WO2021135467A1

WO2021135467A1 - Automated machine learning-based ethereum fuel restriction prediction method, apparatus, computer device, and storage medium

Info

Publication number: WO2021135467A1
Application number: PCT/CN2020/118901
Authority: WO
Inventors: 张楠; 王健宗; 瞿晓阳
Original assignee: 平安科技（深圳）有限公司
Priority date: 2020-07-31
Filing date: 2020-09-29
Publication date: 2021-07-08
Also published as: CN111898766B; CN111898766A

Abstract

An Ethereum fuel restriction prediction method based on automated machine learning, an apparatus, a computer device, and a storage medium, relating to blockchain smart contract technology, and comprising: obtaining from among target network addresses the network addresses of all published Ethereum smart contracts so as to acquire a set of target smart contract codes having completed verification, and transaction information corresponding to each target smart contract code; after screening all transaction information, inputting corresponding feature set to an automated machine learning model for training of an automated machine learning model; if a current smart contract code is detected, obtaining the current feature set corresponding to the current smart contract code and inputting same to the automated machine learning model for calculation, and obtaining a corresponding Ethernet fuel restriction. The method realizes automatic dimensionality reduction of automatically-screened features on the basis of smart contract codes, so as to predict Ethereum fuel restrictions. Manual intervention is avoided, reducing labor costs, and prediction accuracy is improved.

Description

Ethereum fuel limit prediction method, device, computer equipment and storage medium based on automatic machine learning

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on July 31, 2020, the application number is 202010761121.3, and the invention title is "Ethereum Fuel Restriction Prediction Method and Device Based on Automatic Machine Learning". The entire content of the application is approved The reference is incorporated in this application.

Technical field

This application relates to the field of blockchain technology, and in particular to an Ethereum fuel limit prediction method, device, computer equipment, and storage medium based on automatic machine learning.

Background technique

As a relatively successful blockchain project-Bitcoin, its core innovation is to transfer value from a long distance without trusting a third party. But the disadvantage of Bitcoin is that it does not support Turing's complete scripting language. That is to say, Bitcoin can only be stored in a distributed environment, but it has not been able to store and calculate in a distributed environment. In response to this problem, Vitalik and others launched Ethereum. Compared with Bitcoin, the biggest difference between Ethereum is that Ethereum is a Turing-complete scripting language that allows developers to develop any application on it and implement smart contracts.

Ethereum implements an operating environment on the blockchain, which is called the Ethereum Virtual Machine. Every node participating in the Ethereum network will run the Ethereum virtual machine as part of the block verification protocol. These nodes will verify each transaction covered in the block and run the code triggered by the transaction (code in the smart contract) in the Ethereum virtual machine. All nodes on each network will perform the same calculations and store the same values. In the process of executing these codes and calculations, each command such as addition, hash, etc. will have a specific consumption. On Ethereum, fuel is used for counting. For example, the operation of adding on Ethereum needs to consume 3 Fuel.

Because a certain amount of fuel needs to be consumed in the code execution process, and the fuel consumption is also related to the state of the smart contract. Therefore, the user pays a certain amount of fuel in advance before each transaction. Simply put, this prepaid amount is called a fuel limit in Ethereum. In the process of authentication and calculation of nodes on the network, if the amount of fuel used for the calculation of the user's transaction is less than or equal to the set fuel limit, then the transaction will be processed. On the contrary, if the total fuel consumption exceeds the fuel limit, the fuel provided by the user will be used up, and even all operations in the process will be restored. Therefore, it is very important to ensure the accuracy of the fuel limit.

As a branch of artificial intelligence, machine learning is also a hot research topic. Machine learning algorithm is a kind of algorithm that automatically analyzes and obtains rules from data, and uses the rules to predict unknown data. The biggest advantage of machine learning is that work efficiency is greatly improved. Machine learning cannot solve problems that humans cannot solve, but it can accept a large amount of data, quickly establish connections based on the data, and make predictions. Therefore, when a large amount of data is collected, using machine learning to make predictions is more efficient and more accurate. There are already tens of millions of transactions in Ethereum today. It is undoubtedly a better way to use machine learning to discover the laws in these data. Therefore, it is feasible to use machine learning to predict the fuel limit of transactions on Ethereum. But for the Ethereum transaction program data including the smart contract URL, the inventor realized that the features are not obvious. Manual feature engineering and manual selection of machine learning models require a lot of work and it is difficult to ensure versatility and accuracy.

Summary of the invention

The embodiments of the application provide an Ethereum fuel limit prediction method, device, computer equipment, and storage medium based on automatic machine learning, aiming to solve the problem that the Ethereum transaction program data in the prior art includes the network address of the smart contract, and the data characteristics are not Obviously, manual feature engineering and manual selection of machine learning models require a lot of work, and it is difficult to ensure versatility and accuracy.

In the first aspect, an embodiment of the present application provides an Ethereum fuel limit prediction method based on automatic machine learning, which includes:

Call the preset stored breadth-first algorithm and the preset target URL, and obtain the network addresses of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm;

Acquiring, according to the network address, a set of verified target smart contract codes and transaction information corresponding to each target smart contract code in the target smart contract code set;

Call the pre-stored information field screening strategy, and after information screening of the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening strategy is used to screen the smart contract code The core features in the corresponding transaction information form a feature set;

Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract;

If the current smart contract code uploaded by the client is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

Input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding target Receiving end.

