A blockchain-based distributed integrated learning method

Abstract

The application discloses a kind of distributed integrated learning methods based on blockchain, based on the three-layer blockchain structure designed by mini-block, integrated block and key block, the miner in network is trained on the training set after with replacement sampling by consensus protocol, aggregate model from other miners, finally the information of base model and integrated model is recorded in blockchain, so that the whole process of model training, model integration, model evaluation is integrated into blockchain consensus, so that the whole integrated learning process is automatically executed in blockchain network.Therefore, the application can improve the utilization rate of computing power in the blockchain network based on the useful work of machine learning proof mechanism, while avoiding the introduction of central node in the model aggregation process, maximizing the degree of decentralization of blockchain.

Description

Technical Field

[0001] This invention relates to the field of blockchain technology, and in particular to a distributed ensemble learning method based on blockchain. Background Technology

[0002] Blockchain is a distributed data storage technology that ensures the consistency of data replicas across all nodes in an unreliable peer-to-peer network through consensus protocols and incentive mechanisms. The consensus protocol is the cornerstone of blockchain, and its design fundamentally determines the performance characteristics of the blockchain system, including throughput, consistency, scalability, and robustness. Currently, Proof-of-Work (PoW) is one of the most popular consensus protocols. In PoW-based systems, miners generate new blocks by finding a random number that makes the block's hash value lower than a target value. However, the mining process consumes a large amount of energy, and most of this energy is wasted on meaningless hash calculations.

[0003] Currently, solutions to the sustainability problem of PoW protocols fall into two categories: one is reducing the computational load of the consensus process, and the other is the Useful Proof-of-Work (PoUW) protocol. The latter uses computationally useful tasks as proofs of computation, thereby satisfying the computing power needs of third parties. Existing PoUW protocols primarily employ machine learning and optimization tasks as useful proofs of work. Cutting-edge research also combines federated learning with blockchain consensus, transforming mining pools in PoW systems into mining clusters competing for machine learning training rewards, and aggregating the models from various pools into a global model using the FedAvg algorithm.

[0004] In the aforementioned Proof-of-Work (PoUW) protocols employing machine learning and federated learning, most machine learning-based PoUW schemes select a single model as the winning model and discard the rest, inevitably leading to low computational power utilization. As for existing consensus mechanisms based on federated learning, although the models are trained in a distributed manner by data holders, the robustness of these protocols is constrained by the central nodes because the models are typically aggregated by one or more central nodes. Therefore, a technology is needed to aggregate the computational power of nodes in a decentralized blockchain network. Summary of the Invention

[0005] Purpose of the invention: To address the problems of low computing power utilization and insufficient decentralization in existing PoUW protocols, the purpose of this invention is to design a three-layer blockchain structure consisting of mini-blocks, integrated blocks, and key blocks, as well as a distributed ensemble learning method based on blockchain, to improve the computing power utilization of the blockchain network, while avoiding the introduction of a central node in the model aggregation process, thereby maximizing the decentralization of the blockchain.

[0006] Technical solution: To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0007] The first aspect of this invention provides a three-layer blockchain structure that enables the execution of distributed ensemble learning tasks, comprising:

[0008] MiniBlock: Each miniblock corresponds to a unique base model and contains identifiers for machine learning model parameters, identifiers for the model owner, hash of the previous key block, hash of the machine learning task, and timestamp.

[0009] An ensemble block contains the performance metrics of the ensemble model on the validation dataset, the identifier of the model integrator, the hash of the mini-block corresponding to the base model being ensembled, the hash of the machine learning task, and the timestamp.

[0010] A key block contains the hash of the previous key block, the hash of the machine learning task, the block-producing miner identifier, the performance metrics of the best ensemble model on the test dataset, the timestamp of the key block's generation, the root hash of the hash tree carrying the transaction data, a random number, a task queue, the hashes of each ensemble block participating in the ranking, and their performance metrics.

[0011] A second aspect of this invention provides a method for generating a three-layer blockchain structure that enables distributed ensemble learning task execution, comprising the following steps:

[0012] After miners complete the training of the base model on the training dataset, they generate and broadcast mini-blocks; each mini-block corresponds to a unique base model, which includes identifiers of machine learning model parameters, identifiers of the model owner, hashes of the previous key block, hashes of machine learning tasks, and timestamps.

