WO2024114338A1 - 一种训练行为预测模型的方法及装置 - Google Patents

一种训练行为预测模型的方法及装置 Download PDF

Info

Publication number
WO2024114338A1
WO2024114338A1 PCT/CN2023/130960 CN2023130960W WO2024114338A1 WO 2024114338 A1 WO2024114338 A1 WO 2024114338A1 CN 2023130960 W CN2023130960 W CN 2023130960W WO 2024114338 A1 WO2024114338 A1 WO 2024114338A1
Authority
WO
WIPO (PCT)
Prior art keywords
behavior
model
multiple single
sequences
time
Prior art date
Application number
PCT/CN2023/130960
Other languages
English (en)
French (fr)
Inventor
李晓静
许涛
于飞
陆鑫
Original Assignee
蚂蚁财富(上海)金融信息服务有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 蚂蚁财富(上海)金融信息服务有限公司 filed Critical 蚂蚁财富(上海)金融信息服务有限公司
Publication of WO2024114338A1 publication Critical patent/WO2024114338A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"

Definitions

  • One or more embodiments of the present specification relate to the field of computer technology, and more specifically, to a method and apparatus for training a behavior prediction model in the field of computer technology.
  • the traditional behavior sequence modeling method is to discretize the multi-behavior sequence, convert it into a problem in multiple uniform time intervals, and perform time aggregation on multiple behaviors in each time interval to obtain corresponding features.
  • the above method fuzzifies the original time information of each behavior in the multi-behavior sequence, so that the multi-behavior sequence obtained by modeling is significantly different from the original behavior sequence (unprocessed multi-behavior sequence).
  • the time and type of future behaviors predicted by the modeled multi-behavior sequence are inaccurate. Therefore, it is necessary to provide a more accurate method for modeling multi-behavior sequences.
  • One or more embodiments of the present specification provide a method and device for training a behavior prediction model, which can accurately model the time point information of each behavior in a user's multi-behavior sequence.
  • a method for training a behavior prediction model comprising: splitting a user's behavior sequence to obtain multiple single behavior sequences, each single behavior sequence corresponding to a behavior, and used to record the correspondence between the behavior and the time point; time-coding the multiple single behavior sequences to obtain multiple single behavior time series of the multiple single behavior sequences; inputting the multiple single behavior time series into a behavior prediction model, using the behavior prediction model to model the time point for each single behavior time series in the multiple single behavior time series, and obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series; based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and a first loss value between the multiple single behavior time series, training the behavior prediction model, the behavior prediction model being used to predict the time point and type of at least one future behavior of the user based on the user's behavior sequence.
  • one or more embodiments of this specification split the user's behavior sequence to obtain multiple single behavior sequence; time-code all time points in multiple single behavior sequences to obtain multiple single behavior time series of multiple single behavior sequences; input multiple single behavior time series into the behavior prediction model to obtain the temporal distribution of behaviors corresponding to the multiple single behavior time series predicted by the behavior prediction model; thus, the behavior prediction model can be trained based on the temporal distribution of behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior prediction model focuses on each time point of multiple single behaviors in the user's behavior sequence and models each time point.
  • a more accurate behavior prediction model can be modeled, and the behavior prediction model can be used to predict more accurate time points and types of future behaviors.
  • the user's behavior sequence is split to obtain multiple single behavior sequences, including: splitting the user's behavior sequence according to the type of behavior to obtain the multiple single behavior sequences, and the user's behavior sequence corresponds to multiple behavior types.
  • the user's behavior sequence since the user's behavior sequence corresponds to multiple behavior types, the user's behavior sequence can be split according to the behavior type into multiple single behavior sequences, each of which corresponds to a behavior type.
  • the multiple single behavior sequences can be time-coded to obtain multiple single behavior time series of the multiple single behavior sequences, and the multiple single behavior time series can be used to train the behavior prediction model.
  • the multiple single behavior sequences are time-encoded to obtain multiple single behavior time sequences of the multiple single behavior sequences, including: using trigonometric functions to time-encode the multiple single behavior sequences to obtain the multiple single behavior time sequences.
  • the behavior prediction model performs modeling on each of the multiple single behavior time series at the time point to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series, including: the behavior prediction model performs intensity value on each of the multiple single behavior time series at the time point within a preset time period to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the behavior prediction model includes a behavior time sub-model
  • the multiple single behavior time series are input into the behavior prediction model
  • the behavior prediction model performs modeling of each single behavior time series in the multiple single behavior time series at the time point, and obtains the distribution of behaviors corresponding to the multiple single behavior time series in time, including: inputting the multiple single behavior time series into the behavior prediction model
  • the behavior time sub-model in the model models each single behavior time series in the multiple single behavior time series at the time point, and obtains the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the behavior time sub-model is trained by using the first loss value between the multiple single behavior time series and the distribution of behaviors corresponding to the multiple single behavior time series in time.
  • the distribution of behaviors corresponding to the multiple single behavior time series in time is obtained by modeling the multiple single behavior time series at the time point by the behavior time sub-model.
  • the behavior prediction model is trained based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss values between the multiple single behavior time series, including: training the behavior time sub-model in the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss values between the multiple single behavior time series.
  • the behavior prediction model is trained based on the first loss value, so that the behavior prediction model obtained can better predict the time and type of future behaviors based on the behavior sequence to be predicted.
  • the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model
  • the method also includes: inputting the multiple single behavior sequences into the behavior relationship sub-model to obtain a causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the causal relationship is used to characterize the relationship between the occurrence of the behavior of one single behavior sequence and the occurrence of the behavior of another single behavior sequence in every two single behavior sequences; inputting the causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model, and the prediction sub-model performs sequence prediction based on the causal relationship and the distribution to obtain a predicted behavior sequence; based on the first loss value, and the second loss value between the predicted behavior sequence and the user's behavior sequence, the behavior relationship sub-model and the prediction sub-model are trained.
  • the behavior prediction model when the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model, multiple single behavior sequences can be input into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences; the causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the temporal distribution of behaviors corresponding to the multiple single behavior time series are input into the prediction sub-model, and the predicted user sequence is obtained by the prediction sub-model; thus, the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence.
  • the behavior prediction model not only focuses on the time point information of the user's behavior sequence, but also focuses on the causal relationship between multiple behaviors in the user's behavior sequence, so that the user's behavior sequence can be modeled more accurately.
  • the multiple single behavior sequences are input into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences, including: obtaining the causal relationship between every two single behavior sequences in the multiple single behavior sequences based on the sum of the covariance values of every two single behavior sequences in the multiple single behaviors at all time points.
  • obtaining the causal relationship between every two single-behavior sequences in the above multiple single-behavior sequences is to determine the Granger causal relationship between every two single-behavior sequences.
  • the behavior relationship sub-model and the prediction sub-model are trained, including: determining a third loss value between the predicted label and the real label of the user's behavior sequence, the real label is used to indicate the real information of the user, and the predicted label is used to indicate the information of the user predicted by the prediction sub-model; averaging the second loss value and the third loss value to obtain a fourth loss value; and training the behavior relationship sub-model and the prediction sub-model based on the first loss value and the fourth loss value.
  • the above technical solution compared with training the behavior relationship sub-model and the prediction sub-model based only on the first loss value and the second loss value.
  • the above technical solution also introduces a third loss value between the predicted label and the true label; the average value between the third loss value and the second loss value is determined as the fourth loss value, and the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the fourth loss value. Since the scheme introduces the third loss value, it is equivalent to introducing more information to train the behavior relationship sub-model and the prediction sub-model, so that the obtained behavior relationship sub-model and the prediction sub-model are more accurate.
  • the method before determining the third loss value between the predicted label and the true label of the user's behavior sequence, the method also includes: inputting the causal relationship between every two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model to obtain the predicted label corresponding to the user's behavior sequence.
  • the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence, including: obtaining a fifth loss value based on the first loss value and the second loss value; training the behavior relationship sub-model based on the fifth loss value, and training the prediction sub-model based on the second loss value.
  • a fifth loss value is obtained based on the first loss value and the second loss value, including: determining the first numerical value based on a matrix composed of the second-order partial derivatives of each weight in the behavior relationship sub-model and a vector composed of each weight in the behavior relationship sub-model; and obtaining the fifth loss value based on the first numerical value, the first loss value and the second loss value.
  • the method also includes: inputting the behavior sequence to be predicted into the behavior prediction model, and obtaining the time point and type corresponding to at least one future behavior of the behavior sequence to be predicted predicted by the behavior prediction model.
  • one or more embodiments of the present specification propose a method for training a behavior prediction model, which includes splitting a user's behavior sequence to obtain multiple single behavior sequences; time-coding all time points in the multiple single behavior sequences to obtain multiple single behavior time series of the multiple single behavior sequences; inputting the multiple single behavior time series into the behavior prediction model to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series predicted by the behavior prediction model; thereby training the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior prediction model focuses on the information of each time point of multiple single behaviors in the user's behavior sequence and models each time point. This avoids the fuzzification of the original time information of each behavior in the multiple behavior sequences in the related art, thereby modeling a more accurate behavior prediction model, which can be used to predict more accurate time points and types of future behaviors.
  • the behavior prediction model includes a behavior time sub-model, a behavior relationship sub-model, and a prediction sub-model
  • multiple single behavior sequences can be input into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences; the causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the temporal distribution of the behaviors corresponding to the multiple single behavior time series are input into the prediction sub-model, and the predicted user sequence is obtained by the prediction sub-model; thus, based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence, the behavior relationship sub-model and the prediction sub-model are trained.
  • the behavior prediction model not only focuses on the time point information of the user's behavior sequence, but also focuses on the causal relationship between multiple behaviors in the user's behavior sequence, so that the user's behavior sequence can be modeled more accurately.
  • the obtained behavior prediction model can be used to predict the time point and type corresponding to at least one future behavior of the behavior sequence to be predicted.
  • a device for training a behavior prediction model comprising: a determination module, for splitting a user's behavior sequence to obtain multiple single behavior sequences, each single behavior sequence corresponding to a behavior, and for recording the correspondence between the behavior and the time point; the determination module, for time encoding the multiple single behavior sequences to obtain multiple single behavior time series of the multiple single behavior sequences; the determination module, for inputting the multiple single behavior time series into a behavior prediction model, and having the behavior prediction model perform modeling of the time point for each single behavior time series in the multiple single behavior time series to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series; a training module, for training the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and a first loss value between the multiple single behavior time series, the behavior prediction model being used to predict the time point and type of at least one future behavior of the user based on the user's behavior sequence.
  • the determining module is used to determine the The user's behavior sequence is split to obtain the multiple single behavior sequences, and the user's behavior sequence corresponds to the types of multiple behaviors.
  • the determination module is further used to perform time encoding on the multiple single behavior sequences using trigonometric functions to obtain the multiple single behavior time sequences.
  • the determination module is also used to use the behavior prediction model to perform an intensity value of the time point within a preset time period on each of the multiple single behavior time series, to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the behavior prediction model includes a behavior time sub-model
  • the determination module is specifically used to input the multiple single behavior time series into the behavior time sub-model in the behavior prediction model, and the behavior time sub-model models each single behavior time series in the multiple single behavior time series at the time point to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the training module is used to train the behavior time sub-model in the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model
  • the determination module is further used to input the multiple single behavior sequences into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the causal relationship is used to characterize the relationship between the occurrence of the behavior of one single behavior sequence and the occurrence of the behavior of another single behavior sequence in every two single behavior sequences;