In the second aspect, an embodiment of the present application provides an Ethereum fuel limit prediction device based on automatic machine learning, which includes:

The target network address acquisition unit is used to call the preset stored breadth-first algorithm and the preset target URL, and obtain the network of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm address;

The target code collection obtaining unit is configured to obtain the verified target smart contract code collection and transaction information corresponding to each target smart contract code in the target smart contract code collection according to the network address;

The feature set acquisition unit is used to call a pre-stored information field screening strategy, and after information screening is performed on the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening The strategy is used to filter the core features of the transaction information corresponding to the smart contract code to form a feature set;

The automatic machine learning model training unit is used to obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein, the automatic machine learning model is used to predict the smart contract The fuel limit for calling functions;

The current feature set obtaining unit is configured to, if the current smart contract code uploaded by the user terminal is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

The fuel limit prediction unit is used to input the current feature set into the automatic machine learning model to perform calculations to obtain the Ethereum fuel limit corresponding to the current smart contract code, and calculate the Ethereum fuel limit corresponding to the current smart contract code. The fuel limit is sent to the corresponding target receiver.

In a third aspect, an embodiment of the present application provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and running on the processor, and the processor executes the computer The following steps are implemented during the program:

Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein, the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract;

In a fourth aspect, the embodiments of the present application also provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, which when executed by a processor causes the processor to perform the following operations :

The embodiment of the application provides an Ethereum fuel limit prediction method, device, computer equipment and storage medium based on automatic machine learning, including the breadth-first search corresponding to the breadth-first algorithm to obtain from the target URL all published on Ethereum The network address of the smart contract; according to the network address, obtain the target smart contract code set that has been verified, and the transaction information corresponding to each target smart contract code in the target smart contract code set; call the information field screening strategy, and the target smart contract code After the corresponding transaction information is filtered, the feature set corresponding to each target smart contract code is obtained; the feature set corresponding to each target smart contract code is obtained and input to the automatic machine learning model to be trained for training, and the automatic machine learning model is obtained; Upload the current smart contract code to the client, and obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; input the current feature set into the automatic machine learning model for calculation, and obtain the Ethereum fuel corresponding to the current smart contract code Limit, send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding target receiver. After realizing the automatic dimensionality reduction based on the automatic screening of the smart contract code, the Ethereum fuel limit is predicted, which not only avoids manual intervention and reduces labor costs, but also improves the accuracy of prediction.

Description of the drawings

In order to explain the technical solutions of the embodiments of the present application more clearly, the following will briefly introduce the drawings used in the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present application. Ordinary technicians can obtain other drawings based on these drawings without creative work.

FIG. 1 is a schematic diagram of an application scenario of an Ethereum fuel limit prediction method based on automatic machine learning provided by an embodiment of the application;

FIG. 2 is a schematic flowchart of an Ethereum fuel limit prediction method based on automatic machine learning provided by an embodiment of the application;

3 is a schematic block diagram of an Ethereum fuel limit prediction device based on automatic machine learning provided by an embodiment of the application;

Fig. 4 is a schematic block diagram of a computer device provided by an embodiment of the application.

Detailed ways

The technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of them. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

It should be understood that when used in this specification and appended claims, the terms "including" and "including" indicate the existence of the described features, wholes, steps, operations, elements and/or components, but do not exclude one or The existence or addition of multiple other features, wholes, steps, operations, elements, components, and/or collections thereof.

It should also be understood that the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in the specification of this application and the appended claims, unless the context clearly indicates other circumstances, the singular forms "a", "an" and "the" are intended to include plural forms.

It should be further understood that the term "and/or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations .

Please refer to Figures 1 and 2. Figure 1 is a schematic diagram of an application scenario of an Ethereum fuel limit prediction method based on automatic machine learning provided by an embodiment of this application; Figure 2 is an Ethereum fuel based on automatic machine learning provided by an embodiment of this application Schematic diagram of the flow of the restriction prediction method. The Ethereum fuel restriction prediction method based on automatic machine learning is applied to the server, and the method is executed by the application software installed in the server.

As shown in Figure 2, the method includes steps S110 to S160.

S110: Call a preset stored breadth-first algorithm and a preset target URL, and obtain network addresses of all smart contracts that have been published on Ethereum from the target URL through a breadth-first search corresponding to the breadth-first algorithm.

In this embodiment, the target URL (the target URL is specifically implemented as http://etherscan.io/) stores the verified smart contracts on Ethereum and the transaction information related to these smart contracts, and the breadth-first algorithm The corresponding breadth-first search can collect the network addresses of all smart contracts that have been published on Ethereum from the content of the web pages at all levels of the target URL. Through the breadth-first search method, the traversal acquisition of the network addresses of all smart contracts that have been published on Ethereum is realized, and the amount of collected network address collection data is also richer.

