CN110414710A - Trend forecasting method and device based on artificial intelligence - Google Patents
Trend forecasting method and device based on artificial intelligence Download PDFInfo
- Publication number
- CN110414710A CN110414710A CN201910534829.2A CN201910534829A CN110414710A CN 110414710 A CN110414710 A CN 110414710A CN 201910534829 A CN201910534829 A CN 201910534829A CN 110414710 A CN110414710 A CN 110414710A
- Authority
- CN
- China
- Prior art keywords
- data
- information
- stock
- study
- candidate target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
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
- G06N3/08—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"
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Abstract
The embodiment of the invention provides a kind of trend forecasting method and device based on artificial intelligence.On the one hand, this method comprises: determining candidate target to be predicted；Extract the historical yield information and history public feelings information of the candidate target；The historical yield information and the history public feelings information are input to Study on Stock Prediction Model, wherein the Study on Stock Prediction Model is obtained using Encode-Decode from encryption algorithm training；The future profits information of the candidate target is predicted using the Study on Stock Prediction Model.Through the invention, it solves in the prior art through accuracy rate low technical problem when Linear Model for Prediction stock.
Description
[technical field]
The present invention relates to computer field more particularly to a kind of trend forecasting methods and device based on artificial intelligence.
[background technique]
Traditional factor, which is selected stocks, has hysteresis quality based on historical stock data, and is strongly dependent upon stock invester personal experience.
It is linear model that the factor in the prior art, which is selected stocks, and when building is easy to cause over-fitting to historical data, then
Relative attenuation is showed in firm offer, in addition personal consider that the stock factor can only be the index of correlation of specific stock, it is not as depth
The factor is selected stocks like that automatically from the variation numerical value of stock, and feature is extracted.In other words, traditional factor is selected stocks in the presence of certain
Bottleneck.Traditional factor model sorts out four aspects: market ere-ment, the style factor, the financial factor, factor of momentum.Each
There is careful classification again in the factor of aspect.Although can be according to individual's analysis and risk analysis building because of subsystem, with more
Factor building quantized combinations make personal policy-making.
But this is very high for personal skill requirement, and is many times difficult to consider comprehensively.Since the factor is selected stocks
The more accumulation of the personal experience based on stock invester, the historical data factor are selected stocks there are hysteresis quality, are difficult to accomplish through magnanimity stock
The comprehensive analysis of the rationalization of ticket data.
For the above problem present in the relevant technologies, at present it is not yet found that the solution of effect.
[summary of the invention]
In view of this, the embodiment of the invention provides a kind of trend forecasting method and device based on artificial intelligence.
On the one hand, the embodiment of the invention provides a kind of trend forecasting methods based on artificial intelligence, which comprises
Determine candidate target to be predicted；Extract the historical yield information and history public feelings information of the candidate target；By the history
Avail information and the history public feelings information are input to Study on Stock Prediction Model, wherein the Study on Stock Prediction Model is to use
Encode-Decode is obtained from encryption algorithm training；The future of the candidate target is predicted using the Study on Stock Prediction Model
Avail information.
Optionally, before by the historical yield information to Study on Stock Prediction Model, the method also includes: acquisition stock
Stock certificate data of the sample in predetermined period, wherein the stock certificate data includes the avail information of first time period, first time
The public feelings information of section and the avail information of second time period, wherein the second time period is in the nature of first time period
Between after, the stock sample conforms to a predetermined condition in the situation of Profit of the second time period；By the first time period
The public feelings information of avail information and the first time period is made as input label data, the avail information of the second time period
The Study on Stock Prediction Model is obtained using Encode-Decode algorithm training initial model for output label data.
It optionally, include: to believe the income of the first time period using Encode-Decode algorithm training initial model
Breath and the public feelings information of the first time period are compressed to the hidden layer of the initial model, in hidden layer to the input label
Data are encoded, and the first data are exported；By the output layer of first data decompression to the initial model, in output layer pair
First data are decoded, and export the second data, wherein the hidden layer is connected with the output layer；By described second
Data obtain third data multiplied by loss function；The input label data are updated to the third data, and in conjunction with described
Output label data train initial model, obtain the Study on Stock Prediction Model.