  • the causal relationship between every two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series are input into the prediction sub-model, and the prediction sub-model performs sequence prediction based on the causal relationship and the distribution to obtain a predicted behavior sequence;
  • the training module is also used to train the behavior relationship sub-model and the prediction sub-model based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence.
  • the determination module is also used to obtain the causal relationship between each two single behavior sequences in the multiple single behavior sequences based on the sum of the covariance values of each two single behavior sequences in the multiple single behavior sequences at all time points.
  • the training module is further used to determine a third loss value between the predicted label and the true label of the user's behavior sequence, the true label being used to indicate the The user's real information
  • the prediction label is used to indicate the information of the user predicted by the prediction sub-model
  • the fourth loss value is obtained by averaging the second loss value and the third loss value
  • the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the fourth loss value.
  • the determination module is also used to input the causal relationship between every two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model to obtain the prediction label corresponding to the user's behavior sequence.
  • the training module is also used to obtain a fifth loss value based on the first loss value and the second loss value; train the behavior relationship sub-model based on the fifth loss value, and train the prediction sub-model based on the second loss value.
  • the determination module is also used to determine the first numerical value based on the matrix composed of the second-order partial derivatives of each weight in the behavior relationship sub-model and the vector composed of each weight in the behavior relationship sub-model; and obtain the fifth loss value based on the first numerical value, the first loss value and the second loss value.
  • the determination module is also used to input the behavior sequence to be predicted into the behavior prediction model to obtain the time point and type corresponding to at least one future behavior of the behavior sequence to be predicted predicted by the behavior prediction model.
  • an electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the electronic device executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect.
  • a computer-readable storage medium which stores instructions.
  • the instructions When the instructions are executed on a computer or a processor, the computer or the processor executes the method in the first aspect or any possible implementation of the first aspect.
  • a computer program product comprising instructions is provided.
  • the computer program product runs on the computer or processor, the computer or processor executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect.
  • FIG1 is a schematic diagram of time aggregation of multiple behavior sequences provided by one or more embodiments of this specification.
  • FIG. 2 is a schematic flow chart of a method for training a behavior prediction model provided by one or more embodiments of this specification. Process map.
  • FIG3 is a schematic diagram of splitting a user's behavior sequence provided by one or more embodiments of this specification.
  • FIG. 4 is a schematic diagram of the structure of a behavior prediction model provided by one or more embodiments of this specification.
  • FIG5 is a schematic diagram of the structure of a device for training a behavior prediction model provided by one or more embodiments of this specification.
  • FIG. 6 is a schematic diagram of the structure of an electronic device provided by one or more embodiments of this specification.
  • FIG1 is a schematic diagram of time aggregation of multiple behavior sequences provided by one or more embodiments of this specification.
  • a multi-behavior sequence corresponds to multiple types of behaviors, and each type of behavior corresponds to a time point.
  • each type of behavior corresponds to a time point.
  • the time point when the time point is 0 minutes, it is a pentagonal type of behavior; when the time point is 1 minute, it is a circular type of behavior; when the time point is 2.2 minutes, it is a triangular type of behavior; when the time point is 3 minutes, it is a triangular type of behavior; when the time point is 3.5 minutes, it is a pentagonal type of behavior; and so on, until the end, when the time point is 27 minutes, it is a rectangular type of behavior.
  • the pentagonal type is an online purchase type
  • the circular type is a transfer type
  • the triangular type is a fund subscription type
  • the rectangular type is a chat type.
  • the multiple behavior sequences are divided equally according to the preset time intervals to obtain multiple behavior sequences at uniform time intervals.
  • Time aggregation is performed in the behavior sequences at each time interval of the multiple behavior sequences at uniform time intervals, and the number of occurrences of each behavior at each time interval is obtained.
  • the multiple behavior sequences are divided equally and the time aggregation is performed to obtain: 0 to 4 minutes: 2 online purchase behaviors, 1 transfer behavior, 2 fund subscription behaviors, and 0 chat behaviors; 4 to 8 minutes: 0 online purchase behaviors, 1 transfer behavior, 1 fund subscription behavior, and 2 chat behaviors; 8 to 12 minutes: 2 online purchase behaviors, 1 transfer behavior, 0 fund subscription behaviors, and 1 chat behavior; 12 to 16 minutes: 1 online purchase behavior, 1 transfer behavior, 1 fund subscription behavior, and 1 chat behavior; 16 to 20 minutes: 1 online purchase behavior, 2 transfer behaviors, 0 fund subscription behaviors, and 0 chat behaviors; 20 to 24 minutes: 2 online purchase behaviors, 1 transfer behavior, 0 fund subscription behaviors, and 0 chat behaviors; Behavior: 0 times, transfer behavior: 0 times, fund subscription behavior: 2 times and chat behavior: 1 time; and 24-28 minutes: online purchase behavior: 1 time, transfer behavior: 0 times, fund subscription behavior: 2 times and chat behavior: 2 times.
  • the prediction is the time period of a certain type of behavior in the future, not the time point. Therefore, this prediction method is inaccurate in predicting the time information of future behavior.
  • FIG2 is a schematic flowchart of a method for training a behavior prediction model provided by one or more embodiments of this specification.
  • a method for training a behavior prediction model provided in one or more embodiments of this specification can be applied to any electronic device, which may be a computer terminal (a display device with data processing function, a mobile phone terminal) or a server, etc.
  • the determination process of the behavior prediction model of one or more embodiments of the present specification is implemented based on the recurrent neural networks (RNN) in deep learning, which can specifically be long short-term memory (LSTM) units and gated recurrent units (GRU), etc.
  • RNN recurrent neural networks
  • LSTM long short-term memory
  • GRU gated recurrent units
  • the method 200 includes steps 202 to 208 .
  • the electronic device splits the user's behavior sequence to obtain multiple single behavior sequences, each of which corresponds to a behavior and is used to record the corresponding relationship between the behavior and a time point.
  • the user's behavior sequence in the above solution refers to the multi-behavior sequence shown in FIG. 1
  • the user's behavior sequence like the multi-behavior sequence, corresponds to multiple types of behaviors.
  • each type of behavior in the user's behavior sequence has a corresponding time point, that is, the specific time when a certain type of behavior occurs.
  • step 202 includes: the electronic device splits the behavior sequence of the user according to the type of behavior to obtain the multiple single behavior sequences, and the behavior sequence of the user corresponds to the types of multiple behaviors.
  • the user's behavior sequence since the user's behavior sequence corresponds to multiple behavior types, the user's behavior sequence can be split according to the behavior type into multiple single behavior sequences, each of which corresponds to a behavior type.
  • the multiple single behavior sequences can be time-coded to obtain multiple single behavior time series of the multiple single behavior sequences, and the multiple single behavior time series can be used to train the behavior prediction model.
  • FIG3 is a schematic diagram of splitting a user's behavior sequence provided by one or more embodiments of this specification.
  • the multiple behavior sequences shown in FIG1 are split into multiple single behavior sequences.
  • the types of the multiple behaviors corresponding to the user's behavior sequence are specifically the pentagonal type, i.e., the online purchase type, the circular type, i.e., the transfer type, the triangular type, i.e., the fund subscription type, and the rectangular type, i.e., the chat type.
  • the user's behavior sequence is split into multiple single behavior sequences according to the type of behavior, namely: the sequence with sequence number 2, the sequence with sequence number 3, the sequence with sequence number 4, and the sequence with sequence number 5, that is, the user's behavior sequence is split into a behavior sequence of the online purchase type, a behavior sequence of the transfer type, and a behavior sequence of the fund subscription type. and chat-type behavior sequences.
  • the electronic device performs time coding on the multiple single-behavior sequences to obtain multiple single-behavior time sequences of the multiple single-behavior sequences.
  • step 204 includes: the electronic device uses a trigonometric function to time-code the multiple single-behavior sequences to obtain the multiple single-behavior time sequences.
  • the electronic device uses the following trigonometric function formula (1) to time-code each single-behavior sequence in the multiple single-behavior sequences to obtain multiple single-behavior time sequences.
  • ⁇ T represents the mapping of a certain behavior sequence within a period of time
  • b 1 ,..., b d are phase difference parameters ⁇ 1 ,..., ⁇ d are frequency parameters
  • d is the dimension after mapping a certain behavior sequence using trigonometric functions
  • t- ⁇ represents a certain time period.
  • the electronic device inputs the multiple single behavior time series into a behavior prediction model, and the behavior prediction model models each single behavior time series in the multiple single behavior time series at the time point to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • step 206 includes: the behavior prediction model performs intensity value analysis at the time point within a preset time period on each of the multiple single behavior time series to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the behavior prediction model performs the intensity value of each single behavior time series in the preset time period at the time point to obtain the distribution of behaviors corresponding to the multiple single behavior time series in time. That is, the probability density distribution function of the behaviors corresponding to each single behavior time series in the multiple single behavior time series in a certain period of time expressed by the following formula (2) is obtained.
  • ⁇ i represents a certain period of time
  • p * ( ⁇ i ) is the probability density distribution function of the behavior corresponding to a single behavior time series in a certain period of time
  • the probability density distribution function of the behavior corresponding to a single behavior time series in a certain period of time is related to the historical behavior in the single behavior sequence
  • ⁇ * (t i-1 + ⁇ i ) is the intensity function (intensity value) of a behavior occurring in the time period from t i-1 to ⁇ i , that is, the probability of a behavior occurring
  • s represents a certain time point in the time period from t i-1 to ⁇ i .
  • step 206 includes: the electronic device inputs the multiple single behavior time series into the behavior time sub-model in the behavior prediction model, The behavior time sub-model models each single behavior time series in the multiple single behavior time series at the time point, and obtains the distribution of behaviors corresponding to the multiple single behavior time series in time.
  • step 206 is a more specific implementation method of step 206.
  • the behavior prediction model includes a behavior time sub-model
  • multiple single behavior time series can be input into the behavior time sub-model, and the behavior time sub-model models each single behavior time series in the multiple single behavior time series at the time point.
  • the electronic device trains the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior prediction model is used to predict the time point and type of at least one future behavior of the user based on the user's behavior sequence.
  • step 208 includes: the electronic device trains the behavior time sub-model in the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior time sub-model is trained by using the first loss value between the multiple single behavior time series and the distribution of behaviors corresponding to the multiple single behavior time series in time.
  • the distribution of behaviors corresponding to the multiple single behavior time series in time is obtained by modeling the multiple single behavior time series at the time point by the behavior time sub-model.
  • the method when the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model, the method also includes: the electronic device inputs the multiple single behavior sequences into the behavior relationship sub-model to obtain a causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the causal relationship is used to characterize the relationship between the occurrence of a behavior of one single behavior sequence and the occurrence of a behavior of another single behavior sequence in every two single behavior sequences; the electronic device inputs the causal relationship between every two single behavior sequences in the multiple single behavior sequences, and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model, and the prediction sub-model performs sequence prediction based on the causal relationship and the distribution to obtain a predicted behavior sequence; the electronic device trains the behavior relationship sub-model and the prediction sub-model based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence.
  • the behavior prediction model when the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model, multiple single behavior sequences can be input into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences; the causal relationship between every two single behavior sequences in the multiple single behavior sequences, as well as the temporal distribution of behaviors corresponding to the multiple single behavior time series are input into the prediction sub-model, and the predicted user sequence is obtained by the prediction sub-model; thereby, the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence.
  • the behavior prediction model not only pays attention to the time point information of the user's behavior sequence, but also pays attention to the causal relationship between multiple behaviors in the user's behavior sequence, so that it can To model the user's behavior sequence more accurately.
  • the electronic device inputs the multiple single behavior sequences into the behavior relationship sub-model to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences, including: the electronic device obtains the causal relationship between every two single behavior sequences in the multiple single behavior sequences based on the sum of the covariance values of every two single behavior sequences in the multiple single behavior sequences at all time points.
  • obtaining the causal relationship between every two single-behavior sequences in the above multiple single-behavior sequences specifically refers to determining the Granger causal relationship between every two single-behavior sequences.
  • the electronic device obtains the causal relationship between every two single-behavior sequences in the multiple single-behavior sequences according to the following Granger causality formula (3).
  • Xi are any two different single-behavior sequences in multiple single-behavior sequences; i is the i-th behavior in the single-behavior sequence; any two single-behavior sequences in multiple single-behavior sequences can form a set of sequences in k groups; represents the Granger causality between two single-behavior sequences in the kth group of sequences; represents the covariance of the ith behavior between two single-behavior sequences in the kth group of sequences.