In an embodiment, step S110 includes:

Obtain all the network addresses of all smart contracts that have been published on Ethereum in the first-level webpage of the target URL to form a first-level network address set;

Visit all the second-level webpages adjacent to the first-level webpage, and obtain all the network addresses of all smart contracts that have been published on Ethereum in the second-level webpage to form a second-level network address set; visit all in order The third-level webpage adjacent to the second-level webpage until the access to all the nth-level webpages adjacent to the n-1th-level webpage to obtain the third-level network address set to the nth-level network address set respectively ; Wherein, the value of n is equal to the total page level of the target URL;

The first-level network address set to the nth-level network address set form the network addresses of all smart contracts that have been published on Ethereum in the target URL.

In this embodiment, the breadth-first search algorithm (also called breadth-first search) is one of the simplest graph search algorithms, and this algorithm is also the prototype of many important graph algorithms. The core idea of the breadth-first algorithm is: starting from the initial node, apply operators to generate the first-level nodes, check whether the target node is among these subsequent nodes, and if not, use production rules to expand all the first-level nodes one by one. Get the second-level nodes, and check whether the second-level nodes contain the target node one by one. If not, use the operator to expand all the nodes of the second layer one by one..., expand in this way, and check until the target node is found. Using the breadth-first algorithm, the search depth is small, and each node is only visited once, and the node is always accessed by the shortest path, which improves the efficiency of obtaining the network addresses of all smart contracts that have been published on Ethereum.

S120: Obtain, according to the network address, a target smart contract code set that has been verified, and transaction information corresponding to each target smart contract code in the target smart contract code set.

In this embodiment, the smart contract that has been verified on the target website is tagged with a verified tag. At this time, smart contract codes with verified tags can be quickly screened out to form a target smart contract code set. The transaction information corresponding to each smart contract has multiple field values. The values of these fields may all be related to fuel limit prediction, or some of them may be related to fuel limit prediction. In the subsequent steps, the transaction information corresponding to the smart contract needs to be included. Core fields are filtered.

In an embodiment, step S120 includes:

Name and store each target smart contract code in the target smart contract code set for which the network address has been obtained and verified;

Obtain the block height of the exchange, the hash value of the transaction, the fuel limit, the fuel actually used to execute the transaction separately, and the input data of the exchange function used in the transaction information of each target smart contract code; among them, the exchange The input data using the function includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction.

In this embodiment, since there are several versions of some smart contracts on the blockchain, the version names are the same, but the code of each version may be different. At this time, the name is based on the smart contract name + smart contract version number + smart contract upload time, so that smart contracts with the same version name but different codes can be distinguished by different names.

Obtaining the transaction information of the smart contract mainly includes: the block height of the exchange, the hash value of the transaction, the fuel limit, the fuel actually used to execute the transaction separately, and the input data of the exchange to use the function of the exchange; among them, the exchange uses The input data of the function includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction. According to the analysis of the importance of data, the input data of the general exchange use function has the most relevance to the fuel limit prediction, so you can set the information field filtering strategy for filtering out the value of the core field in the target smart contract code for subsequent practical use.

S130. Invoke a pre-stored information field screening strategy, and after information screening is performed on the transaction information corresponding to each target smart contract code, a feature set corresponding to each target smart contract code is obtained; wherein, the information field screening strategy is used to screen intelligence The core features in the transaction information corresponding to the contract code constitute a feature set.

In this embodiment, since the transaction information corresponding to each target smart contract code includes many fields, in order to reduce the data dimension, the transaction information corresponding to each target smart contract code can be filtered to obtain the corresponding target smart contract code. The corresponding feature set.

In an embodiment, step S130 includes:

Obtain the core feature field set included in the information field screening strategy; wherein, the core feature field set includes the block height field of the exchange, the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function field, and the exchange execution The FOR loop count field in the function and the number field of the variable in the transaction;

The transaction information corresponding to each target smart contract code is filtered according to the core feature field set, and the feature set corresponding to each target smart contract code is obtained.

In this embodiment, in the Ethereum virtual machine environment, each operation needs to consume part of the fuel. For example, an addition operation on the Ethereum needs to consume 3 fuels and so on. The above example is the fuel consumed by a single operation. The fuel consumed by the functions involved in the smart contract code is not the sum of the fuel consumed by the above data operations. However, it is known that the corresponding values of the core feature fields included in the information field screening strategy are positively correlated with fuel consumption.

In specific implementation, the core feature fields that are positively related to fuel consumption include the exchange block height field, the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR cycles in the exchange execution function, and the transaction. The number field of the middle variable is set in the information field screening strategy to be used to screen the corresponding values of the core feature fields included in the feature set. After the transaction information corresponding to each target smart contract code is screened accordingly, the result of the screening is the feature set corresponding to each target smart contract code. Through this screening process, the data features are effectively reduced in dimensionality.

S140. Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract.

In this embodiment, machine learning allows the algorithm to automatically find a set of rules from the data, so as to extract the features of the data that are helpful for classification/clustering/decision-making. With the development of machine learning, manual intervention is required. There are more and more parts, and AutoML (that is, automatic machine learning) automates the entire process of machine learning models from construction to application, and finally an end-to-end model (end to end) is obtained.