Optionally, carrying out coding to the input label data in hidden layer includes: to use with minor function to the input
Label data is encoded: h1=f (xW1+b1), wherein x is input label data, W1It is first layer network weight, b1It is inclined
Difference, h1It is the first data.
Optionally, being decoded in output layer to first data includes: to use with minor function to first data
It is decoded: O2=g (h1W2+b2), wherein h1It is the first data, O2It is the second data, W2It is second layer network weight, b2It is inclined
Difference.
Optionally, include: by second data multiplied by following loss function multiplied by loss function by second data:Wherein, ε is loss amount, O2It is the second data, x is input label data.
Optionally, predict that the future profits information of the candidate target includes: using institute using the Study on Stock Prediction Model
It states historical yield information and the history public feelings information constructs the valence of the candidate target based on the Encode-Decode algorithm
Lattice characteristic information, wherein the probability that the price feature information is used to characterize multiple future profits rates of the candidate target is bent
Line；According to the probability curve, the highest earning rate of select probability is determined as the following receipts of the candidate target in income section
Beneficial rate.
On the other hand, the embodiment of the invention provides a kind of trend prediction device based on artificial intelligence, described device packet
It includes: determining module, for determining candidate target to be predicted；Extraction module, for extracting the historical yield of the candidate target
Information and history public feelings information；Input module, for the historical yield information and the history public feelings information to be input to stock
Ticket prediction model, wherein the Study on Stock Prediction Model is obtained using Encode-Decode from encryption algorithm training；Prediction
Module, for predicting the future profits information of the candidate target using the Study on Stock Prediction Model.
Optionally, described device further include: acquisition module, for the input module by the historical yield information and
The history public feelings information is input to before Study on Stock Prediction Model, acquires stock certificate data of the stock sample in predetermined period,
In, the stock certificate data includes the avail information of first time period, the public feelings information of first time period and second time period
Avail information, wherein the second time period is after the natural time of first time period, and the stock sample is described second
The situation of Profit of period conforms to a predetermined condition；Training module, for by the avail information of the first time period and described
The public feelings information of one period as input label data, the avail information of the second time period as output label data,
Using Encode-Decode algorithm training initial model, the Study on Stock Prediction Model is obtained.
Optionally, the training module includes: coding unit, for by the avail information of the first time period and described
The public feelings information of first time period is compressed to the hidden layer of the initial model, carries out in hidden layer to the input label data
Coding exports the first data；Decoding unit, for by the output layer of first data decompression to the initial model, defeated
Layer is decoded first data out, exports the second data, wherein the hidden layer is connected with the output layer；Training
Unit, for second data multiplied by loss function, to be obtained third data；The input label data are updated to described
Third data, and in conjunction with output label data training initial model, obtain the Study on Stock Prediction Model.
Optionally, the coding unit includes: coded sub-units, for using with minor function to the input label data
It is encoded: h1=f (xW1+b1), wherein x is input label data, W1It is first layer network weight, b1It is deviation, h1It is
One data.
Optionally, the decoding unit includes: decoding subunit, is carried out with minor function to first data for using
Decoding: O2=g (h1W2+b2), wherein h1It is the first data, O2It is the second data, W2It is second layer network weight, b2It is deviation.
Optionally, the training unit includes: trained subelement, is used for second data multiplied by following loss letter
Number:Wherein, ε is loss amount, O2It is the second data, x is input label data.
Optionally, the prediction module includes: construction unit, for using the historical yield information and the history carriage
Feelings information constructs the price feature information of the candidate target based on the Encode-Decode algorithm, wherein the price is special
Reference ceases the probability curve for characterizing multiple future profits rates of the candidate target；Determination unit, for according to described general
Rate curve highest earning rate of select probability in income section is determined as the future profits rate of the candidate target.
According to still another embodiment of the invention, a kind of storage medium is additionally provided, meter is stored in the storage medium
Calculation machine program, wherein the computer program is arranged to execute the step in any of the above-described embodiment of the method when operation.
According to still another embodiment of the invention, a kind of electronic device, including memory and processor are additionally provided, it is described
Computer program is stored in memory, the processor is arranged to run the computer program to execute any of the above-described
Step in embodiment of the method.