  • the electronic device trains the behavior relationship sub-model and the prediction sub-model based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence, including: the electronic device determines a third loss value between the predicted label and the real label of the user's behavior sequence, the real label is used to indicate the real information of the user, and the predicted label is used to indicate the information of the user predicted by the prediction sub-model; the electronic device averages the second loss value and the third loss value to obtain a fourth loss value; the electronic device trains the behavior relationship sub-model and the prediction sub-model based on the first loss value and the fourth loss value.
  • the above technical solution compared with training the behavior relationship sub-model and the prediction sub-model based only on the first loss value and the second loss value.
  • the above technical solution also introduces a third loss value between the predicted label and the true label; the average value between the third loss value and the second loss value is determined as the fourth loss value, and the behavior relationship sub-model and the prediction sub-model are trained based on the first loss value and the fourth loss value. Since the scheme introduces the third loss value, it is equivalent to introducing more information to train the behavior relationship sub-model and the prediction sub-model, so that the obtained behavior relationship sub-model and the prediction sub-model are more accurate.
  • the method before the electronic device determines the third loss value between the predicted label and the true label of the user's behavior sequence, the method also includes: the electronic device inputs the causal relationship between every two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model to obtain the predicted label corresponding to the user's behavior sequence.
  • the true label of the user's behavior sequence may be the user's identity information and/or occupation information.
  • the acquisition of the user's identity information and/or occupational information requires the user's consent.
  • the electronic device is based on the first loss value, the predicted behavior sequence and the The second loss value between the user's behavior sequence is used to train the behavior relationship sub-model and the prediction sub-model, including: the electronic device obtains a fifth loss value based on the first loss value and the second loss value; the electronic device trains the behavior relationship sub-model based on the fifth loss value, and trains the prediction sub-model based on the second loss value.
  • the electronic device obtains a fifth loss value based on the first loss value and the second loss value, including: the electronic device determines the first numerical value based on a matrix formed by the second-order partial derivatives of each weight in the behavior relationship sub-model and a vector formed by each weight in the behavior relationship sub-model; the electronic device obtains the fifth loss value based on the first numerical value, the first loss value and the second loss value.
  • the electronic device obtains the first value according to the following Pareto optimal solution formula (4).
  • H is a matrix composed of the second-order partial derivatives of each weight in the behavior relationship sub-model, specifically a Hessian matrix
  • ⁇ is the first value
  • f is the vector value of the behavior relationship sub-model
  • ⁇ is the vector value composed of each weight in the behavior relationship sub-model.
  • the method further includes: the electronic device inputs the behavior sequence to be predicted into the behavior prediction model, and obtains the time point and type corresponding to at least one future behavior of the behavior sequence to be predicted predicted by the behavior prediction model.
  • the specific process is to use the trained behavior prediction model.
  • the trained behavior prediction model can be used to predict any behavior sequence to obtain the time point and type corresponding to at least one future behavior of the behavior sequence.
  • the electronic device splits the user sequence to be predicted according to the type of behavior to obtain multiple single-behavior user sequences; the electronic device time-encodes the multiple single-behavior user sequences to obtain multiple single-behavior user time sequences, and the behavior sequence to be detected includes multiple single-behavior user sequences and multiple single-behavior user time sequences.
  • the electronic device time-encodes the multiple single-behavior user sequences to obtain the multiple single-behavior user time sequences, including: the electronic device time-encodes the multiple single-behavior user sequences using trigonometric functions to obtain the multiple single-behavior user time sequences.
  • the electronic device inputs multiple single-behavior user time series into the behavior time sub-model to obtain the temporal distribution of the behaviors corresponding to the multiple single-behavior user time series; the electronic device inputs multiple single-behavior user sequences into the behavior relationship sub-model to obtain the causal relationship between every two single-behavior user sequences in the multiple single-behavior user sequences; the electronic device inputs the distribution and the causal relationship into the prediction sub-model to obtain the time point and type corresponding to at least one future behavior of the behavior sequence to be predicted.
  • FIG. 4 is a schematic diagram of the structure of a behavior prediction model provided by one or more embodiments of this specification.
  • the process of obtaining the behavior prediction model is described in detail.
  • the user's behavior The sequence is split to obtain multiple single behavior sequences; the multiple single behavior sequences are time-encoded to obtain multiple single behavior time series; the time points in the multiple single behavior time series are modeled by the behavior time sub-model to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series; the multiple single behavior sequences are input into the behavior relationship sub-model, and the causal relationship between every two single behavior sequences in the multiple single behavior sequences is obtained by the behavior relationship sub-model; the prediction sub-model can predict the predicted behavior sequence and the prediction label according to the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the causal relationship between every two single behavior sequences in the multiple single behavior sequences.
  • the behavior time sub-model, behavior relationship sub-model and prediction sub-model can be trained based on the first loss value and the second loss value; the fourth loss value can also be obtained based on the average of the second loss value and the third loss value, and the behavior time sub-model, behavior relationship sub-model and prediction sub-model can be trained based on the first loss value and the fourth loss value.
  • FIG5 is a schematic diagram of the structure of a device for training a behavior prediction model provided by one or more embodiments of this specification.
  • the device 500 includes: a determination module 501, which is used to split the user's behavior sequence to obtain multiple single behavior sequences, each single behavior sequence corresponds to a behavior, and is used to record the correspondence between the behavior and the time point; the determination module 501 is also used to time-code the multiple single behavior sequences to obtain multiple single behavior time series of the multiple single behavior sequences; the determination module 501 is also used to input the multiple single behavior time series into a behavior prediction model, and the behavior prediction model performs modeling on each single behavior time series in the multiple single behavior time series at the time point to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series; a training module 502 is used to train the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series, and the behavior prediction model is used to predict the time point and type of at least one future behavior of the user based on the user's behavior sequence.
  • a determination module 501 which is used to split the user's behavior
  • the determination module 501 is used to split the behavior sequence of the user according to the type of behavior to obtain the multiple single behavior sequences, and the behavior sequence of the user corresponds to multiple behavior types.
  • the determination module 501 is further configured to perform time coding on the multiple single behavior sequences using a trigonometric function to obtain the multiple single behavior time sequences.
  • the determination module 501 is also used to perform, by the behavior prediction model, an intensity value of each single behavior time series in the multiple single behavior time series at the time point within a preset time period to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the behavior prediction model includes a behavior time sub-model
  • the determination module 501 is further used to A single behavior time series is input into the behavior time sub-model in the behavior prediction model, and the behavior time sub-model models each single behavior time series in the multiple single behavior time series at the time point to obtain the temporal distribution of the behaviors corresponding to the multiple single behavior time series.
  • the training module 502 is used to train the behavior time sub-model in the behavior prediction model based on the temporal distribution of the behaviors corresponding to the multiple single behavior time series and the first loss value between the multiple single behavior time series.
  • the behavior prediction model also includes a behavior relationship sub-model and a prediction sub-model.
  • the determination module 501 is further used to input the multiple single behavior sequences into the behavior relationship sub-model to obtain the causal relationship between the occurrence of the behaviors of each two single behavior sequences in the multiple single behavior sequences, and the causal relationship is used to characterize the relationship between the occurrence of the behaviors of one single behavior sequence and the other single behavior sequence in each two single behavior sequences; the causal relationship between each two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series are input into the prediction sub-model, and the prediction sub-model performs sequence prediction based on the causal relationship and the distribution to obtain a predicted behavior sequence; the training module 502 is further used to train the behavior relationship sub-model and the prediction sub-model based on the first loss value and the second loss value between the predicted behavior sequence and the user's behavior sequence.
  • the determination module 501 is further configured to obtain the causal relationship between every two single behavior sequences in the multiple single behavior sequences according to the sum of covariance values of every two single behavior sequences in the multiple single behavior sequences at all time points.
  • the training module 502 is also used to determine a third loss value between the predicted label and the real label of the user's behavior sequence, the real label is used to indicate the real information of the user, and the predicted label is used to indicate the information of the user predicted by the prediction sub-model; averaging the second loss value and the third loss value to obtain a fourth loss value; and training the behavior relationship sub-model and the prediction sub-model based on the first loss value and the fourth loss value.
  • the determination module 501 is also used to input the causal relationship between every two single behavior sequences in the multiple single behavior sequences and the temporal distribution of the behaviors corresponding to the multiple single behavior time series into the prediction sub-model to obtain the prediction label corresponding to the user's behavior sequence.
  • the training module 502 is further used to obtain a fifth loss value based on the first loss value and the second loss value; train the behavior relationship sub-model based on the fifth loss value, and train the prediction sub-model based on the second loss value.
  • the determination module 501 is also used to determine the first numerical value based on the matrix composed of the second-order partial derivatives of each weight in the behavior relationship sub-model and the vector composed of each weight in the behavior relationship sub-model; and obtain the fifth loss value based on the first numerical value, the first loss value and the second loss value.
  • the determination module 501 is further configured to input the behavior sequence to be predicted into the behavior prediction model to obtain a time point and a type corresponding to at least one future behavior of the behavior sequence to be predicted predicted by the behavior prediction model.
  • FIG. 6 is a schematic diagram of the structure of an electronic device provided by one or more embodiments of this specification.
  • the electronic device 600 includes: a memory 601, a processor 602, and a computer program 603 stored in the memory 601 and running on the processor 602, wherein when the processor 602 executes the computer program 603, the electronic device can execute any one of the methods for training a behavior prediction model described above.
  • One or more embodiments of this specification may divide the functional modules of the electronic device according to the above method examples.
  • each functional module may be corresponded to, or two or more functions may be integrated into one processing module.
  • the above integrated module may be implemented in the form of hardware. It should be noted that the division of modules in one or more embodiments of this specification is schematic and is only a logical function division. There may be other division methods in actual implementation.
  • the electronic device may include: a determination module and a training module, etc. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, which will not be repeated here.
  • the electronic device provided in one or more embodiments of this specification is used to execute the above-mentioned method of training a behavior prediction model, and thus can achieve the same effect as the above-mentioned implementation method.
  • the electronic device may include a processing module and a storage module.
  • the processing module may be used to control and manage the actions of the electronic device.
  • the storage module may be used to support the electronic device in executing mutual program codes and data.
  • the processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits disclosed in conjunction with one or more embodiments of the present specification.
  • the processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc.
  • the storage module may be a memory.
  • the electronic device provided in one or more embodiments of the present specification may specifically be a chip, a component or a module, and the electronic device may include a connected processor and a memory; wherein the memory is used to store instructions, and when the electronic device is running, the processor may call and execute the instructions so that the chip executes any one of the methods for training a behavior prediction model described above.
  • One or more embodiments of the present specification provide a computer-readable storage medium having instructions stored therein.
  • the instructions When the instructions are executed on a computer or a processor, the computer or the processor executes any one of the methods for training a behavior prediction model described above.
  • One or more embodiments of the present specification also provide a computer program product comprising instructions, which, when executed on a computer or a processor, enables the computer or the processor to execute the above-mentioned related steps to implement any of the methods for training a behavior prediction model described above.
  • the electronic device, computer-readable storage medium, computer program product or chip containing instructions provided in one or more embodiments of this specification are used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above and will not be repeated here.
  • the disclosed devices and methods can be implemented in other ways.
  • the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.
  • Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本说明书实施例提供了一种训练行为预测模型的方法及装置。本说明书的实施例将包括多种行为的类型的序列(用户的行为序列)按照行为的类型进行拆分，获取多个单行为序列；对多个单行为序列中的每个单行为序列的所有时间点进行时间编码，获取多个单行为序列对应的多个单行为时间序列；利用行为预测模型对多个单行为时间序列中的每个单行为时间序列进行建模，关注多个单行为时间序列中的每个单行为时间序列的所有时间点，以获取行为预测模型预测的多个单行为时间序列对应的行为在时间上的分布情况；基于多个单行为时间序列对应的行为在时间上的分布情况与多个单行为时间序列之间的损失值，来训练行为预测模型。