The application of machine learning requires a lot of manual intervention, which is manifested in various aspects of machine learning such as feature extraction, model selection, and parameter adjustment. Automatic machine learning (AutoML) attempts to learn these important steps related to features, models, optimization, and evaluation automatically, so that machine learning models can be applied without manual intervention.

For transaction fuel prediction tasks, the input data structure is complex, and it also contains features that are difficult to quantify such as code text. Traditional machine learning regression methods cannot directly accept text as features, and it is difficult to perform feature engineering and model selection. , If you use automated machine learning methods, the difficulty of the problem can be greatly reduced.

In an embodiment, step S140 includes:

Call the pre-stored principal component analysis algorithm to select the main feature of the feature set corresponding to each target smart contract code, and obtain the dimensionality reduction feature set corresponding to each feature set;

The automatic machine learning model to be trained is sequentially subjected to model training, model selection/combination, and hyperparameter tuning according to the dimensionality reduction feature set to obtain an automatic machine learning model.

In this embodiment, the principal component analysis algorithm, PCA, is mainly used for data dimensionality reduction. In the process of training the automatic machine learning model, the data dimension of the training set can not be too much. At this time, the main feature selection can be performed on the feature set corresponding to each target smart contract code, and the dimensionality reduction feature set corresponding to each feature set can be obtained. Realize data dimensionality reduction.

Then, according to the dimensionality reduction feature set, the automatic machine learning model to be trained is sequentially performed model training, model selection/combination, and hyperparameter tuning to obtain the automatic machine learning model.

From the sequence of the automatic machine learning process, it is initially data preparation, including data collection and cleaning, and then feature engineering, which includes feature selection, feature extraction (reducing features, commonly used methods such as PCA), and feature combination (Combine/construct multiple features into a new feature); In the subsequent model construction, the most critical is model selection. After hyperparameter optimization, many methods can be adopted. The simplest method is grid search, which is commonly used. Methods include reinforcement learning, evolutionary algorithms, Bayesian optimization, and gradient descent to narrow the search space; finally, AutoML automates the process of model evaluation by introducing early stopping, reducing the accuracy of the model, and sharing parameters.

The task of data collection is not to search and collect real data, but also to generate simulation data to expand the training data set. New technologies that can be used include countering neural networks, and reinforcement learning frameworks can also be used to optimize control. The parameters of the generated data, so that the generated data can more effectively assist the training of the model. The data cleaning is to automatically complete the work that was manually completed before including missing value completion, outlier processing, feature normalization, and different coding of categorical features.

For automatic model selection, the traditional method is to select one or more models from traditional models, such as KNN, SVM, and decision trees, or combine them with the best model. The current research hotspot of AutoML is neural architecture search. It is Neural Architecture Search), that is, without manual intervention, the model automatically generates a network structure that is most effective for the current task.

After the model is selected, the most commonly used is the grid search, that is, according to a fixed interval, in the search space, AutoML will use random sampling, first evaluate the importance of each hyperparameter, and then focus on the important The parameters are fine-tuned.

After the automatic machine learning model to be trained is trained through the dimensionality reduction feature set, an automatic machine learning model for predicting the fuel limit of the function called by the smart contract can be obtained.

S150: If the current smart contract code uploaded by the user terminal is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy.

In this embodiment, in order to predict the Ethereum fuel limit corresponding to other smart contract codes, the user can directly input the current smart contract code on the user side at this time, and then the user side sends the current smart contract code to the server. Pre-processing the fuel limit prediction data by the server. At this time, reference may also be made to step S130 to obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy. Prediction through this streamlined current feature set effectively reduces the data dimension and improves the efficiency of subsequent predictions.

S160. Input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding The target receiver.

In this embodiment, the current feature set is input to the automatic machine learning model for calculation, and the automatic machine learning model can extract the fuel consumption limit corresponding to each function or data operation in the current feature set, thereby accurately predicting Get the Ethereum fuel limit corresponding to the current smart contract code. At this time, in order to timely notify the user of the prediction result, the Ethereum fuel limit corresponding to the current smart contract code can be sent to the corresponding target receiving end.

In an embodiment, step S160 includes:

Performing main feature selection on the current feature set according to the principal component analysis algorithm, and obtaining a current dimensionality reduction feature set corresponding to the current feature set;

Inputting the current dimensionality reduction feature set to the model selection model in the automatic machine learning model for calculation to obtain a target model;

The current feature set is input to the target model for calculation, and the Ethereum fuel limit corresponding to the current smart contract code is obtained.

In this embodiment, when the fuel limit prediction is performed through automatic machine learning, it mainly realizes the automatic selection of models. The traditional method is to select one or more models with the best combination effect from the traditional models, and Currently, AutoML can search through neural architecture, that is, without manual intervention, the model automatically generates a network structure that is most effective for the current task, which greatly improves the accuracy of the prediction results.

This method realizes the automatic dimensionality reduction based on the automatic screening of the smart contract code to predict the Ethereum fuel limit, which not only avoids manual intervention and reduces labor costs, but also improves the accuracy of the prediction.