[0013] After the miner aggregates the base model and performs performance evaluation on the validation dataset, it generates and broadcasts ensemble blocks; each ensemble block contains the performance metrics of the ensemble model on the validation dataset, the identifier of the model integrator, the hash of the mini-block corresponding to the ensembled base model, the hash of the machine learning task, and a timestamp;

[0014] After evaluating the performance metrics of the collected ensemble models on the test dataset, miners select the optimal ensemble model, generate and broadcast key blocks; the key block includes the hash of the previous key block, the hash of the machine learning task, the block-producing miner identifier, the performance metrics of the optimal ensemble model on the test dataset, the timestamp of the key block generation, the root hash of the hash tree carrying transaction data, a random number, a task queue, the hashes of each ensemble block participating in the ranking, and their performance metrics.

[0015] A third aspect of this invention provides a blockchain-based distributed ensemble learning method, comprising the following steps:

[0016] Step 1: The task publisher prepares training, validation, and test datasets in advance and publishes the machine learning task through blockchain transactions. The task includes the hash values ​​of these three datasets and scripts for training the base model, integrating the models, and evaluating the performance.

[0017] Step 2: The miner verifies the task information and forwards it to other miners.

[0018] Step 3: Once a miner produces the previous key block or receives a valid previous key block from another miner, it downloads the training dataset from the task publisher. After downloading, the miner begins training the base model, producing mini-blocks and broadcasting them.

[0019] Step 4: The task publisher publishes the verification dataset. After receiving the message, the miners start aggregating the base model and evaluate its performance on the verification dataset, generate an integrated block and broadcast it.

[0020] Step 5: The task publisher publishes the test dataset. After receiving the message, the miners begin to evaluate the performance metrics of the collected ensemble models on the test dataset, select the optimal ensemble model, calculate the key block hash, generate the key block, and broadcast it.

[0021] Step 6: The task publisher obtains the performance indicators of the best ensemble model and the corresponding base model parameters from the key blocks, and obtains an ensemble model with ideal performance by aggregating the base models.

[0022] Preferably, in one embodiment of the present invention, the machine learning task published by the user through a blockchain transaction includes a transaction used as training fees, which transfers the training fees to a virtual address.

[0023] Preferably, in one embodiment of the invention, miners who generate key blocks and mini-blocks receive rewards, the training cost of the task issuer is distributed equally among all miners who generate the base model used by the winning ensemble model, and miners who generate key blocks are rewarded with newly generated tokens.

[0024] Alternatively, in one embodiment of the present invention, miners can use their local private data together with the public training data provided by the task publisher for base model training, thereby improving the quality of the base model and the quality of the ensemble model.

[0025] Preferably, in one embodiment of the present invention, the ensemble model is obtained by aggregating the base model using the Bagging algorithm.

[0026] A fourth aspect of the present invention provides a computer system including a memory, a processor, and a computer program / instructions stored in the memory and executable on the processor, wherein the computer program / instructions, when executed by the processor, implement the steps of the aforementioned method.

[0027] A fifth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method.

[0028] A sixth aspect of the present invention provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the aforementioned method.

[0029] The blockchain-based distributed ensemble learning method provided by this invention has the following beneficial effects:

[0030] (1) This invention designs a three-layer blockchain structure and embeds the distributed ensemble learning process into the generation, verification and propagation process of three types of blocks, thereby realizing the execution of distributed ensemble learning tasks in the blockchain.

[0031] (2) Compared with the existing machine learning-based PoUW protocol, this invention realizes the automated and distributed execution of the ensemble learning process in the blockchain network, and forms an ensemble model with better performance by aggregating base models trained by multiple miners; and further improves the quality of base models and ensemble models by allowing miners to use private data.

[0032] (3) Compared with the federated learning proof mechanism, this invention does not require the aggregation base model by the mining pool administrator, can operate on the public chain, and has a higher degree of decentralization.

[0033] (4) This invention encourages miners to participate in maintaining the blockchain, training the model, integrating the model and evaluating the model by designing an incentive mechanism, thereby enabling the blockchain based on this invention to operate robustly and sustainably.