Description

一种训练行为预测模型的方法及装置 技术领域
本说明书一个或多个实施例涉及计算机技术领域，并且更具体地，涉及计算机技术领域中的一种训练行为预测模型的方法及装置。
背景技术
在日常生产或生活中，用户往往会执行各种各样的行为。通过数据分析技术，人们可以对用户的历史行为进行分析处理，以预测用户未来行为发生的时间。例如，可以将某一用户曾经点击过的商品，按照点击的时间顺序组成点击行为的时间序列。通过对该点击行为的时间序列进行分析，可以预测用户在未来点击某一商品的时间等。
针对某一用户的非均匀、连续时间域上的多行为序列，传统的行为序列建模方法是将多行为序列离散化，转化为多段均匀时间间隔上的问题，对每段时间间隔内的多个行为进行时间聚合并求得相应的特征。上述方式在对多行为序列进行处理的过程中，将多行为序列中各个行为的原始时间信息进行了模糊化，这样建模得到的多行为序列与原行为序列(未处理的多行为序列)存在较大差异。从而导致利用该建模的多行为序列预测的未来行为的时间和类型不准确。因此，需要提供更加准确的对多行为序列进行建模的方法。
发明内容
本说明书一个或多个实施例提供了一种训练行为预测模型的方法及装置，该方法能够准确地对用户的多行为序列中各个行为的时间点信息进行建模。
第一方面，提供了一种训练行为预测模型的方法，该方法包括：对用户的行为序列进行拆分，得到多个单行为序列，各个单行为序列对应于一种行为，用于记录该行为和时间点的对应关系；对该多个单行为序列进行时间编码，得到该多个单行为序列的多个单行为时间序列；将该多个单行为时间序列输入行为预测模型，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况；基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型进行训练，该行为预测模型用于基于该用户的行为序列预测该用户未来至少一种行为的时间点和类型。
上述技术方案中，本说明书一个或多个实施例对用户的行为序列进行拆分，得到多 个单行为序列；对多个单行为序列中的所有时间点进行时间编码，得到多个单行为序列的多个单行为时间序列；将多个单行为时间序列输入到行为预测模型，得到行为预测模型预测的多个单行为时间序列对应的行为在时间上的分布情况；从而可以基于多个单行为时间序列对应的行为在时间上的分布情况与多个单行为时间序列之间的第一损失值，对行为预测模型进行训练。也就是说，行为预测模型关注用户的行为序列中多个单行为的各个时间点，对各个时间点进行建模。从而可以建模得到更加准确的行为预测模型，利用该行为预测模型可以预测更加准确的未来行为的时间点和类型。
结合第一方面，在某些可能的实现方式中，对用户的行为序列进行拆分，得到多个单行为序列，包括：根据行为的类型对该用户的行为序列进行拆分，得到该多个单行为序列，该用户的行为序列对应于多种行为的类型。
上述技术方案中，由于用户的行为序列对应于多种行为的类型，因此可以按照行为的类型将用户的行为序列进行拆分，拆分成多个单行为序列，多个单行为中各个单行为对应于一种行为的类型。从而可以对多个单行为序列进行时间编码得到多个单行为序列的多个单行为时间序列，从而多个单行为时间序列可以用于对行为预测模型的训练。
结合第一方面和上述实现方式，在某些可能的实现方式中，对该多个单行为序列进行时间编码，得到该多个单行为序列的多个单行为时间序列，包括：利用三角函数对该多个单行为序列进行时间编码，得到该多个单行为时间序列。
上述技术方案中，利用三角函数对多个单行为序列进行时间编码，可以得到多个单行为序列中各个单行为序列在时间上所有时间特征，得到多个单行为序列的多个单行为时间序列。
结合第一方面和上述实现方式，在某些可能的实现方式中，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况，包括：由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行预设时间段内该时间点的强度值，得到该多个单行为时间序列对应的行为在时间上的分布情况。
应理解，上述技术方案是为了得到多个单行为时间序列对应的行为在时间上的分布情况，实则是得到多个单行为时间序列对应的行为在某时间段内的概率密度分布函数。
结合第一方面和上述实现方式，在某些可能的实现方式中，该行为预测模型包括行为时间子模型，将该多个单行为时间序列输入行为预测模型，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况，包括：将该多个单行为时间序列输入该行为预测 模型中的该行为时间子模型，由该行为时间子模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况。
上述技术方案中，是在行为预测模型包括行为时间子模型的情况下，具体利用该多个单行为时间序列以及该多个单行为时间序列对应的行为在时间上的分布情况之间的第一损失值对该行为时间子模型进行训练的方案。其中，该多个单行为时间序列对应的行为在时间上的分布情况是由该行为时间子模型对该多个单行为时间序列进行该时间点的建模得到的。
结合第一方面和上述实现方式，在某些可能的实现方式中，基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型进行训练，包括：基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型中的该行为时间子模型进行训练。
上述技术方案中，基于第一损失值对行为预测模型进行训练，这样得到的行为预测模型可以更好地基于待预测的行为序列预测未来行为的时间和类型。
结合第一方面和上述实现方式，在某些可能的实现方式中，该行为预测模型还包括行为关系子模型和预测子模型，该方法还包括：将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列之间的因果关系，该因果关系用于表征在每两个单行为序列中一个单行为序列的行为的发生与另一个单行为序列的行为的发生之间的关系；将该多个单行为序列中每两个单行为序列之间的因果关系，以及该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由该预测子模型基于该因果关系和该分布情况进行序列预测，得到预测的行为序列；基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练。
上述技术方案中，在行为预测模型还包括行为关系子模型和预测子模型时，可以将多个单行为序列输入行为关系子模型，得到多个单行为序列中每两个单行为序列之间的因果关系；将多个单行为序列中每两个单行为序列之间的因果关系，以及多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由预测子模型得到预测的用户序列；从而基于第一损失值，以及预测的行为序列和用户的行为序列之间的第二损失值，对行为关系子模型和预测子模型进行训练。也就是说，行为预测模型不仅关注用户的行为序列的时间点信息，也关注用户的行为序列中多个行为之间的因果关系，从而可以对用户的行为序列进行更加准确地建模。
结合第一方面和上述实现方式，在某些可能的实现方式中，将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列之间的因果关系，包括：根据该多个单行为中每两个单行为序列在所有时间点上的协方差值之和，得到该多个单行为序列中每两个单行为序列之间的因果关系。
应理解，上述得到多个单行为序列中每两个单行为序列之间的因果关系是确定每两个单行为序列之间的格兰杰因果关系。
结合第一方面和上述实现方式，在某些可能的实现方式中，基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练，包括：确定预测标签与该用户的行为序列的真实标签之间的第三损失值，该真实标签用于指示该用户的真实信息，该预测标签用于指示该预测子模型预测的该用户的信息；对该第二损失值和该第三损失值求取平均值得到第四损失值；基于该第一损失值和该第四损失值对该行为关系子模型和该预测子模型进行训练。
上述技术方案中，相对于只基于第一损失值和第二损失值来训练行为关系子模型和预测子模型。上述技术方案还引入了预测标签与真实标签之间的第三损失值；将第三损失值和第二损失值之间的平均值确定为第四损失值，基于第一损失值和第四损失值来训练行为关系子模型和预测子模型。由于该方案引入了第三损失值，相当于引入更多的信息训练行为关系子模型和预测子模型，这样得到的行为关系子模型和预测子模型更加准确。
结合第一方面和上述实现方式，在某些可能的实现方式中，确定预测标签与该用户的行为序列的真实标签之间的第三损失值之前，该方法还包括：将该多个单行为序列中每两个单行为序列之间的因果关系与该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，得到该用户的行为序列对应的该预测标签。
结合第一方面和上述实现方式，在某些可能的实现方式中，基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练，包括：根据该第一损失值和该第二损失值，得到第五损失值；基于该第五损失值训练该行为关系子模型，以及基于该第二损失值训练该预测子模型。
结合第一方面和上述实现方式，在某些可能的实现方式中，根据该第一损失值和该第二损失值，得到第五损失值，包括：根据该行为关系子模型中的各个权重的二阶偏导构成的矩阵和该行为关系子模型中的各个权重构成的向量，确定第一数值；根据该第一数值、该第一损失值和该第二损失值得到该第五损失值。
应理解，上述确定第一数值的过程是利用帕累托最优解原理来实现的。
结合第一方面和上述实现方式，在某些可能的实现方式中，该方法还包括：将待预测的行为序列输入该行为预测模型，得到该行为预测模型预测的该待预测的行为序列的未来至少一个行为对应的时间点和类型。
综上，本说明书一个或多个实施例提出了一种训练行为预测模型的方法，对用户的行为序列进行拆分，得到多个单行为序列；对多个单行为序列中的所有时间点进行时间编码，得到多个单行为序列的多个单行为时间序列；将多个单行为时间序列输入到行为预测模型，得到行为预测模型预测的多个单行为时间序列对应的行为在时间上的分布情况；从而可以基于多个单行为时间序列对应的行为在时间上的分布情况与多个单行为时间序列之间的第一损失值，对行为预测模型进行训练。也就是说，行为预测模型关注了用户的行为序列中多个单行为的各个时间点信息，对各个时间点进行建模。而避免相关技术中对多行为序列中各个行为的原始时间信息进行模糊化，从而可以建模得到更加准确的行为预测模型，利用该行为预测模型可以预测更加准确的未来行为的时间点和类型。