Through the invention, Study on Stock Prediction Model is obtained from encryption algorithm training by using Encode-Decode, from coding
Algorithm can extract the nonlinear characteristic that data movement is potentially abstracted, then using Study on Stock Prediction Model predicting candidate object
Future profits information constructs nonlinear mind for earning rate from coding Encode-Decode algorithm model by deep learning
Through network model, solve in the prior art through accuracy rate low technical problem when Linear Model for Prediction stock.
[Detailed description of the invention]
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this field
For those of ordinary skill, without any creative labor, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is a kind of hardware block diagram of trend prediction mobile terminal based on artificial intelligence of the embodiment of the present invention；
Fig. 2 is the flow chart of the trend forecasting method according to an embodiment of the present invention based on artificial intelligence；
Fig. 3 is flow chart of the embodiment of the present invention using Encode-Decode algorithm training initial model；
Fig. 4 is the structural block diagram of the trend prediction device according to an embodiment of the present invention based on artificial intelligence.
[specific embodiment]
Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and in combination with Examples.It should be noted that not conflicting
In the case of, the features in the embodiments and the embodiments of the present application can be combined with each other.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.
Embodiment 1
Embodiment of the method provided by the embodiment of the present application one can be in mobile terminal, terminal or similar fortune
It calculates and is executed in device.For running on mobile terminals, Fig. 1 is a kind of trend based on artificial intelligence of the embodiment of the present invention
Predict the hardware block diagram of mobile terminal.As shown in Figure 1, mobile terminal 10 may include one or more (only shows in Fig. 1
One) (processor 102 can include but is not limited to the place of Micro-processor MCV or programmable logic device FPGA etc. to processor 102
Manage device) and memory 104 for storing data, optionally, above-mentioned mobile terminal can also include for communication function
Transmission device 106 and input-output equipment 108.It will appreciated by the skilled person that structure shown in FIG. 1 is only to show
Meaning, does not cause to limit to the structure of above-mentioned mobile terminal.For example, mobile terminal 10 may also include it is more than shown in Fig. 1
Perhaps less component or with the configuration different from shown in Fig. 1.
Memory 104 can be used for storing computer program, for example, the software program and module of application software, such as this hair
The corresponding computer program of the trend forecasting method based on artificial intelligence in bright embodiment, processor 102 pass through operation storage
Computer program in memory 104 realizes above-mentioned method thereby executing various function application and data processing.
Memory 104 may include high speed random access memory, may also include nonvolatile memory, as one or more magnetic storage fills
It sets, flash memory or other non-volatile solid state memories.In some instances, memory 104 can further comprise relative to place
The remotely located memory of device 102 is managed, these remote memories can pass through network connection to mobile terminal 10.Above-mentioned network
Example includes but is not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
Transmitting device 106 is used to that data to be received or sent via a network.Above-mentioned network specific example may include
The wireless network that the communication providers of mobile terminal 10 provide.In an example, transmitting device 106 includes a Network adaptation
Device (Network Interface Controller, referred to as NIC), can be connected by base station with other network equipments to
It can be communicated with internet.In an example, transmitting device 106 can for radio frequency (Radio Frequency, referred to as
RF) module is used to wirelessly be communicated with internet.
A kind of trend forecasting method based on artificial intelligence is provided in the present embodiment, and Fig. 2 is to implement according to the present invention
The flow chart of the trend forecasting method based on artificial intelligence of example, as shown in Fig. 2, the process includes the following steps:
Step S202 determines candidate target to be predicted；
In the present embodiment, candidate target is the stock that user wants prediction future trend, can be by choosing stock mark
Know or fund mark is to determine.
Step S204 extracts the historical yield information and history public feelings information of the candidate target；
The history public feelings information of the present embodiment is the information for influencing income, that is, influences the relevant information of income, as news is sent out
Cloth number, if publication financial statement, if purchase is purchased, sales achievement, profit margin, number of diplomatizing, if there are works
Quotient punishes, if generates commercial dispute etc..It can be acquired from various disclosed media or information channel (such as government website)
And acquisition.
The historical yield information and the history public feelings information are input to Study on Stock Prediction Model by step S206, wherein
The Study on Stock Prediction Model is obtained using Encode-Decode from encryption algorithm training；
Step S208 predicts the future profits information of the candidate target using the Study on Stock Prediction Model.