The embodiment of the present application also provides an Ethereum fuel limit prediction device based on automatic machine learning. The Ethereum fuel limit prediction device based on automatic machine learning is used to execute any of the aforementioned methods for predicting Ethereum fuel limit based on automatic machine learning. Examples. Specifically, please refer to FIG. 3, which is a schematic block diagram of an Ethereum fuel limit prediction device based on automatic machine learning provided by an embodiment of the present application. The Ethereum fuel limit prediction device 100 based on automatic machine learning can be configured in a server.

As shown in FIG. 3, the Ethereum fuel limit prediction device 100 based on automatic machine learning includes: a target network address acquisition unit 110, a target code set acquisition unit 120, a feature set acquisition unit 130, an automatic machine learning model training unit 140, and current features The set acquisition unit 150 and the fuel limit prediction unit 160.

The target network address obtaining unit 110 is used to call a preset stored breadth-first algorithm and a preset target URL, and obtain the information of all smart contracts that have been published on Ethereum from the target URL through a breadth-first search corresponding to the breadth-first algorithm website address.

In an embodiment, the target network address obtaining unit 110 includes:

The first-level obtaining unit is used to obtain all the network addresses of all smart contracts that have been published on Ethereum in the first-level webpage of the target URL to form a first-level network address set;

Traverse the next-level acquisition unit to access all second-level webpages adjacent to the first-level webpage, and obtain all the network addresses of all smart contracts that have been published on Ethereum in the second-level webpage to form a second-level webpage. Level network address set; sequentially visit all third-level webpages adjacent to the second-level webpage until accessing all nth-level webpages adjacent to the n-1th-level webpage to obtain the third-level network respectively The address set to the nth-level network address set; wherein the value of n is equal to the total number of page levels of the target URL;

The network address combination unit is configured to form the network addresses of all smart contracts that have been published on Ethereum in the target URL from the first-level network address set to the nth-level network address set.

The target code set obtaining unit 120 is configured to obtain a verified target smart contract code set and transaction information corresponding to each target smart contract code in the target smart contract code set according to the network address.

In an embodiment, the target code set obtaining unit 120 includes:

The naming storage unit is used to respectively name and store each target smart contract code in the target smart contract code set for which the network address has been verified;

The transaction information analysis and acquisition unit is used to acquire the block height of the exchange included in the transaction information of each target smart contract code, the hash value of the transaction, the fuel limit, the fuel actually used to execute the transaction separately, and the function used by the exchange The input data of the function used by the exchange includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction.

The feature set acquisition unit 130 is configured to call a pre-stored information field screening strategy, and after information screening is performed on the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field The screening strategy is used to screen the core features of the transaction information corresponding to the smart contract code to form a feature set.

In an embodiment, the feature set obtaining unit 130 includes:

The information field screening strategy obtaining unit is used to obtain the core feature field set included in the information field screening strategy; wherein, the core feature field set includes the field of the block height of the exchange, the field of the number of times the exchange executes the SHA256 function, and the field of transaction The number of times field of SHA3 function execution, the number of FOR loops field in the exchange execution function and the number field of the variables in the transaction;

The core feature field acquisition unit is used to screen the transaction information corresponding to each target smart contract code according to the core feature field set to obtain the feature set corresponding to each target smart contract code.

The automatic machine learning model training unit 140 is used to obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein, the automatic machine learning model is used to predict the smart contract The fuel limit of the called function.

For transaction fuel prediction tasks, the input data structure is complex, and it also contains features that are difficult to quantify, such as code text. Traditional machine learning regression methods cannot directly accept text as features, and it is difficult to perform feature engineering and model selection. , If you use automated machine learning methods, the difficulty of the problem can be greatly reduced.

In an embodiment, the automatic machine learning model training unit 140 includes:

The data dimensionality reduction unit is used to call the pre-stored principal component analysis algorithm to select the main feature of the feature set corresponding to each target smart contract code to obtain the dimensionality reduction feature set corresponding to each feature set;

The model training unit is used to sequentially perform model training, model selection/combination, and hyperparameter tuning on the automatic machine learning model to be trained according to the dimensionality reduction feature set to obtain the automatic machine learning model.

The current feature set obtaining unit 150 is configured to, if the current smart contract code uploaded by the user terminal is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy.

In this embodiment, in order to predict the Ethereum fuel limit corresponding to other smart contract codes, the user can directly input the current smart contract code on the user side at this time, and then the user side sends the current smart contract code to the server. Pre-processing the fuel limit prediction data by the server. At this time, you can also refer to the feature set obtaining unit 130 to obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy. Prediction through this streamlined current feature set effectively reduces the data dimension and improves the efficiency of subsequent predictions.

The fuel limit prediction unit 160 is configured to input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and calculate the Ethereum fuel limit corresponding to the current smart contract code. The fuel limit is sent to the corresponding target receiver.

In an embodiment, the fuel limit prediction unit 160 includes:

The current dimensionality reduction unit is configured to perform main feature selection on the current feature set according to the principal component analysis algorithm, and obtain a current dimensionality reduction feature set corresponding to the current feature set;

A target model selection unit, configured to input the current dimensionality reduction feature set into a model selection model in the automatic machine learning model for calculation to obtain a target model;

The current prediction unit is used to input the current feature set to the target model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code.

This device realizes automatic dimensionality reduction based on the automatic screening of smart contract codes to predict the Ethereum fuel limit, which not only avoids manual intervention and reduces labor costs, but also improves the accuracy of prediction.