此外，在行为预测模型包括行为时间子模型、行为关系子模型和预测子模型时，可以将多个单行为序列输入行为关系子模型，得到多个单行为序列中每两个单行为序列之间的因果关系；将多个单行为序列中每两个单行为序列之间的因果关系，以及多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由预测子模型得到预测的用户序列；从而基于第一损失值，以及预测的行为序列和用户的行为序列之间的第二损失值，对行为关系子模型和预测子模型进行训练。也就是说，行为预测模型不仅关注用户的行为序列的时间点信息，也关注用户的行为序列中多个行为之间的因果关系，从而可以对用户的行为序列进行更加准确地建模。
最后，可以利用得到的行为预测模型预测待预测的行为序列的未来至少一个行为对应的时间点和类型。
第二方面，提供了一种训练行为预测模型的装置，该装置包括：确定模块，用于对用户的行为序列进行拆分，得到多个单行为序列，各个单行为序列对应于一种行为，用于记录该行为和时间点的对应关系；该确定模块，还用于对该多个单行为序列进行时间编码，得到该多个单行为序列的多个单行为时间序列；该确定模块，还用于将该多个单行为时间序列输入行为预测模型，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况；训练模块，用于基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型进行训练，该行为预测模型用于基于该用户的行为序列预测该用户未来至少一种行为的时间点和类型。
结合第二方面，在某些可能的实现方式中，该确定模块，用于根据行为的类型对该 用户的行为序列进行拆分，得到该多个单行为序列，该用户的行为序列对应于多种行为的类型。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于利用三角函数对该多个单行为序列进行时间编码，得到该多个单行为时间序列。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行预设时间段内该时间点的强度值，得到该多个单行为时间序列对应的行为在时间上的分布情况。
应理解，上述方案是为了得到多个单行为时间序列对应的行为在时间上的分布情况，实则是得到多个单行为时间序列对应的行为在某时间段内的概率密度分布函数。
结合第二方面和上述实现方式，在某些可能的实现方式中，该行为预测模型包括行为时间子模型，该确定模块，具体还用于将该多个单行为时间序列输入该行为预测模型中的该行为时间子模型，由该行为时间子模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况。
结合第二方面和上述实现方式，在某些可能的实现方式中，该训练模块，用于基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型中的该行为时间子模型进行训练。
结合第二方面和上述实现方式，在某些可能的实现方式中，该行为预测模型还包括行为关系子模型和预测子模型，该确定模块，还用于将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列之间的因果关系，该因果关系用于表征在每两个单行为序列中一个单行为序列的行为的发生与另一个单行为序列的行为的发生之间的关系；将该多个单行为序列中每两个单行为序列之间的因果关系，以及该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由该预测子模型基于该因果关系和该分布情况进行序列预测，得到预测的行为序列；该训练模块，还用于基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于根据该多个单行为中每两个单行为序列在所有时间点上的协方差值之和，得到该多个单行为序列中每两个单行为序列之间的因果关系。
结合第二方面和上述实现方式，在某些可能的实现方式中，该训练模块，还用于确定预测标签与该用户的行为序列的真实标签之间的第三损失值，该真实标签用于指示该 用户的真实信息，该预测标签用于指示该预测子模型预测的该用户的信息；对该第二损失值和该第三损失值求取平均值得到第四损失值；基于该第一损失值和该第四损失值对该行为关系子模型和该预测子模型进行训练。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于将该多个单行为序列中每两个单行为序列之间的因果关系与该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，得到该用户的行为序列对应的该预测标签。
结合第二方面和上述实现方式，在某些可能的实现方式中，该训练模块，还用于根据该第一损失值和该第二损失值，得到第五损失值；基于该第五损失值训练该行为关系子模型，以及基于该第二损失值训练该预测子模型。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于根据该行为关系子模型中的各个权重的二阶偏导构成的矩阵和该行为关系子模型中的各个权重构成的向量，确定第一数值；根据该第一数值、该第一损失值和该第二损失值得到该第五损失值。
应理解，上述确定第一数值的过程是利用帕累托最优解原理来实现的。
结合第二方面和上述实现方式，在某些可能的实现方式中，该确定模块，还用于将待预测的行为序列输入该行为预测模型，得到该行为预测模型预测的该待预测的行为序列的未来至少一个行为对应的时间点和类型。
第三方面，提供一种电子设备，包括存储器、处理器以及存储在该存储器中并在该处理器上运行的计算机程序，其中，该处理器执行该计算机程序时，使得该电子设备执行上述第一方面或第一方面任意一种可能的实现方式中的方法。
第四方面，提供了一种计算机可读存储介质，该计算机可读存储介质中存储有指令，当该指令在计算机或处理器上运行时，使得该计算机或处理器执行上述第一方面或第一方面任意一种可能的实现方式中的方法。
第五方面，提供了一种包含指令的计算机程序产品，当该计算机程序产品在该计算机或处理器上运行时，使得该计算机或处理器执行上述第一方面或第一方面任意一种可能的实现方式中的方法。
附图说明
图1是本说明书一个或多个实施例提供的一种对多行为序列进行时间聚合的示意图。
图2是本说明书一个或多个实施例提供的一种训练行为预测模型的方法的示意性流 程图。
图3是本说明书一个或多个实施例提供的一种对用户的行为序列进行拆分的示意图。
图4是本说明书一个或多个实施例提供的一种行为预测模型的结构示意图。
图5是本说明书一个或多个实施例提供的一种训练行为预测模型的装置的结构示意图。
图6是本说明书一个或多个实施例提供的一种电子设备的结构示意图。
具体实施方式
下面将结合附图，对本说明书一个或多个实施例中的技术方案进行清楚、详尽地描述。其中，在本说明书一个或多个实施例的描述中，“多个”是指两个或多于两个。术语“第一”、“第二”仅用于描述目的，而不能理解为暗示或暗示相对重要性或隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者多个该特征。文本中的“和/或”仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。
图1是本说明书一个或多个实施例提供的一种对多行为序列进行时间聚合的示意图。
示例性的，如图1所示的多行为序列，该多行为序列对应多种行为的类型，每一种行为的类型对应一个时间点。如图1所示，时间点为0分钟时，是五角形类型的行为；时间点为1分钟时，是圆形类型的行为；时间点为2.2分钟时，是三角形类型的行为；时间点为3分钟时，是三角形类型的行为；时间点为3.5分钟时，是五角形类型的行为；依次类推，直到最后，时间点为27分钟时，是矩形类型的行为。其中，五角形类型为线上购买类型，圆形类型为转账类型，三角形类型为基金申购类型，以及矩形类型为聊天类型。
相关技术中，针对图1所示的多行为序列，将多行为序列按照预设的时间间隔进行等分，得到多段均匀时间上的行为序列。在多段均匀时间上的行为序列的每一段时间上的行为序列中进行时间聚合，并求得每一段时间上各个行为发生的次数。假设预设的时间间隔为4分钟，则在多行为序列进行等分，以及时间聚合后得到：0～4分钟：线上购买行为2次，转账行为1次，基金申购行为2次和聊天行为0次；4～8分钟：线上购买行为0次，转账行为1次，基金申购行为1次和聊天行为2次；8～12分钟：线上购买行为2次，转账行为1次，基金申购行为0次和聊天行为1次；12～16分钟：线上购买行为1次，转账行为1次，基金申购行为1次和聊天行为1次；16～20分钟：线上购买行为1次，转账行为2次，基金申购行为0次和聊天行为0次；20～24分钟：线上购买 行为0次，转账行为0次，基金申购行为2次和聊天行为1次以及24～28分钟：线上购买行为1次，转账行为0次，基金申购行为2次和聊天行为2次。
根据上述每隔4分钟统计的多行为序列的规律，预测得到的是未来某一类型的行为发生的时间段，而非时间点。因此，该预测方式在预测未来行为发生的时间信息上是不精确的。
图2是本说明书一个或多个实施例提供的一种训练行为预测模型的方法的示意性流程图。
应理解，本说明书一个或多个实施例提供的一种训练行为预测模型的方法可以应用于任意一种电子设备，该电子设备可以是计算机终端(具有处理数据功能的显示设备、手机终端)或服务器等。
还应理解，本说明书一个或多个实施例行为预测模型的确定过程是基于深度学习中的循环神经网络(Recurrent Neural Networks，RNN)实现的，具体可以是长短期记忆(Long Short Term Memory，LSTM)单元和门控循环单元(Gated Recurrent Unit，GRU)等。
示例性的，如图2所示，该方法200包括步骤202至步骤208。
202，电子设备对用户的行为序列进行拆分，得到多个单行为序列，各个单行为序列对应于一种行为，用于记录该行为和时间点的对应关系。
应理解，上述方案中的用户的行为序列指图1所示的多行为序列，该用户的行为序列与多行为序列一样，都对应多种类型的行为。
