network layer corresponding to the network layer in the sensory cortex module.

[0097] Optionally, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the slow wave stage of the slow wave cycle in which the neuron node is currently located, and if the membrane potential corresponding to the neuron node exceeds a preset pulse threshold, a pulse is issued to the neuron node of the next network layer that has a connection relationship with the neuron node; the slow wave stage of the current slow wave cycle includes: a rising stage and a falling stage; wherein, if the slow wave stage of the current slow wave cycle is the rising stage, the update amplitude of the membrane potential corresponding to the neuron node each time is greater than the update amplitude of the membrane potential corresponding to the neuron node when it is currently in the falling stage; the electrical signal includes: a slow wave rhythmic signal, and at least one of the pulses output by the neuron node of the previous network layer that has a connection relationship with the neuron node.

[0098] Optionally, for each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node based on the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the scaling factor corresponding to the slow wave stage of the slow wave cycle in which the neuron node is currently located.

[0099] Optionally, the prefrontal cortex module and the hippocampus module are composed of a spiking neural network model, and the sensory cortex module is composed of a spiking neural network model or an artificial neural network model.

[0100] Figure 6 A schematic diagram of a task execution device provided for this specification includes: The task data acquisition module 601 is used to acquire the task data of the task to be executed, wherein the task data is at least one of image data, text data and audio data; The task execution module 602 is used to input the task data into a pre-trained target model so as to process the task data through the target model to execute the task to be executed and obtain an execution result. The target model is obtained based on the target prefrontal cortex module trained by the above-mentioned continuous learning method based on memory consolidation.

[0101] This specification also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 A continuous learning and task execution method based on memory consolidation is provided.

[0102] This manual also provides Figure 7 The one shown corresponds to Figure 1 A schematic diagram of the electronic device. Figure 7 As mentioned above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The continuous learning and task execution method based on memory consolidation. Of course, in addition to the software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0103] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0104] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0105] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0106] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0107] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0109] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0110] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0111] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0112] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0113] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0114] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0115] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0116] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0117] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0118] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.

Claims

1. A continuous learning method based on memory consolidation, characterized in that: The method is applied to a multi-task continuous learning system, which includes: a prefrontal cortex module, a hippocampus module, and a sensory cortex module, wherein the prefrontal cortex module, the hippocampus module, and the sensory cortex module include network layers, and for each network layer, the network layer includes neuron nodes, and the method includes: Acquire training sample sets for different tasks, wherein each training sample set includes: at least one of image data, text data, and audio data; According to each training sample set, multiple rounds of parameter adjustment are performed on the module parameters of the prefrontal cortex module to obtain a target prefrontal cortex module, and the task is executed through the target prefrontal cortex module; wherein, For each round of parameter adjustment, a target training sample set used in the current round of parameter adjustment is selected from each training sample set, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; The sensory cortex module is trained according to the target training sample set to obtain a trained sensory cortex module, the module parameters of the trained sensory cortex module are transferred to the hippocampus module, and the connection weights between the neuronal nodes contained in the hippocampus module are extracted as the target weights, the target weights are loaded into the prefrontal cortex module to be adjusted, and the preset slow wave rhythm signal is input into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuronal nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuronal nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

2. The method according to claim 1, characterized in that There is a one-to-one correspondence between each network layer included in the hippocampus module and each network layer included in the sensory cortex module. For each network layer included in the hippocampus module, there is a one-to-one correspondence between each neuron node included in the network layer and each neuron node included in the network layer corresponding to the network layer in the sensory cortex module.

3. The method according to claim 1, characterized in that For each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node according to the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the slow wave stage of the slow wave cycle in which the neuron node is currently located, and if the membrane potential corresponding to the neuron node exceeds a preset pulse threshold, a pulse is issued to the neuron node of the next network layer that is connected to the neuron node; The slow wave stage of the current slow wave cycle includes: a rising stage and a falling stage; wherein, if the slow wave stage of the current slow wave cycle is the rising stage, the update amplitude of the membrane potential corresponding to the neuron node each time is greater than the update amplitude of the membrane potential corresponding to the neuron node when it is currently in the falling stage; the electrical signal includes: a slow wave rhythmic signal, and at least one of the pulses output by the neuron node of the upper network layer that is connected to the neuron node.

4. The method according to claim 3, characterized in that For each neuron node included in the prefrontal cortex module to be adjusted, the neuron node updates the membrane potential corresponding to the neuron node based on the received electrical signal, the connection weight between the neuron node and the neuron node of the previous network layer, and the scaling factor corresponding to the slow wave stage of the slow wave cycle in which the neuron node is currently located.

5. The method according to claim 1, characterized in that The prefrontal cortex module and the hippocampus module are composed of pulse neural network models, and the sensory cortex module is composed of a pulse neural network model or an artificial neural network model.

6. A task execution method, characterized in that: include: Acquire task data of the task to be executed, wherein the task data is at least one of image data, text data, and audio data; The task data is input into a pre-trained target model so that the task data is processed by the target model to execute the task to be executed and obtain an execution result. The target model is obtained based on a target prefrontal cortex module trained by the method described in any one of claims 1 to 5 above.

7. A continuous learning device based on memory consolidation, characterized in that: include: An acquisition module, used to acquire each training sample set for different tasks, wherein each training sample set includes: at least one of image data, text data, and audio data; A training module is used to perform multiple rounds of parameter adjustment on the module parameters of the prefrontal cortex module according to the training sample sets to obtain a target prefrontal cortex module, and perform tasks through the target prefrontal cortex module; wherein, for each round of parameter adjustment, a target training sample set used in the current round of parameter adjustment is selected from the training sample sets, and the prefrontal cortex module after the previous round of parameter adjustment is used as the prefrontal cortex module to be adjusted in the current round of parameter adjustment; the sensory cortex module is trained according to the target training sample set to obtain a trained sensory cortex module, and the module parameters of the trained sensory cortex module are transferred to the hippocampal module block, and extract the connection weights between the neuron nodes contained in the hippocampus module as the target weight, load the target weight into the prefrontal cortex module to be adjusted, and input the preset slow wave rhythm signal into the prefrontal cortex module to be adjusted, so as to adjust the connection weights between the neuron nodes in the prefrontal cortex module to be adjusted according to the pulse emission sequence between the neuron nodes when the prefrontal cortex module to be adjusted receives the slow wave rhythm signal, so as to obtain the prefrontal cortex module after this round of adjustment; when it is determined that the preset termination condition is met, the prefrontal cortex module after this round of adjustment is used as the target prefrontal cortex module.

8. A task execution device, characterized in that: include: A task data acquisition module, used to acquire task data of the task to be executed, wherein the task data is at least one of image data, text data, and audio data; A task execution module, used for inputting the task data into a pre-trained target model so as to process the task data through the target model to execute the task to be executed and obtain an execution result, wherein the target model is obtained based on a target prefrontal cortex module trained by the method described in any one of claims 1 to 5 above.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of claims 1 to 6 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method described in any one of claims 1 to 6 is implemented.

Images