The above-mentioned Ethereum fuel limit prediction device based on automatic machine learning can be implemented in the form of a computer program, which can be run on a computer device as shown in FIG. 4.

Please refer to FIG. 4, which is a schematic block diagram of a computer device according to an embodiment of the present application. The computer device 500 is a server, and the server may be an independent server or a server cluster composed of multiple servers.

4, the computer device 500 includes a processor 502, a memory, and a network interface 505 connected through a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.

The non-volatile storage medium 503 can store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, the processor 502 can execute the Ethereum fuel limit prediction method based on automatic machine learning.

The processor 502 is used to provide calculation and control capabilities, and support the operation of the entire computer device 500.

The internal memory 504 provides an environment for the operation of the computer program 5032 in the non-volatile storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute the Ethereum fuel limit prediction method based on automatic machine learning. .

The network interface 505 is used for network communication, such as providing data information transmission. Those skilled in the art can understand that the structure shown in FIG. 4 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device 500 to which the solution of the present application is applied. The specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

Wherein, the processor 502 is configured to run a computer program 5032 stored in a memory to implement the Ethereum fuel limit prediction method based on automatic machine learning disclosed in the embodiment of the present application.

Those skilled in the art can understand that the embodiment of the computer device shown in FIG. 4 does not constitute a limitation on the specific configuration of the computer device. In other embodiments, the computer device may include more or less components than those shown in the figure. Or some parts are combined, or different parts are arranged. For example, in some embodiments, the computer device may only include a memory and a processor. In such embodiments, the structures and functions of the memory and the processor are consistent with the embodiment shown in FIG. 4, and will not be repeated here.

It should be understood that, in this embodiment of the application, the processor 502 may be a central processing unit (Central Processing Unit, CPU), and the processor 502 may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSPs), Application Specific Integrated Circuit (ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor.

In another embodiment of the present application, a computer-readable storage medium is provided. The computer-readable storage medium may be non-volatile or volatile. The computer-readable storage medium stores a computer program, where the computer program is executed by a processor to implement the Ethereum fuel limit prediction method based on automatic machine learning disclosed in the embodiments of the present application.

Those skilled in the art can clearly understand that, for the convenience and conciseness of description, the specific working process of the above-described equipment, device, and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here. A person of ordinary skill in the art may be aware that the units and algorithm steps of the examples described in the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both, in order to clearly illustrate the hardware and software Interchangeability, in the above description, the composition and steps of each example have been generally described in accordance with the function. Whether these functions are executed by hardware or software depends on the specific application and design constraint conditions of the technical solution. Professionals and technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of this application.

In the several embodiments provided in this application, it should be understood that the disclosed equipment, device, and method may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods, or the units with the same function may be combined into one. Units, for example, multiple units or components can be combined or integrated into another system, or some features can be omitted or not implemented. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may also be electrical, mechanical or other forms of connection.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments of the present application.

In addition, the functional units in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or software functional unit.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application is essentially or the part that contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium. It includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), magnetic disk or optical disk and other media that can store program codes.

The above are only specific implementations of this application, but the protection scope of this application is not limited to this. Anyone familiar with the technical field can easily think of various equivalents within the technical scope disclosed in this application. Modifications or replacements, these modifications or replacements shall be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

An Ethereum fuel limit prediction method based on automatic machine learning, which includes:

Call the preset stored breadth-first algorithm and the preset target URL, and obtain the network addresses of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm;

Acquiring, according to the network address, a set of verified target smart contract codes and transaction information corresponding to each target smart contract code in the target smart contract code set;

Call the pre-stored information field screening strategy, and after information screening of the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening strategy is used to screen the smart contract code The core features in the corresponding transaction information form a feature set;

Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract;

If the current smart contract code uploaded by the client is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

Input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding target Receiving end.
The Ethereum fuel limit prediction method based on automatic machine learning according to claim 1, wherein the breadth-first search corresponding to the breadth-first algorithm obtains from the target URL the network that has issued all smart contracts on the Ethereum Address, including;

Obtain all the network addresses of all smart contracts that have been published on Ethereum in the first-level webpage of the target URL to form a first-level network address set;

Visit all the second-level webpages adjacent to the first-level webpage, and obtain all the network addresses of all smart contracts that have been published on Ethereum in the second-level webpage to form a second-level network address set; visit all in order The third-level webpage adjacent to the second-level webpage until the access to all the nth-level webpages adjacent to the n-1th-level webpage to obtain the third-level network address set to the nth-level network address set respectively ; Wherein, the value of n is equal to the total page level of the target URL;

The first-level network address set to the nth-level network address set form the network addresses of all smart contracts that have been published on Ethereum in the target URL.
The method for predicting Ethereum fuel limit based on automatic machine learning according to claim 1, wherein said obtaining the verified target smart contract code set according to the network address, and each target smart contract code set in the target smart contract code set. The transaction information corresponding to the contract code includes:

Name and store each target smart contract code in the target smart contract code set for which the network address has been obtained and verified;