[0034] (5) The performance of the integrated model generated by the present invention increases with the increase in the number of miners using private data to train the base model in the blockchain.

[0035] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0036] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0037] Figure 1 A schematic diagram of a three-layer block structure for enabling distributed ensemble learning task execution according to an embodiment of the present invention;

[0038] Figure 2 This is a flowchart illustrating the overall process of a blockchain-based distributed ensemble learning method provided according to an embodiment of the present invention.

[0039] Figure 3 This is a schematic diagram of each stage of the task execution process in the blockchain-based distributed ensemble learning method provided according to an embodiment of the present invention;

[0040] Figure 4 This diagram illustrates the performance improvement of the ensemble model generated by the embodiments of the present invention relative to the base model and the model trained directly on public data under different proportions of public data.

[0041] Figure 5 This diagram illustrates the accuracy of the integrated model generated by the embodiments of the present invention under different network sizes and connectivity conditions. Detailed Implementation

[0042] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art should understand that the embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0043] The following describes an embodiment of the blockchain-based distributed ensemble learning method according to the accompanying drawings. Addressing the problems of low computing power utilization and insufficient decentralization in existing PoUW protocols mentioned in the background section, this invention provides a blockchain-based distributed ensemble learning method that uses the Bagging algorithm to fuse models trained by multiple miners. The disclosed three-layer blockchain structure and distributed ensemble learning method can automatically execute model training, model ensemble, and model evaluation processes in a fully decentralized public blockchain.

[0044] Specifically, Figure 1 This invention provides a three-layer blockchain structure that enables the execution of distributed ensemble learning tasks, according to an embodiment of the present invention.

[0045] like Figure 1 As shown, the three-layer blockchain structure includes three data structures and the connections between them:

[0046] Mini blocks (represented as) This is used to record model ownership and hashes of model parameters. Each mini-block corresponds to a unique base model and contains the model owner's identifier. The identifier is calculated by concatenating the machine learning model parameters with the miner identifier. Hash of the previous critical block Machine learning task hashing and timestamp .

[0047] Integrated circuit (represented as) This section records the base models required for each ensemble model and the performance metrics of the ensemble model on the validation dataset. An ensemble block points to multiple mini-blocks; aggregating the base models corresponding to the mini-blocks yields an ensemble model. The ensemble block contains the performance metrics of that ensemble model on the validation dataset. On performance indicators Identifiers of model integrators Hash of the aggregated base model Machine learning task hashing and timestamp .

[0048] Key block (represented as) This is used to store blockchain transactions, task queues, and ensemble model rankings, and includes the hash of the previous key block. Machine learning task hashing Block miner identifier The optimal ensemble model on the test dataset On performance indicators timestamps of key block generation The root hash of the hash tree that carries transaction data Nonce, Task Queue, and hashes of each ensemble block participating in the ranking. as well as The corresponding ensemble model on the test dataset On performance indicators .

[0049] This invention provides a method for generating a three-layer blockchain structure that enables distributed ensemble learning tasks. The main steps include: miners training a base model on a training dataset, generating and broadcasting the generated blockchain structure. Figure 1 Mini-blocks in the structure shown; miner aggregation base model, and after performance evaluation on the validation dataset, generated and broadcast. Figure 1 The structure shown includes ensemble blocks; after miners evaluate the performance metrics of the collected ensemble models on the test dataset, they select the optimal ensemble model, generate and broadcast it. Figure 1 The key blocks in the structure shown.

[0050] like Figure 2 and Figure 3 As shown, Figure 2 This invention provides a blockchain-based distributed ensemble learning method. Figure 3 The diagram illustrates the various stages of the task execution process in a blockchain-based distributed ensemble learning method.

[0051] Figure 2In step ①, “Task Release”, the task releaser prepares training, validation and test datasets in advance, and releases the machine learning task through blockchain transactions. The task includes the hash values ​​of the three datasets and the scripts for training the base model, integrating the model and evaluating the performance, as well as a transaction to pay the training fee, which transfers the training fee to a virtual address.

[0052] Figure 2 In step ②, “Task Verification and On-Chain,” miners verify the task information and forward it to other miners. The task will eventually enter the task queue of the key block.