还应理解，用户的行为序列中每种类型的行为均有对应的时间点，即某种类型的行为具体发生的时间。
一种可能的实现方式中，步骤202，包括：电子设备根据行为的类型对该用户的行为序列进行拆分，得到该多个单行为序列，该用户的行为序列对应于多种行为的类型。
上述技术方案中，由于用户的行为序列对应于多种行为的类型，因此可以按照行为的类型将用户的行为序列进行拆分，拆分成多个单行为序列，多个单行为中各个单行为对应于一种行为的类型。从而可以对多个单行为序列进行时间编码得到多个单行为序列的多个单行为时间序列，从而多个单行为时间序列可以用于对行为预测模型的训练。
图3是本说明书一个或多个实施例提供的一种对用户的行为序列进行拆分的示意图。
示例性的，如图3具体是将图1所示的多行为序列(具体是序号为①的用户的行为序列)拆分后的多个单行为序列。其中，用户的行为序列对应多种行为的类型具体是五角形类型即线上购买类型、圆形类型即转账类型、三角形类型即基金申购类型和矩形类型即聊天类型。根据行为的类型将用户的行为序列拆分成多个单行为序列，分别为：序号为②的序列、序号为③的序列、序号为④的序列和序号为⑤的序列，即将用户的行为序列拆分成线上购买类型的行为序列、转账类型的行为序列、基金申购类型的行为序列 和聊天类型的行为序列。
204，电子设备对所述多个单行为序列进行时间编码，得到所述多个单行为序列的多个单行为时间序列。
一种可能的实现方式中，步骤204，包括：电子设备利用三角函数对该多个单行为序列进行时间编码，得到该多个单行为时间序列。
上述技术方案中，利用三角函数对多个单行为序列进行时间编码，可以得到多个单行为序列中各个单行为序列在时间上所有时间特征，得到多个单行为序列的多个单行为时间序列。
具体地，电子设备利用如下的三角函数的公式(1)对多个单行为序列中的各个单行为序列进行时间编码，得到多个单行为时间序列。
其中，ΦT表示某行为序列在一段时间内的映射；b1,…,bd为相位差参数ω1,…,ωd为频率参数；d为对某行为序列使用三角函数映射后的维度；t-τ表示某一时间段。
206，电子设备将所述多个单行为时间序列输入行为预测模型，由该行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况。
一种可能的实现方式中，步骤206，包括：由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行预设时间段内该时间点的强度值，得到该多个单行为时间序列对应的行为在时间上的分布情况。
应理解，上述技术方案是为了得到多个单行为时间序列对应的行为在时间上的分布情况，实则是得到多个单行为时间序列对应的行为在某时间段内的概率密度分布函数。
应理解，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行预设时间段内该时间点的强度值，得到该多个单行为时间序列对应的行为在时间上的分布情况。也就是得到如下公式(2)表示的多个单行为时间序列中各个单行为时间序列对应的行为在某一段时间上的概率密度分布函数。
其中，τi表示某一段时间；p*(τi)为某单行为时间序列对应的行为在某一段时间上的概率密度分布函数；表示某单行为时间序列对应的行为在某一段时间上的概率密度分布函数与该单行为序列中的历史行为有关；λ*(ti-1+τi)为在ti-1到τi的时间段内某行为发生的强度函数(强度值)，即某行为发生的概率；s表示在ti-1到τi的时间段内的某一时间点。
一种可能的实现方式中，在该行为预测模型包括行为时间子模型的情况下，步骤206，包括：电子设备将该多个单行为时间序列输入该行为预测模型中的该行为时间子模型， 由该行为时间子模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况。
上述技术方案中，是步骤206的一种更加具体的实现方式，具体是在行为预测模型包括行为时间子模型的情况下，可以将多个单行为时间序列输入到该行为时间子模型，由该行为时间子模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模。
208，电子设备基于所述多个单行为时间序列对应的行为在时间上的分布情况与所述多个单行为时间序列之间的第一损失值，对该行为预测模型进行训练，该行为预测模型用于基于该用户的行为序列预测该用户未来至少一种行为的时间点和类型。
一种可能的实现方式中，在该行为预测模型包括行为时间子模型的情况下，步骤208，包括：电子设备基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型中的该行为时间子模型进行训练。
上述技术方案中，是在行为预测模型包括行为时间子模型的情况下，具体利用该多个单行为时间序列以及该多个单行为时间序列对应的行为在时间上的分布情况之间的第一损失值对该行为时间子模型进行训练的方案。其中，该多个单行为时间序列对应的行为在时间上的分布情况是由该行为时间子模型对该多个单行为时间序列进行该时间点的建模得到的。
一种可能的实现方式中，在该行为预测模型还包括行为关系子模型和预测子模型的情况下，该方法还包括：电子设备将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列之间的因果关系，该因果关系用于表征在每两个单行为序列中一个单行为序列的行为的发生与另一个单行为序列的行为的发生之间的关系；电子设备将该多个单行为序列中每两个单行为序列之间的因果关系，以及该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由该预测子模型基于该因果关系和该分布情况进行序列预测，得到预测的行为序列；电子设备基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练。
上述技术方案中，在行为预测模型还包括行为关系子模型和预测子模型时，可以将多个单行为序列输入行为关系子模型，得到多个单行为序列中每两个单行为序列之间的因果关系；将多个单行为序列中每两个单行为序列之间的因果关系，以及多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由预测子模型得到预测的用户序列；从而基于第一损失值，以及预测的行为序列和用户的行为序列之间的第二损失值，对行为关系子模型和预测子模型进行训练。也就是说，行为预测模型不仅关注用户的行为序列的时间点信息，也关注用户的行为序列中多个行为之间的因果关系，从而可 以对用户的行为序列进行更加准确地建模。
一种可能的实现方式中，电子设备将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列之间的因果关系，包括：电子设备根据该多个单行为序列中每两个单行为序列在所有时间点上的协方差值之和，得到该多个单行为序列中每两个单行为序列之间的因果关系。
应理解，上述得到多个单行为序列中每两个单行为序列之间的因果关系具体是确定每两个单行为序列之间的格兰杰因果关系。
还应理解，电子设备根据如下格兰杰因果关系的公式(3)得到该多个单行为序列中每两个单行为序列之间的因果关系。
其中，Xi和分别为多个单行为序列中任意两个不同的单行为序列；i为单行为序列中的第i个行为；多个单行为序列中任意两个单行为序列可以组成k组中的一组序列；表示第k组序列中两个单行为序列之间的格兰杰因果关系；表示第k组序列中两个单行为序列之间的第i个行为的协方差。
一种可能的实现方式中，电子设备基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练，包括：电子设备确定预测标签与该用户的行为序列的真实标签之间的第三损失值，该真实标签用于指示该用户的真实信息，该预测标签用于指示该预测子模型预测的该用户的信息；电子设备对该第二损失值和该第三损失值求取平均值得到第四损失值；电子设备基于该第一损失值和该第四损失值对该行为关系子模型和该预测子模型进行训练。
上述技术方案中，相对于只基于第一损失值和第二损失值来训练行为关系子模型和预测子模型。上述技术方案还引入了预测标签与真实标签之间的第三损失值；将第三损失值和第二损失值之间的平均值确定为第四损失值，基于第一损失值和第四损失值来训练行为关系子模型和预测子模型。由于该方案引入了第三损失值，相当于引入更多的信息训练行为关系子模型和预测子模型，这样得到的行为关系子模型和预测子模型更加准确。
一种可能的实现方式中，在电子设备确定预测标签与该用户的行为序列的真实标签之间的第三损失值之前，该方法还包括：电子设备将该多个单行为序列中每两个单行为序列之间的因果关系与该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，得到该用户的行为序列对应的该预测标签。
可选地，用户的行为序列的真实标签可以是用户的身份信息和/或职业信息。
其中，用户的身份信息和/或职业信息的获取需要得到用户的同意。
一种可能的实现方式中，电子设备基于该第一损失值，以及该预测的行为序列和该 用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练，包括：电子设备根据该第一损失值和该第二损失值，得到第五损失值；电子设备基于该第五损失值训练该行为关系子模型，以及基于该第二损失值训练该预测子模型。
一种可能的实现方式中，电子设备根据该第一损失值和该第二损失值，得到第五损失值，包括：电子设备根据该行为关系子模型中的各个权重的二阶偏导构成的矩阵和该行为关系子模型中的各个权重构成的向量，确定第一数值；电子设备根据该第一数值、该第一损失值和该第二损失值得到该第五损失值。
应理解，上述确定第一数值的过程是利用帕累托最优解原理来实现的。
具体地，电子设备根据如下帕累托最优解公式(4)得到该第一数值。
其中，H为行为关系子模型中的各个权重的二阶偏导构成的矩阵，具体是黑森矩阵；υ为第一数值；f为行为关系子模型的向量值；为各个权重的目标梯度；β为行为关系子模型中各个权重组成的向量值。
一种可能的实现方式中，该方法还包括：电子设备将待预测的行为序列输入该行为预测模型，得到该行为预测模型预测的该待预测的行为序列的未来至少一个行为对应的时间点和类型。
上述技术方案中，具体是使用训练得到的行为预测模型的过程。可以利用训练好的行为预测模型对任意一个行为序列进行预测得到该行为序列的未来至少一个行为对应的时间点和类型。