Scheme through this embodiment obtains Prediction of Stock Index mould from encryption algorithm training by using Encode-Decode
Type can extract the nonlinear characteristic that data movement is potentially abstracted from encryption algorithm, then be predicted using Study on Stock Prediction Model
The future profits information of candidate target is constructed from coding Encode-Decode algorithm model for earning rate by deep learning
Nonlinear neural network model is solved and is asked in the prior art by accuracy rate low technology when Linear Model for Prediction stock
Topic.
In the present embodiment, it is also necessary to the training Study on Stock Prediction Model, by the historical yield information and the history carriage
Before feelings information input to Study on Stock Prediction Model, further includes:
S11 acquires stock certificate data of the stock sample in predetermined period, wherein the stock certificate data includes at the first time
Avail information, the public feelings information of first time period and the avail information of second time period of section, wherein second time
For section after the natural time of first time period, the stock sample meets predetermined item in the situation of Profit of the second time period
Part；
Predetermined condition is the stock that surging stock refers to that earning rate is high in deep bid, in one month for predetermined period, such as:
The maximum n branch stock of accumulated growth rate adds up to increase the most n branch stock of market value, and the most n branch stock of land number of days can also
Being divided according to column, screening several surging stock in each column.Stock certificate data includes: the 1st to 15 day income letter
Breath, the 1st to 15 day public feelings information, the 16th to 30 day avail information.
S12, using the public feelings information of the avail information of the first time period and the first time period as input label
Data, the avail information of the second time period is as output label data, using Encode-Decode algorithm training introductory die
Type obtains the Study on Stock Prediction Model.
Input label data in this programme include the financial value of first time period, influence value (such as public sentiment of extraneous factor
Information), pass through weighted sum.Output label data are the financial value of second time period, and first time period and second time period can
To be the adjacent or non-conterminous period, financial value is made of one group of data, including two dimensions of time and income.
The feature (time, income) of the present embodiment is all the value type after quantization, and feature is deposited with the result predicted is needed
In some influent factors, the present embodiment is by extracting non-linear element therein, and then training obtains Study on Stock Prediction Model.From
Encode the high-order feature that Encode-Decode passes through deep neural network model learning training data；For the number not marked
According to, can be used it is unsupervised from coding to learn sparse high-order feature and reconstruct oneself.Self-encoding encoder is neural network
One kind being compressed to hidden layer by that will input, then unzips to output layer, and reconstruct is originally inputted, and is a kind of unsupervised learning method,
Using backpropagation training pattern, it is mainly used to carry out Data Dimensionality Reduction, feature learning, go hot-tempered, production modeling etc..From coding mind
It is generally made of input layer, hidden layer and output layer through network.
In an embodiment of the present embodiment, initial model is trained using Encode-Decode algorithm, Fig. 3 is
The embodiment of the present invention trains the flow chart of initial model using Encode-Decode algorithm, as shown in Figure 3, comprising:
The public feelings information of the avail information of the first time period and the first time period is compressed to described first by S302
The hidden layer of beginning model encodes the input label data in hidden layer, exports the first data；
Optionally, carrying out coding to the input label data in hidden layer includes: to use with minor function to the input
Label data is encoded: h1=f (xW1+b1), wherein x is input label data, W1It is first layer network weight, b1It is inclined
Difference, h1It is the first data.Since there are different degrees of redundancies for input information itself, these are removed by study from coding
Redundancy feature is input to all useful features in hidden layer.If activation primitive is linear function, compression process is just
It is PCA process (principal component analytical method is a kind of data compression algorithm), the activation primitive of the hidden layer of the present embodiment is non-thread
Property function,
S304, by the output layer of first data decompression to the initial model, in output layer to first data
It is decoded, exports the second data, wherein the hidden layer is connected with the output layer in an initial model case；
Optionally, being decoded in output layer to first data includes: to use with minor function to first data
It is decoded: O2=g (h1W2+b2), wherein h1It is the first data, O2It is the second data, W2It is second layer network weight, b2It is inclined
Difference.
S306 obtains third data by second data multiplied by loss function；
Optionally, include: by second data multiplied by following loss function multiplied by loss function by second data:Wherein, ε is loss amount, O2It is the second data, x is input label data.