[0053] Figure 2 Steps ③ through ⑤ constitute the task execution process. Figure 3 The first stage (Phase 1, Base Model Training) corresponds to Figure 2 Step ③ "Base model training" Figure 3 Phase 2 (Ensemble Block Generation) corresponds to Figure 2 Step 4, "Generate Integration Block", Figure 3 The third phase (Phase 3, Key Model Generation) corresponds to Figure 2 Step 5: "Generate key blocks".

[0054] In this embodiment of the invention, the task execution process includes the following three stages:

[0055] When the block height is key block Once generated by a miner, the first phase of the task execution process will be triggered. When the miner... Received Afterwards, if Through verification, miners The base model will begin training. If the miner... Upon receiving one or more valid forks, select the longest fork (i.e., the one with the largest terminal critical block height). If the terminal critical blocks of these forks have the same height, select the one with the optimal height. The fork containing the key block becomes the main chain, and tasks at the current height are executed on that main chain. Once the miners... confirm It will choose to be The first task in the task queue is to download the public training dataset. ,Will With private datasets Merge into local dataset Then, the miners In local dataset Execute the training script and generate the parameters of the base model. Once the base model is ready, miners timestamp Miner's identifier Task hash Model parameter identifiers and Encapsulated in In, and broadcast to the blockchain network Before the second phase, all miners receive mini-blocks and broadcast them to neighboring miners, but do not transmit the base model externally to avoid model theft.

[0056] Task publishers can Public validation dataset This triggers the second phase of the task execution process. Once the honest miner... Upon receiving the validation dataset, it will refuse to accept new mini-blocks and perform the following steps: (1) Download the validation dataset And according to from Identifiers extracted Get model parameters (2) miners Verify all collected mini-blocks; if a mini-block points to an invalid key block, or the downloaded model parameters... with identifier If there is a mismatch, or if the performance metrics of the base model corresponding to the mini-block on the validation dataset are lower than the minimum tolerance value given by the task publisher, then the mini-block is considered invalid, and the miner... All invalid mini-blocks will be discarded; (3) miners Use the validation dataset The performance metrics are obtained by evaluating the ensemble model, which is an aggregation of the base models corresponding to all valid mini-blocks. (4) In Encapsulated timestamp miner identifier Task hash Performance indicators obtained from the evaluation The hash of the mini-blocks corresponding to all aggregated base models And broadcast it to the blockchain network. Before the start of the third phase, if the miners... Receive any integrated blocks from other miners ,miner verify It then broadcasts the integrated block to nearby miners.

[0057] Task publishers can Public test dataset This triggers the third phase of the task execution process. (Miner) Received test dataset Then, the test dataset will be used. Evaluate all verified integrated blocks, and then attempt to generate a new key block. (Assume the miner...) It has been verified. ,and point to ,and Include ,in yes The number of mini-blocks it points to, It is not greater than positive integers, miners Generate. Then the miners calculate Performance metrics and from a series Find the optimal performance index Assuming That is the optimal performance indicator. Then the miners... Load all components of the key blocks into the candidate blocks. In the middle. By changing the random number and calculating the hash value of the candidate block, the miner will eventually find one that satisfies... Effective key blocks and upon receiving valid key blocks from other miners Former Broadcaster If a valid key block is found and broadcast Received valid key blocks from other miners Then the miners abandoned and enter block height Machine learning tasks The first stage of the execution process. Here... It is a static threshold that controls the difficulty of generating key blocks, equivalent to the target value in PoW consensus.

[0058] Figure 2 In step 6, "Task complete," the task issuer can use the block height... Key blocks on After downloading the base model parameters from the blockchain network and aggregating the base model, the task publisher will obtain an integrated model with ideal performance.

[0059] refer to Figure 4 and Figure 5 To demonstrate the performance of the present invention in practical work, actual tests and data recordings were conducted on typical embodiments of the present invention, and the analysis results are as follows.