一种可能的实现方式中，电子设备按照行为的类型将待预测的用户序列进行拆分，得到多个单行为用户序列；电子设备对多个单行为用户序列进行时间编码，得到多个单行为用户时间序列，该待检测的行为序列包括多个单行为用户序列和多个单行为用户时间序列。
可选地，电子设备对多个单行为用户序列进行时间编码，得到多个单行为用户时间序列，包括：电子设备利用三角函数对多个单行为用户序列进行时间编码，得到多个单行为用户时间序列。
一种可能的实现方式中，电子设备将多个单行为用户时间序列输入行为时间子模型，得到多个单行为用户时间序列对应的行为在时间上的分布情况；电子设备将多个单行为用户序列输入该行为关系子模型，得到多个单行为用户序列中每两个单行为用户序列之间的因果关系；电子设备将该分布情况和该因果关系输入预测子模型得到该待预测的行为序列的未来至少一个行为对应的时间点和类型。
图4是本说明书一个或多个实施例提供的一种行为预测模型的结构示意图。
如图4所示，具体描述了得到行为预测模型的过程。如图4，具体是将用户的行为 序列进行拆分得到多个单行为序列；对多个单行为序列进行时间编码，得到多个单行为时间序列；由行为时间子模型对多个单行为时间序列中的时间点进行建模，得到多个单行为时间序列对应的行为在时间上的分布情况；将多个单行为序列输入行为关系子模型，由行为关系子模型得到多个单行为序列中每两个单行为序列之间的因果关系；预测子模型可以根据多个单行为时间序列对应的行为在时间上的分布情况和多个单行为序列中每两个单行为序列之间的因果关系预测得到预测的行为序列，以及预测标签。
确定多个单行为时间序列与多个单行为时间序列对应的行为在时间上的分布情况之间的第一损失值；确定用户的行为序列与预测的行为序列之间的第二损失值；确定真实标签与预测标签的第三损失值。
可以基于第一损失值和第二损失值训练行为时间子模型、行为关系子模型和预测子模型；也可以基于第二损失值和第三损失值的平均值得到第四损失值，基于第一损失值和第四损失值训练行为时间子模型、行为关系子模型和预测子模型。
图5是本说明书一个或多个实施例提供的一种训练行为预测模型的装置的结构示意图。
示例性的，如图5所示，该装置500包括：确定模块501，用于对用户的行为序列进行拆分，得到多个单行为序列，各个单行为序列对应于一种行为，用于记录该行为和时间点的对应关系；该确定模块501，还用于对该多个单行为序列进行时间编码，得到该多个单行为序列的多个单行为时间序列；该确定模块501，还用于将该多个单行为时间序列输入行为预测模型，由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况；训练模块502，用于基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型进行训练，该行为预测模型用于基于该用户的行为序列预测该用户未来至少一种行为的时间点和类型。
可选地，该确定模块501，用于根据行为的类型对该用户的行为序列进行拆分，得到该多个单行为序列，该用户的行为序列对应于多种行为的类型。
可选地，该确定模块501，还用于利用三角函数对该多个单行为序列进行时间编码，得到该多个单行为时间序列。
可选地，该确定模块501，还用于由该行为预测模型对该多个单行为时间序列中各个单行为时间序列进行预设时间段内该时间点的强度值，得到该多个单行为时间序列对应的行为在时间上的分布情况。
应理解，上述方案是为了得到多个单行为时间序列对应的行为在时间上的分布情况，实则是得到多个单行为时间序列对应的行为在某时间段内的概率密度分布函数。
可选地，该行为预测模型包括行为时间子模型，该确定模块501，还用于将该多个 单行为时间序列输入该行为预测模型中的该行为时间子模型，由该行为时间子模型对该多个单行为时间序列中各个单行为时间序列进行该时间点的建模，得到该多个单行为时间序列对应的行为在时间上的分布情况。
可选地，该训练模块502，用于基于该多个单行为时间序列对应的行为在时间上的分布情况与该多个单行为时间序列之间的第一损失值，对该行为预测模型中的该行为时间子模型进行训练。
可选地，该行为预测模型还包括行为关系子模型和预测子模型，该确定模块501，还用于将该多个单行为序列输入该行为关系子模型，得到该多个单行为序列中每两个单行为序列的行为的发生之间的因果关系，该因果关系用于表征在每两个单行为序列中一个单行为序列的行为的发生与另一个单行为序列之间的关系；将该多个单行为序列中每两个单行为序列之间的因果关系，以及该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，由该预测子模型基于该因果关系和该分布情况进行序列预测，得到预测的行为序列；该训练模块502，还用于基于该第一损失值，以及该预测的行为序列和该用户的行为序列之间的第二损失值，对该行为关系子模型和该预测子模型进行训练。
可选地，该确定模块501，还用于根据该多个单行为中每两个单行为序列在所有时间点上的协方差值之和，得到该多个单行为序列中每两个单行为序列之间的因果关系。
可选地，该训练模块502，还用于确定预测标签与该用户的行为序列的真实标签之间的第三损失值，该真实标签用于指示该用户的真实信息，该预测标签用于指示该预测子模型预测的该用户的信息；对该第二损失值和该第三损失值求取平均值得到第四损失值；基于该第一损失值和该第四损失值对该行为关系子模型和该预测子模型进行训练。
可选地，该确定模块501，还用于将该多个单行为序列中每两个单行为序列之间的因果关系与该多个单行为时间序列对应的行为在时间上的分布情况输入该预测子模型，得到该用户的行为序列对应的该预测标签。
可选地，该训练模块502，还用于根据该第一损失值和该第二损失值，得到第五损失值；基于该第五损失值训练该行为关系子模型，以及基于该第二损失值训练该预测子模型。
可选地，该确定模块501，还用于根据该行为关系子模型中的各个权重的二阶偏导构成的矩阵和该行为关系子模型中的各个权重构成的向量，确定第一数值；根据该第一数值、该第一损失值和该第二损失值得到该第五损失值。
应理解，上述确定第一数值的过程是利用帕累托最优解原理来实现的。
可选地，该确定模块501，还用于将待预测的行为序列输入该行为预测模型，得到该行为预测模型预测的该待预测的行为序列的未来至少一个行为对应的时间点和类型。
图6是本说明书一个或多个实施例提供的一种电子设备的结构示意图。
示例性的，如图6所示，该电子设备600包括：存储器601、处理器602以及存储在该存储器601中并在处理器602上运行的计算机程序603，其中，该处理器602执行该计算机程序603时，使得该电子设备可执行前述介绍的任意一种训练行为预测模型的方法。
本说明书一个或多个实施例可以根据上述方法示例对电子设备进行功能模块的划分，例如，可以对应各个功能模块，也可以将两个或两个以上的功能集成在一个处理模块中，上述集成的模块可以采用硬件的形式实现。需要说明的是，本说明书一个或多个实施例中对模块的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。
在采用对应各个功能划分各个功能模块的情况下，该电子设备可以包括：确定模块和训练模块等。需要说明的是，上述方法实施例涉及的各个步骤的所有相关内容的可以援引到对应功能模块的功能描述，在此不再赘述。
本说明书一个或多个实施例提供的电子设备，用于执行上述一种训练行为预测模型的方法，因此可以达到与上述实现方法相同的效果。
在采用集成的单元的情况下，电子设备可以包括处理模块、存储模块。其中，处理模块可以用于对电子设备的动作进行控制管理。存储模块可以用于支持电子设备执行相互程序代码和数据等。
其中，处理模块可以是处理器或控制器，其可以实现或执行结合本说明书一个或多个实施例公开内容所藐视的各种示例性的逻辑方框，模块和电路。处理器也可以是实现计算功能的组合，例如包括一个或多个微处理器组合，数字信号处理(digital signal processing，DSP)和微处理器的组合等等，存储模块可以是存储器。
本说明书一个或多个实施例提供的电子设备具体可以是芯片，组件或模块，该电子设备可包括相连的处理器和存储器；其中，存储器用于存储指令，当电子设备运行时，处理器可调用并执行指令，以使芯片执行前述介绍的任意一种训练行为预测模型的方法。
本说明书一个或多个实施例提供一种计算机可读存储介质，该计算机可读存储介质中存储有指令，当该指令在计算机或处理器上运行时，使得该计算机或处理器执行前述介绍的任意一种训练行为预测模型的方法。
本说明书一个或多个实施例还提供了一种包含指令的计算机程序产品，当该计算机程序产品在计算机或处理器上运行时，使得计算机或处理器执行上述相关步骤，以实现前述介绍的任意一种训练行为预测模型的方法。
其中，本说明书一个或多个实施例提供的电子设备、计算机可读存储介质、包含指令的计算机程序产品或芯片均用于执行上文所提供的对应的方法，因此，其所能达到的有益效果可参考上文所提供的对应的方法中的有益效果，此处不再赘述。
通过以上实施方式的描述，所属领域的技术人员可以了解到，为描述的方便和简洁，仅以上述各功能模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能模块完成，即将装置的内部结构划分成不同的功能模块，以完成以上描述的全部或者部分功能。
在本说明书一个或多个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，模块或单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个装置，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
上述对本说明书特定实施例进行了描述。其它实施例在所附权利要求书的范围内。在一些情况下，在权利要求书中记载的动作或步骤可以按照不同于实施例中的顺序来执行并且仍然可以实现期望的结果。另外，在附图中描绘的过程不一定要求示出的特定顺序或者连续顺序才能实现期望的结果。在某些实施方式中，多任务处理和并行处理也是可以的或者可能是有利的。
以上内容，仅为本说明书中一个或多个具体的实施方式，但本说明书一个或多个实施例的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本说明书一个或多个实施例揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本说明书一个或多个实施例的保护范围之内。因此，本说明书一个或多个实施例的保护范围应以权利要求的保护范围为准。