In initial model, if it is 1*n that input layer, which is dimension, then the dimension of output layer is also 1*n, the dimension of hidden layer is
1*m.Since the purpose from coding neural network is to make output consistent with the dimension of input, it is from the training for encoding neural network
Means, what is be more absorbed in is the function of the feature extraction of hidden layer, if n > m, i.e., hidden layer number of nodes is less than input layer
Number of nodes, input layer to hidden layer are a kind of dimensionality reduction operations, and self-encoding encoder at this time is known as owing complete from encoding, and work as activation primitive
For linear function and loss function is quadratic form error and when hidden layer only has one layer, input layer to the dimensionality reduction behaviour between hidden layer
It is equivalent to principal component analysis PCA.If n≤m, the output information of the network model trained at this time may be replicated directly
Information is inputted, but if regularization correction is added, then canonical self-encoding encoder can be obtained.
The input label data are updated to the third data by S308, and in conjunction with output label data training
Initial model obtains the Study on Stock Prediction Model.
In an embodiment of the present embodiment, the future of the candidate target is predicted using the Study on Stock Prediction Model
Avail information includes: to be based on the Encode-Decode algorithm using the historical yield information and the history public feelings information
Construct the price feature information of the candidate target, wherein the price feature information is for characterizing the more of the candidate target
The probability curve of a future profits rate；According to the probability curve, the highest earning rate of select probability is determined as in income section
The future profits rate of the candidate target.
It is extracted by feature of the neural network to candidate target, and provides every expected stock return with these features
The confidence level (" probability " in different earning rate sections) in rate section determines according to these confidence levels and recommends stock, comprising: extracts
The historical yield information and history public feelings information of candidate target extract market raw information as input layer data and input nerve net
Network.For example, can then be made if necessary to the pricing information for excavating personal share with the earning rate of personal share past T phase and public opinion index
It is input in neural network for initial data.Price according to historical yield information and history public feelings information building candidate target is special
Reference breath learns the nonlinear characteristic that pricing information contains by autoencoder network.Most base in this certain applications deep learning
From coding to construct price feature information, price feature information is one and walks power curve, including time and income two this algorithm
A dimension, by price feature information input to model of selecting stocks.The earning rate for exporting candidate target, the valence extracted using previous step
Lattice feature predicts the expected yield of T+1 phase, i.e., predicts the stock yield of some following day of trade.
In order to enable output item is more clear, continuous earning rate numerical intervals can be exported and fall in different incomes
" probability " in rate section, the probability as earning rate falls in the section 10%-20% is 2%, and the probability for falling in the section 1%-5% is
30%, the probability for falling in the section 5%-10% is 65%.The principle of the confidence level of this programme is according to prediction input data and instruction
Practice the similarity and determination of the sample data used, similarity is higher, illustrates that model more agrees with, confidence level is higher.It can choose
Maximum probability or income highest topN are as prediction result；When candidate target is more, this probability value, root can also be passed through
The corresponding combination is constructed according to the different investment objectives and investment types, if robustness select probability is maximum, radical type selection, probability
Under the premise of preset value, earning rate is highest etc..
Through the above description of the embodiments, those skilled in the art can be understood that according to above-mentioned implementation
The method of example can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but it is very much
In the case of the former be more preferably embodiment.Based on this understanding, technical solution of the present invention is substantially in other words to existing
The part that technology contributes can be embodied in the form of software products, which is stored in a storage
In medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, calculate
Machine, server or network equipment etc.) execute method described in each embodiment of the present invention.
Embodiment 2
A kind of trend prediction device based on artificial intelligence is additionally provided in the present embodiment, and the device is for realizing above-mentioned
Embodiment and preferred embodiment, the descriptions that have already been made will not be repeated.As used below, term " module " can be real
The combination of the software and/or hardware of existing predetermined function.Although device described in following embodiment is preferably realized with software,
But the realization of the combination of hardware or software and hardware is also that may and be contemplated.
Fig. 4 is the structural block diagram of the trend prediction device according to an embodiment of the present invention based on artificial intelligence, such as Fig. 4 institute
Show, which includes:
Determining module 40, for determining candidate target to be predicted；
Extraction module 42, for extracting the historical yield information and history public feelings information of the candidate target；
Input module 44, for the historical yield information and the history public feelings information to be input to Prediction of Stock Index mould
Type, wherein the Study on Stock Prediction Model is obtained using Encode-Decode from encryption algorithm training；
Prediction module 46, for predicting the future profits information of the candidate target using the Study on Stock Prediction Model.