[0060] Figure 4 This reflects the performance improvement of the ensemble model (BagChain) generated by this invention relative to the base model (Base) and the model trained solely on public data (Public) under different proportions of public data. It can be seen that the ensemble model outperforms any one of its base models in terms of accuracy on the test dataset. This is especially true when the public dataset ratio is [not specified]. When the value is less than 0.3, the model trained on the public dataset performs worse than the base model trained using both private and public data. This demonstrates the value of this invention because task publishers can obtain high-quality machine learning models from it when all miners have private datasets available for machine learning tasks. It can also be observed that as the proportion of public datasets increases... The accuracy of both the base model and the ensemble model in this invention has improved due to the increase in the proportion of public datasets. The increase in the number of data samples in the miners' local training dataset leads to more accurate base and ensemble models. Furthermore, the difference between the ensemble models obtained after training the base models for 20 and 30 epochs is negligible, indicating that 20 epochs are sufficient to generate a sufficiently good base model. The aggregated model exhibits ideal performance, demonstrating that training the base model in this invention is lightweight. Therefore, more resource-constrained nodes can participate in model training without fine-tuning the machine learning model; their models can be aggregated to improve the performance of the ensemble model in this invention.

[0061] Figure 5 The figure reflects the accuracy of the ensemble model generated by this invention under different network sizes and connectivity levels. The sparse network in the figure represents a network model with poorer connectivity than the fully connected synchronous network. It can be seen that the average accuracy on the test dataset increases with the increase in the number of miners. This is because as the network size expands, more private data and computing power are invested in training the base model. Furthermore, a higher private dataset ratio significantly improves the accuracy of the ensemble model generated in this invention, indicating that the performance of the ensemble model is improved under higher levels of miner participation and private data abundance. Moreover, comparing the accuracy under the two network models shows that the performance of this invention is essentially unaffected in the sparse network model, highlighting the robustness of this invention in harsher network environments.

[0062] This invention discloses a blockchain-based distributed ensemble learning method. It designs a three-layer blockchain structure consisting of mini-blocks, ensemble blocks, and key blocks. Through a designed consensus protocol, miners in the network train base models on training sets with replacement and aggregate models from other miners. Finally, the information of the base models and ensemble models is recorded on the blockchain, thus integrating the entire process of model training, model ensemble, and model evaluation into the blockchain consensus mechanism. This method automates and distributes the entire ensemble learning process within the blockchain network, improving the computational efficiency of the useful proof-of-work mechanism based on machine learning in the blockchain network. Simultaneously, it avoids introducing a central node during model aggregation, maximizing the decentralization of the blockchain.

[0063] This invention provides a computer system including a memory, a processor, and a computer program / instructions stored in the memory and executable on the processor. When the computer program / instructions are executed by the processor, they implement the steps of the aforementioned method.

[0064] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method.

[0065] This invention provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the aforementioned method.

Claims

1. A three-layer blockchain structure enabling distributed ensemble learning task execution, characterized in that, include: Mini-blocks are used to record the ownership of the model and the hash of the model parameters. Each mini-block corresponds to a unique base model and contains the identifier of the machine learning model parameters, the identifier of the model owner, the hash of the previous key block, the hash of the machine learning task, and a timestamp. An ensemble block is used to record the base models required for each ensemble model and to record the performance metrics of the ensemble model on the validation dataset. An ensemble block points to multiple mini-blocks, and the base models corresponding to the mini-blocks are aggregated to obtain an ensemble model. An ensemble block contains the performance metrics of the ensemble model on the validation dataset, the identifier of the model integrator, the hash of the mini-block corresponding to the integrated base model, the machine learning task hash, and the timestamp. Key blocks are used to store blockchain transactions, task queues, and ensemble model rankings. They include the hash of the previous key block, machine learning task hashes, block-producing miner identifiers, performance metrics of the best ensemble model on the test dataset, timestamps of key block generation, root hashes of the hash tree carrying transaction data, random numbers, task queues, hashes of each ensemble block participating in the ranking, and their performance metrics.