Obtain the block height of the exchange, the hash value of the transaction, the fuel limit, the fuel actually used to execute the transaction separately, and the input data of the exchange function used in the transaction information of each target smart contract code; among them, the exchange The input data using the function includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction.
The Ethereum fuel limit prediction method based on automatic machine learning according to claim 3, wherein said calling a pre-stored information field screening strategy, after information screening of the transaction information corresponding to each target smart contract code, is The feature set corresponding to the target smart contract code includes:

Obtain the core feature field set included in the information field screening strategy; wherein, the core feature field set includes the block height field of the exchange, the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function field, and the exchange execution The FOR loop count field in the function and the number field of the variable in the transaction;

The transaction information corresponding to each target smart contract code is filtered according to the core feature field set, and the feature set corresponding to each target smart contract code is obtained.
The method for predicting Ethereum fuel limit based on automatic machine learning according to claim 4, wherein said acquiring the feature set corresponding to each target smart contract code is input to the automatic machine learning model to be trained for training to obtain the automatic machine learning model, include:

Call the pre-stored principal component analysis algorithm to select the main feature of the feature set corresponding to each target smart contract code, and obtain the dimensionality reduction feature set corresponding to each feature set;

The automatic machine learning model to be trained is sequentially subjected to model training, model selection/combination, and hyperparameter tuning according to the dimensionality reduction feature set to obtain an automatic machine learning model.
The method for predicting Ethereum fuel limit based on automatic machine learning according to claim 5, wherein said inputting said current feature set into said automatic machine learning model to perform calculations to obtain said current smart contract code Ethereum fuel restrictions include:

Performing main feature selection on the current feature set according to the principal component analysis algorithm, and obtaining a current dimensionality reduction feature set corresponding to the current feature set;

Inputting the current dimensionality reduction feature set to the model selection model in the automatic machine learning model for calculation to obtain a target model;

The current feature set is input to the target model for calculation, and the Ethereum fuel limit corresponding to the current smart contract code is obtained.
The Ethereum fuel limit prediction method based on automatic machine learning according to claim 1, wherein the said network address is obtained by naming and storing each target smart contract code in the target smart contract code set that has been verified. include:

Name and store each target smart contract code in the target smart contract code set for which the network address has been verified by the smart contract name + smart contract version number + smart contract upload time.
The method for predicting Ethereum fuel limit based on automatic machine learning according to claim 6, wherein said inputting said current feature set into said target model for calculation to obtain Ethereum fuel corresponding to said current smart contract code Restrictions include:

An automatic machine learning model is used to extract the fuel limit corresponding to each function or data operation in the current feature set to obtain the Ethereum fuel limit corresponding to the current smart contract code.
An Ethereum fuel limit prediction device based on automatic machine learning, which includes:

The target network address acquisition unit is used to call the preset stored breadth-first algorithm and the preset target URL, and obtain the network of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm address;

The target code collection obtaining unit is configured to obtain the verified target smart contract code collection and transaction information corresponding to each target smart contract code in the target smart contract code collection according to the network address;

The feature set acquisition unit is used to call a pre-stored information field screening strategy, and after information screening is performed on the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening The strategy is used to filter the core features of the transaction information corresponding to the smart contract code to form a feature set;

The automatic machine learning model training unit is used to obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein, the automatic machine learning model is used to predict the smart contract The fuel limit for calling functions;

The current feature set obtaining unit is configured to, if the current smart contract code uploaded by the user terminal is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

The fuel limit prediction unit is used to input the current feature set into the automatic machine learning model to perform calculations to obtain the Ethereum fuel limit corresponding to the current smart contract code, and calculate the Ethereum fuel limit corresponding to the current smart contract code. The fuel limit is sent to the corresponding target receiver.
A computer device includes a memory, a processor, and a computer program stored on the memory and capable of running on the processor, wherein the processor implements the following steps when the processor executes the computer program:

Call the preset stored breadth-first algorithm and the preset target URL, and obtain the network addresses of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm;

Acquiring, according to the network address, a set of verified target smart contract codes and transaction information corresponding to each target smart contract code in the target smart contract code set;

Call the pre-stored information field screening strategy, and after information screening of the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening strategy is used to screen the smart contract code The core features in the corresponding transaction information form a feature set;

Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract;

If the current smart contract code uploaded by the client is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

Input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding target Receiving end.
The computer device according to claim 10, wherein the breadth-first search corresponding to the breadth-first algorithm obtains the network addresses of all smart contracts that have been published on Ethereum from the target website, comprising;

Obtain all the network addresses of all smart contracts that have been published on Ethereum in the first-level webpage of the target URL to form a first-level network address set;

Visit all the second-level webpages adjacent to the first-level webpage, and obtain all the network addresses of all smart contracts that have been published on Ethereum in the second-level webpage to form a second-level network address set; visit all in order The third-level webpage adjacent to the second-level webpage until the access to all the nth-level webpages adjacent to the n-1th-level webpage to obtain the third-level network address set to the nth-level network address set respectively ; Wherein, the value of n is equal to the total page level of the target URL;

The first-level network address set to the nth-level network address set form the network addresses of all smart contracts that have been published on Ethereum in the target URL.
The computer device according to claim 10, wherein said obtaining the verified target smart contract code set and the transaction information corresponding to each target smart contract code in the target smart contract code set according to the network address comprises:

Name and store each target smart contract code in the target smart contract code set for which the network address has been obtained and verified;

Obtain the block height of the exchange, the hash value of the transaction, the fuel limit, the fuel actually used to execute the transaction separately, and the input data of the exchange function used in the transaction information of each target smart contract code; among them, the exchange The input data using the function includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction.
The computer device according to claim 12, wherein said calling a pre-stored information field screening strategy, after information screening of transaction information corresponding to each target smart contract code, obtains a feature set corresponding to each target smart contract code, include:

Obtain the core feature field set included in the information field screening strategy; wherein, the core feature field set includes the block height field of the exchange, the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function field, and the exchange execution The FOR loop count field in the function and the number field of the variable in the transaction;

The transaction information corresponding to each target smart contract code is filtered according to the core feature field set, and the feature set corresponding to each target smart contract code is obtained.
The computer device according to claim 13, wherein said obtaining the feature set corresponding to each target smart contract code and inputting it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model comprises:

Call the pre-stored principal component analysis algorithm to select the main feature of the feature set corresponding to each target smart contract code, and obtain the dimensionality reduction feature set corresponding to each feature set;

The automatic machine learning model to be trained is sequentially subjected to model training, model selection/combination, and hyperparameter tuning according to the dimensionality reduction feature set to obtain an automatic machine learning model.
The computer device according to claim 14, wherein the inputting the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code comprises:

Performing main feature selection on the current feature set according to the principal component analysis algorithm, and obtaining a current dimensionality reduction feature set corresponding to the current feature set;

Inputting the current dimensionality reduction feature set to the model selection model in the automatic machine learning model for calculation to obtain a target model;

The current feature set is input to the target model for calculation, and the Ethereum fuel limit corresponding to the current smart contract code is obtained.
The computer device according to claim 10, wherein said naming and storing each target smart contract code in the target smart contract code set for which the verification of the network address has been completed comprises:

Each target smart contract code in the target smart contract code collection that has been verified by the network address is respectively named and stored by the smart contract name + smart contract version number + smart contract upload time.
The computer device according to claim 15, wherein said inputting said current feature set into said target model for calculation to obtain the Ethereum fuel limit corresponding to said current smart contract code comprises:

An automatic machine learning model is used to extract the fuel limit corresponding to each function or data operation in the current feature set to obtain the Ethereum fuel limit corresponding to the current smart contract code.
A computer-readable storage medium, wherein the computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the following operations:

Call the preset stored breadth-first algorithm and the preset target URL, and obtain the network addresses of all smart contracts that have been published on Ethereum from the target URL through the breadth-first search corresponding to the breadth-first algorithm;

Acquiring, according to the network address, a set of verified target smart contract codes and transaction information corresponding to each target smart contract code in the target smart contract code set;

Call the pre-stored information field screening strategy, and after information screening of the transaction information corresponding to each target smart contract code, the feature set corresponding to each target smart contract code is obtained; wherein, the information field screening strategy is used to screen the smart contract code The core features in the corresponding transaction information form a feature set;

Obtain the feature set corresponding to each target smart contract code and input it to the automatic machine learning model to be trained for training to obtain the automatic machine learning model; wherein the automatic machine learning model is used to predict the fuel limit of the function called by the smart contract;

If the current smart contract code uploaded by the client is detected, obtain the current feature set corresponding to the current smart contract code according to the information field screening strategy; and

Input the current feature set into the automatic machine learning model for calculation to obtain the Ethereum fuel limit corresponding to the current smart contract code, and send the Ethereum fuel limit corresponding to the current smart contract code to the corresponding target Receiving end.
The computer-readable storage medium according to claim 18, wherein the breadth-first search corresponding to the breadth-first algorithm obtains the network addresses of all smart contracts that have been published on Ethereum from the target website, comprising;

Obtain all the network addresses of all smart contracts that have been published on Ethereum in the first-level webpage of the target URL to form a first-level network address set;

Visit all the second-level webpages adjacent to the first-level webpage, and obtain all the network addresses of all smart contracts that have been published on Ethereum in the second-level webpage to form a second-level network address set; visit all in order The third-level webpage adjacent to the second-level webpage until the access to all the nth-level webpages adjacent to the n-1th-level webpage to obtain the third-level network address set to the nth-level network address set respectively ; Wherein, the value of n is equal to the total page level of the target URL;

The first-level network address set to the nth-level network address set form the network addresses of all smart contracts that have been published on Ethereum in the target URL.
The computer-readable storage medium according to claim 18, wherein said obtaining a verified target smart contract code set according to the network address, and transaction information corresponding to each target smart contract code in the target smart contract code set ,include:

Name and store each target smart contract code in the target smart contract code set for which the network address has been obtained and verified;

Obtain the block height of the exchange, the hash value of the transaction, the fuel limit, the fuel actually used to execute this transaction separately, and the input data of the exchange function used in the transaction information of each target smart contract code; among them, the exchange The input data using the function includes the number of times the exchange executes the SHA256 function, the number of times the transaction executes the SHA3 function, the number of FOR loops in the exchange execution function, and the number of variables in the transaction.