Optionally, described device further include: acquisition module, for the input module by the historical yield information and
The history public feelings information is input to before Study on Stock Prediction Model, acquires stock certificate data of the stock sample in predetermined period,
In, the stock certificate data includes the avail information of first time period, the public feelings information of first time period and second time period
Avail information, wherein the second time period is after the natural time of first time period, and the stock sample is described second
The situation of Profit of period conforms to a predetermined condition；Training module, for by the avail information of the first time period and described
The public feelings information of one period as input label data, the avail information of the second time period as output label data,
Using Encode-Decode algorithm training initial model, the Study on Stock Prediction Model is obtained.
Optionally, the training module includes: coding unit, for by the avail information of the first time period and described
The public feelings information of first time period is compressed to the hidden layer of the initial model, carries out in hidden layer to the input label data
Coding exports the first data；Decoding unit, for by the output layer of first data decompression to the initial model, defeated
Layer is decoded first data out, exports the second data, wherein the hidden layer is connected with the output layer；Training
Unit, for second data multiplied by loss function, to be obtained third data；The input label data are updated to described
Third data, and in conjunction with output label data training initial model, obtain the Study on Stock Prediction Model.
Optionally, the coding unit includes: coded sub-units, for using with minor function to the input label data
It is encoded: h1=f (xW1+b1), wherein x is input label data, W1It is first layer network weight, b1It is deviation, h1It is
One data.
Optionally, the decoding unit includes: decoding subunit, is carried out with minor function to first data for using
Decoding: O2=g (h1W2+b2), wherein h1It is the first data, O2It is the second data, W2It is second layer network weight, b2It is deviation.
Optionally, the training unit includes: trained subelement, is used for second data multiplied by following loss letter
Number:Wherein, ε is loss amount, O2It is the second data, x is input label data.
Optionally, the prediction module includes: construction unit, for using the historical yield information and the history carriage
Feelings information constructs the price feature information of the candidate target based on the Encode-Decode algorithm, wherein the price is special
Reference ceases the probability curve for characterizing multiple future profits rates of the candidate target；Determination unit, for according to described general
Rate curve highest earning rate of select probability in income section is determined as the future profits rate of the candidate target.
It should be noted that above-mentioned modules can be realized by software or hardware, for the latter, Ke Yitong
Following manner realization is crossed, but not limited to this: above-mentioned module is respectively positioned in same processor；Alternatively, above-mentioned modules are with any
Combined form is located in different processors.
Embodiment 3
In several embodiments provided by the present invention, it should be understood that disclosed system, device and method can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only a kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or group
Part can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown
Or the mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of hardware adds SFU software functional unit.
The above-mentioned integrated unit being realized in the form of SFU software functional unit can store and computer-readable deposit at one
In storage media.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions are used so that a computer
It is each that device (can be personal computer, server or network equipment etc.) or processor (Processor) execute the present invention
The part steps of embodiment the method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-
Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. it is various
It can store the medium of program code.
The embodiments of the present invention also provide a kind of storage medium, computer program is stored in the storage medium, wherein
The computer program is arranged to execute the step in any of the above-described embodiment of the method when operation.
Optionally, in the present embodiment, above-mentioned storage medium can be set to store by executing based on following steps
Calculation machine program:
S1 determines candidate target to be predicted；
S2 extracts the historical yield information and history public feelings information of the candidate target；
The historical yield information and the history public feelings information are input to Study on Stock Prediction Model, wherein the stock by S3
Ticket prediction model is obtained using Encode-Decode from encryption algorithm training；
S4 predicts the future profits information of the candidate target using the Study on Stock Prediction Model.
Optionally, in the present embodiment, above-mentioned storage medium can include but is not limited to: USB flash disk, read-only memory (Read-
Only Memory, referred to as ROM), it is random access memory (Random Access Memory, referred to as RAM), mobile hard
The various media that can store computer program such as disk, magnetic or disk.
The embodiments of the present invention also provide a kind of electronic device, including memory and processor, stored in the memory
There is computer program, which is arranged to run computer program to execute the step in any of the above-described embodiment of the method
Suddenly.