2. A method for generating a three-layer blockchain structure that enables distributed ensemble learning task execution, characterized in that, Includes the following steps: After miners complete the training of the base model on the training dataset, they generate and broadcast mini-blocks. The mini-blocks are used to record the ownership of the model and the hash of the model parameters. Each mini-block corresponds to a unique base model and includes the identifier of the machine learning model parameters, the identifier of the model owner, the hash of the previous key block, the hash of the machine learning task, and a timestamp. After the miner aggregation base model is modeled and its performance is evaluated on the validation dataset, an integrated block is generated and broadcast. The ensemble block is used to record the base models required for each ensemble model and to record the performance metrics of the ensemble model on the validation dataset. The ensemble block points to multiple mini-blocks, and the base models corresponding to the mini-blocks are aggregated to obtain an ensemble model. Each ensemble block contains the performance metrics of the ensemble model on the validation dataset, the identifier of the model integrator, the hash of the mini-block corresponding to the integrated base model, the machine learning task hash, and the timestamp. After miners evaluate the performance metrics of the collected ensemble models on the test dataset, they select the optimal ensemble model and generate and broadcast key blocks. The key blocks are used to store blockchain transactions, task queues, and ensemble model rankings. They include the hash of the previous key block, the hash of the machine learning task, the miner's identifier, the performance metrics of the optimal ensemble model on the test dataset, the timestamp of the key block's generation, the root hash of the hash tree carrying the transaction data, a random number, the task queue, the hashes of each ensemble block participating in the ranking, and its performance metrics.

3. A blockchain-based distributed ensemble learning method, characterized in that, Includes the following steps: The task publisher prepares training, validation, and testing datasets in advance and publishes the machine learning task through blockchain transactions. The task includes the hash values ​​of these three datasets and scripts for training the base model, integrating the models, and evaluating the performance. Miners verify task information and forward it to other miners; when a miner generates the previous key block or receives a valid previous key block from another miner, it downloads the training dataset from the task publisher. After the download is complete, the miner starts training the base model, generates mini-blocks, and broadcasts them. The mini-block is used to record the ownership of the model and the hash of the model parameters; the mini-block contains the identifier of the machine learning model parameters, the identifier of the model owner, the hash of the previous key block, the hash of the machine learning task, and a timestamp; The task publisher publicly releases the verification dataset. After receiving the message, miners begin to aggregate the base model and evaluate its performance on the verification dataset, generate an integrated block and broadcast it. The ensemble block is used to record the base models required for each ensemble model and to record the performance metrics of the ensemble model on the validation dataset. The ensemble block points to multiple mini-blocks, and the base models corresponding to the mini-blocks are aggregated to obtain an ensemble model. The ensemble block contains the performance metrics of the ensemble model on the validation dataset, the identifier of the model integrator, the hash of the mini-block corresponding to the integrated base model, the machine learning task hash, and the timestamp. The task publisher publishes the test dataset. After receiving the message, miners begin to evaluate the performance metrics of the collected ensemble models on the test dataset, select the optimal ensemble model, calculate the key block hash, generate the key block, and broadcast it. The key block is used to store blockchain transactions, task queues, and ensemble model rankings. It includes the hash of the previous key block, machine learning task hash, block miner identifier, performance metrics of the best ensemble model on the test dataset, timestamp of key block generation, root hash of the hash tree carrying transaction data, random number, task queue, hash of each ensemble block participating in the ranking, and its performance metrics. The task publisher obtains the performance metrics of the best ensemble model and the corresponding base model parameters from the key blocks, and obtains an ensemble model with ideal performance by aggregating the base models.

4. The blockchain-based distributed ensemble learning method according to claim 3, characterized in that, A machine learning task published by a user through a blockchain transaction includes a transaction used as a training fee, which transfers the training fee to a virtual address.

5. The blockchain-based distributed ensemble learning method according to claim 3, characterized in that, Miners who generate key blocks and mini blocks will receive rewards. The training cost of the task issuer will be distributed equally among all miners who generate the base model used by the winning ensemble model. At the same time, miners who generate key blocks will be rewarded with newly generated tokens.

6. The blockchain-based distributed ensemble learning method according to claim 3, characterized in that, Miners use both their local private data and the public training data provided by the task publisher to train the base model.

7. The blockchain-based distributed ensemble learning method according to claim 3, characterized in that, The ensemble model is obtained by aggregating the base model using the Bagging algorithm.

8. A computer system comprising a memory, a processor, and computer programs / instructions stored in the memory and executable on the processor, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 2-7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 2-7.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 2-7.

Images