Claims (17)

  1. 一种训练行为预测模型的方法，所述方法包括：
    对用户的行为序列进行拆分，得到多个单行为序列，各个所述单行为序列对应于一种行为，用于记录所述行为和时间点的对应关系；
    对所述多个单行为序列进行时间编码，得到所述多个单行为序列的多个单行为时间序列；
    将所述多个单行为时间序列输入行为预测模型，由所述行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行所述时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况；
    基于所述多个单行为时间序列对应的行为在时间上的分布情况与所述多个单行为时间序列之间的第一损失值，对所述行为预测模型进行训练，所述行为预测模型用于基于所述用户的行为序列预测所述用户未来至少一种行为的时间点和类型。
  2. 根据权利要求1所述的方法，所述对用户的行为序列进行拆分，得到多个单行为序列，包括：
    根据行为的类型对所述用户的行为序列进行拆分，得到所述多个单行为序列，所述用户的行为序列对应于多种行为的类型。
  3. 根据权利要求1所述的方法，所述对所述多个单行为序列进行时间编码，得到所述多个单行为序列的多个单行为时间序列，包括：
    利用三角函数对所述多个单行为序列进行时间编码，得到所述多个单行为时间序列。
  4. 根据权利要求1所述的方法，所述由所述行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行所述时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况，包括：
    由所述行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行预设时间段内所述时间点的强度值，得到所述多个单行为时间序列对应的行为在时间上的分布情况。
  5. 根据权利要求1-4中任意一项所述的方法，所述行为预测模型包括行为时间子模型，所述将所述多个单行为时间序列输入行为预测模型，由所述行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行所述时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况，包括：
    将所述多个单行为时间序列输入所述行为预测模型中的所述行为时间子模型，由所述行为时间子模型对所述多个单行为时间序列中各个单行为时间序列进行所述时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况。
  6. 根据权利要求5所述的方法，所述基于所述多个单行为时间序列对应的行为在 时间上的分布情况与所述多个单行为时间序列之间的第一损失值，对所述行为预测模型进行训练，包括：
    基于所述多个单行为时间序列对应的行为在时间上的分布情况与所述多个单行为时间序列之间的第一损失值，对所述行为预测模型中的所述行为时间子模型进行训练。
  7. 根据权利要求5所述的方法，所述行为预测模型还包括行为关系子模型和预测子模型，所述方法还包括：
    将所述多个单行为序列输入所述行为关系子模型，得到所述多个单行为序列中每两个单行为序列之间的因果关系，所述因果关系用于表征在每两个单行为序列中一个单行为序列的行为的发生与另一个单行为序列的行为的发生之间的关系；
    将所述多个单行为序列中每两个单行为序列之间的因果关系，以及所述多个单行为时间序列对应的行为在时间上的分布情况输入所述预测子模型，由所述预测子模型基于所述因果关系和所述分布情况进行序列预测，得到预测的行为序列；
    基于所述第一损失值，以及所述预测的行为序列和所述用户的行为序列之间的第二损失值，对所述行为关系子模型和所述预测子模型进行训练。
  8. 根据权利要求7所述的方法，所述将所述多个单行为序列输入所述行为关系子模型，得到所述多个单行为序列中每两个单行为序列之间的因果关系，包括：
    根据所述多个单行为中每两个单行为序列在所有时间点上的协方差值之和，得到所述多个单行为序列中每两个单行为序列之间的因果关系。
  9. 根据权利要求7所述的方法，所述基于所述第一损失值，以及所述预测的行为序列和所述用户的行为序列之间的第二损失值，对所述行为关系子模型和所述预测子模型进行训练，包括：
    确定预测标签与所述用户的行为序列的真实标签之间的第三损失值，所述真实标签用于指示所述用户的真实信息，所述预测标签用于指示所述预测子模型预测的所述用户的信息；
    对所述第二损失值和所述第三损失值求取平均值得到第四损失值；
    基于所述第一损失值和所述第四损失值对所述行为关系子模型和所述预测子模型进行训练。
  10. 根据权利要求9所述的方法，所述确定预测标签与所述用户的行为序列的真实标签之间的第三损失值之前，所述方法还包括：
    将所述多个单行为序列中每两个单行为序列之间的因果关系与所述多个单行为时间序列对应的行为在时间上的分布情况输入所述预测子模型，得到所述用户的行为序列对应的所述预测标签。
  11. 根据权利要求7-10中任意一项所述的方法，所述基于所述第一损失值，以及 所述预测的行为序列和所述用户的行为序列之间的第二损失值，对所述行为关系子模型和所述预测子模型进行训练，包括：
    根据所述第一损失值和所述第二损失值，得到第五损失值；
    基于所述第五损失值训练所述行为关系子模型，以及基于所述第二损失值训练所述预测子模型。
  12. 根据权利要求11所述的方法，所述根据所述第一损失值和所述第二损失值，得到第五损失值，包括：
    根据所述关系子模型中的各个权重的二阶偏导构成的矩阵和所述关系子模型中的各个权重构成的向量，确定第一数值；
    根据所述第一数值、所述第一损失值和所述第二损失值得到所述第五损失值。
  13. 根据权利要求1所述的方法，所述方法还包括：
    将待预测的行为序列输入所述行为预测模型，得到所述行为预测模型预测的所述待预测的行为序列的未来至少一个行为对应的时间点和类型。
  14. 一种训练行为预测模型的装置，所述装置包括：
    确定模块，用于对用户的行为序列进行拆分，得到多个单行为序列，各个所述单行为序列对应于一种行为，用于记录所述行为和时间点的对应关系；
    所述确定模块，还用于对所述多个单行为序列进行时间编码，得到所述多个单行为序列的多个单行为时间序列；
    所述确定模块，还用于将所述多个单行为时间序列输入行为预测模型，由所述行为预测模型对所述多个单行为时间序列中各个单行为时间序列进行所述时间点的建模，得到所述多个单行为时间序列对应的行为在时间上的分布情况；
    训练模块，用于基于所述多个单行为时间序列对应的行为在时间上的分布情况与所述多个单行为时间序列之间的第一损失值，对所述行为预测模型进行训练，所述行为预测模型用于基于所述用户的行为序列预测所述用户未来至少一种行为的时间点和类型。
  15. 一种电子设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，其中，所述处理器执行所述计算机程序时，使得所述电子设备执行如权利要求1至13中任意一项所述的一种训练行为预测模型的方法。
  16. 一种计算机可读存储介质，所述计算机可读存储介质中存储有指令，当所述指令在计算机或处理器上运行时，使得所述计算机或处理器执行如权利要求1至13中任意一项所述的一种训练行为预测模型的方法。
  17. 一种包含指令的计算机程序产品，当所述计算机程序产品在所述计算机或处理器上运行时，使得所述计算机或处理器执行如权利要求1至13中任意一项所述的一种训练行为预测模型的方法。
PCT/CN2023/130960 2022-11-29 2023-11-10 一种训练行为预测模型的方法及装置 WO2024114338A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202211513095.8A CN115828991A (zh) 2022-11-29 2022-11-29 一种训练行为预测模型的方法及装置
CN202211513095.8 2022-11-29

Publications (1)

Publication Number Publication Date
WO2024114338A1 true WO2024114338A1 (zh) 2024-06-06

Family

ID=85532727

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/130960 WO2024114338A1 (zh) 2022-11-29 2023-11-10 一种训练行为预测模型的方法及装置

Country Status (2)

Country Link
CN (1) CN115828991A (zh)
WO (1) WO2024114338A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115828991A (zh) * 2022-11-29 2023-03-21 蚂蚁财富(上海)金融信息服务有限公司 一种训练行为预测模型的方法及装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190087470A1 (en) * 2015-10-28 2019-03-21 Tongji University System and method for mining user cycle mode
CN111046084A (zh) * 2019-12-18 2020-04-21 重庆大学 一种多元时间序列监测数据的关联规则挖掘方法
CN113821720A (zh) * 2021-07-14 2021-12-21 腾讯科技（深圳）有限公司 一种行为预测方法、装置及相关产品
CN114119151A (zh) * 2021-11-23 2022-03-01 上海交通大学 下一个购物篮个性化推荐方法、***及介质
US20220138537A1 (en) * 2020-11-02 2022-05-05 International Business Machines Corporation Probabilistic nonlinear relationships cross-multi time series and external factors for improved multivariate time series modeling and forecasting
CN115169551A (zh) * 2022-06-30 2022-10-11 支付宝(杭州)信息技术有限公司 行为预测模型的训练方法、风险行为预测方法和装置
CN115828991A (zh) * 2022-11-29 2023-03-21 蚂蚁财富(上海)金融信息服务有限公司 一种训练行为预测模型的方法及装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190087470A1 (en) * 2015-10-28 2019-03-21 Tongji University System and method for mining user cycle mode
CN111046084A (zh) * 2019-12-18 2020-04-21 重庆大学 一种多元时间序列监测数据的关联规则挖掘方法
US20220138537A1 (en) * 2020-11-02 2022-05-05 International Business Machines Corporation Probabilistic nonlinear relationships cross-multi time series and external factors for improved multivariate time series modeling and forecasting
CN113821720A (zh) * 2021-07-14 2021-12-21 腾讯科技（深圳）有限公司 一种行为预测方法、装置及相关产品
CN114119151A (zh) * 2021-11-23 2022-03-01 上海交通大学 下一个购物篮个性化推荐方法、***及介质
CN115169551A (zh) * 2022-06-30 2022-10-11 支付宝(杭州)信息技术有限公司 行为预测模型的训练方法、风险行为预测方法和装置
CN115828991A (zh) * 2022-11-29 2023-03-21 蚂蚁财富(上海)金融信息服务有限公司 一种训练行为预测模型的方法及装置

Also Published As

Publication number Publication date
CN115828991A (zh) 2023-03-21

Similar Documents

Publication Publication Date Title
US10831486B2 (en) Automation of sequences of actions
JP6817426B2 (ja) マシンラーニング基盤の半導体製造の収率予測システム及び方法
US11687352B2 (en) Machine-learning models applied to interaction data for determining interaction goals and facilitating experience-based modifications to interface elements in online environments
US11775412B2 (en) Machine learning models applied to interaction data for facilitating modifications to online environments
WO2024114338A1 (zh) 一种训练行为预测模型的方法及装置
CN108052979A (zh) 对模型预测值进行融合的方法、装置和设备
CN111401722B (zh) 智能决策方法和智能决策***
CN110971659A (zh) 推荐消息的推送方法、装置及存储介质
CN111427974A (zh) 数据质量评估管理方法和装置
Khodadadi et al. ChOracle: A unified statistical framework for churn prediction
WO2020199962A1 (zh) 改进传统零售行业货架摆放的方法
CN110427358B (zh) 数据清洗方法及装置和信息推荐方法及装置
CN116013228A (zh) 一种音乐生成方法、装置、电子设备及其存储介质
CN117217505A (zh) 一种基于书籍领域的资源管理***
CN110097113B (zh) 一种监控展示信息投放***工作状态的方法、装置及***
CN113822693A (zh) 用户购买力评价值的生成方法、装置、设备及存储介质
CN111241821B (zh) 确定用户的行为特征的方法和装置
CN116501979A (zh) 信息推荐方法、装置、计算机设备及计算机可读存储介质
CN111368195A (zh) 一种模型训练方法、装置、设备及存储介质
CN110971973A (zh) 一种视频推送方法、装置及电子设备
JP2004078780A (ja) 予測方法、予測装置、予測プログラム、および予測プログラムを記録した記録媒体
US11593740B1 (en) Computing system for automated evaluation of process workflows
US11762562B2 (en) Performance analysis apparatus and performance analysis method
CN111507471B (zh) 一种模型训练方法、装置、设备及存储介质
CN114490406A (zh) 测试覆盖项管理方法、装置、设备及介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23896490

Country of ref document: EP

Kind code of ref document: A1