Optionally, above-mentioned electronic device can also include transmission device and input-output equipment, wherein the transmission device
It is connected with above-mentioned processor, which connects with above-mentioned processor.
Optionally, in the present embodiment, above-mentioned processor can be set to execute following steps by computer program:
S1 determines candidate target to be predicted；
S2 extracts the historical yield information and history public feelings information of the candidate target；
The historical yield information and the history public feelings information are input to Study on Stock Prediction Model, wherein the stock by S3
Ticket prediction model is obtained using Encode-Decode from encryption algorithm training；
S4 predicts the future profits information of the candidate target using the Study on Stock Prediction Model.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (10)
1. a kind of trend forecasting method based on artificial intelligence, which is characterized in that the described method includes:
Determine candidate target to be predicted；
Extract the historical yield information and history public feelings information of the candidate target；
The historical yield information and the history public feelings information are input to Study on Stock Prediction Model, wherein the Prediction of Stock Index
Model is obtained using Encode-Decode from encryption algorithm training；
The future profits information of the candidate target is predicted using the Study on Stock Prediction Model.
2. the method according to claim 1, wherein believing by the historical yield information and the history public sentiment
Breath is input to before Study on Stock Prediction Model, the method also includes:
Acquire stock certificate data of the stock sample in predetermined period, wherein the stock certificate data includes the income of first time period
The avail information of information, the public feelings information of first time period and second time period, wherein the second time period is first
After the natural time of period, the stock sample conforms to a predetermined condition in the situation of Profit of the second time period；
It is described using the public feelings information of the avail information of the first time period and the first time period as input label data
The avail information of second time period obtains institute using Encode-Decode algorithm training initial model as output label data
State Study on Stock Prediction Model.
3. according to the method described in claim 2, it is characterized in that, using Encode-Decode algorithm training initial model packet
It includes:
The public feelings information of the avail information of the first time period and the first time period is compressed to the initial model
Hidden layer encodes the input label data in hidden layer, exports the first data；
By the output layer of first data decompression to the initial model, first data are decoded in output layer,
Export the second data, wherein the hidden layer is connected with the output layer；
By second data multiplied by loss function, third data are obtained；
The input label data are updated to the third data, and train initial model in conjunction with the output label data,
Obtain the Study on Stock Prediction Model.
4. according to the method described in claim 3, it is characterized in that, carrying out coding packet to the input label data in hidden layer
It includes:
It uses and the input label data is encoded with minor function:
h1=f (xW1+b1), wherein x is input label data, W1It is first layer network weight, b1It is deviation, h1It is the first number
According to.
5. according to the method described in claim 3, it is characterized in that, being decoded in output layer to first data and including:
It uses and first data is decoded with minor function:
O2=g (h1W2+b2), wherein h1It is the first data, O2It is the second data, W2It is second layer network weight, b2It is deviation.
6. according to the method described in claim 3, it is characterized in that, including: multiplied by loss function by second data
By second data multiplied by following loss function:
Wherein, ε is loss amount, O2It is the second data, x is input label data.
7. the method according to claim 1, wherein predicting the candidate target using the Study on Stock Prediction Model
Future profits information include:
The Encode-Decode algorithm, which is based on, using the historical yield information and the history public feelings information constructs the time
Select the price feature information of object, wherein the price feature information is used to characterize multiple future profits of the candidate target
The probability curve of rate；
According to the probability curve, the highest earning rate of select probability is determined as the future of the candidate target in income section
Earning rate.
8. a kind of trend prediction device based on artificial intelligence, which is characterized in that described device includes:
Determining module, for determining candidate target to be predicted；
Extraction module, for extracting the historical yield information and history public feelings information of the candidate target；
Input module, for the historical yield information and the history public feelings information to be input to Study on Stock Prediction Model, wherein
The Study on Stock Prediction Model is obtained using Encode-Decode from encryption algorithm training；
Prediction module, for predicting the future profits information of the candidate target using the Study on Stock Prediction Model.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 7 the method when executing the computer program.
10. a kind of computer storage medium, is stored thereon with computer program, which is characterized in that the computer program is located
The step of reason device realizes method described in any one of claims 1 to 7 when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910534829.2A CN110414710A (en) | 2019-06-20 | 2019-06-20 | Trend forecasting method and device based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910534829.2A CN110414710A (en) | 2019-06-20 | 2019-06-20 | Trend forecasting method and device based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110414710A true CN110414710A (en) | 2019-11-05 |
Family
ID=68359379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910534829.2A Pending CN110414710A (en) | 2019-06-20 | 2019-06-20 | Trend forecasting method and device based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110414710A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178498A (en) * | 2019-12-09 | 2020-05-19 | 北京邮电大学 | Stock fluctuation prediction method and device |
| CN111985704A (en) * | 2020-08-11 | 2020-11-24 | 上海华力微电子有限公司 | Method and device for predicting failure rate of wafer |
| CN113159445A (en) * | 2021-05-07 | 2021-07-23 | 朱小波 | Crime information prediction method and device and electronic equipment |
| CN113269259A (en) * | 2021-05-31 | 2021-08-17 | 北京邮电大学 | Target information prediction method and device |
| WO2023109025A1 (en) * | 2021-12-15 | 2023-06-22 | 北京达佳互联信息技术有限公司 | Delivery information processing method, and resource prediction model training method and apparatus |
-
2019
- 2019-06-20 CN CN201910534829.2A patent/CN110414710A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178498A (en) * | 2019-12-09 | 2020-05-19 | 北京邮电大学 | Stock fluctuation prediction method and device |
| CN111178498B (en) * | 2019-12-09 | 2023-08-22 | 北京邮电大学 | Stock fluctuation prediction method and device |
| CN111985704A (en) * | 2020-08-11 | 2020-11-24 | 上海华力微电子有限公司 | Method and device for predicting failure rate of wafer |
| CN113159445A (en) * | 2021-05-07 | 2021-07-23 | 朱小波 | Crime information prediction method and device and electronic equipment |
| CN113269259A (en) * | 2021-05-31 | 2021-08-17 | 北京邮电大学 | Target information prediction method and device |
| WO2023109025A1 (en) * | 2021-12-15 | 2023-06-22 | 北京达佳互联信息技术有限公司 | Delivery information processing method, and resource prediction model training method and apparatus |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110414710A (en) | Trend forecasting method and device based on artificial intelligence | |
| Zhou et al. | Stock market prediction on high-frequency data using generative adversarial nets. | |
| CN112819604A (en) | Personal credit evaluation method and system based on fusion neural network feature mining | |
| CN114493755B (en) | Self-attention sequence recommendation method fusing time sequence information | |
| CN110415086A (en) | Intelligence financing recommended method based on user's Continuous behavior sequence signature | |
| CN110781401A (en) | Top-n project recommendation method based on collaborative autoregressive flow | |
| CN112116448A (en) | Accurate marketing construction method for three-agriculture client portrait | |
| CN111489259A (en) | Stock market risk prediction intelligent implementation method based on deep learning | |
| CN113313316A (en) | Method and device for outputting prediction data, storage medium and electronic equipment | |
| CN115358838A (en) | Credit time series data modeling method and device based on convolutional neural network | |
| CN116739787B (en) | Transaction recommendation method and system based on artificial intelligence | |
| EP1107157A2 (en) | System and method for performing predictive analysis | |
| CN110390041B (en) | Online learning method and device and computer readable storage medium | |
| Cortez et al. | Time series forecasting by evolutionary neural networks | |
| CN115809928A (en) | Project investment decision-making method and system | |
| CN115883424A (en) | Method and system for predicting traffic data between high-speed backbone networks | |
| CN112163704B (en) | High-quality supplier prediction method for building material tender platform | |
| CN115393098A (en) | Financing product information recommendation method and device | |
| CN113033090B (en) | Push model training method, data push device and storage medium | |
| Cortez et al. | Evolving time series forecasting neural network models | |
| CN114969511A (en) | Content recommendation method, device and medium based on fragments | |
| Banditvilai et al. | Forecasting Models for Thailand’s Electrical Appliances Export Values | |
| Hargreaves | An automated stock investment system using machine learning techniques: an application in Australia | |
| CN111882339A (en) | Prediction model training and response rate prediction method, device, equipment and storage medium | |
| Hamzah et al. | Forecasting Nestle Stock Price by using Brownian Motion Model during Pandemic Covid-19 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191105 |
|
| RJ01 | Rejection of invention patent application after publication |