Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Sequence prediction model training method, prediction system, prediction method and medium

A training method and sequence prediction technology, applied in the field of artificial intelligence, can solve problems such as inability to process asynchronous time series information

Pending Publication Date: 2020-11-10
点内(上海)生物科技有限公司
View PDF7 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the deficiency that the existing technology cannot process asynchronous time series information, the present invention proposes a training method of a model for predicting asynchronous time series information, and a system and forecasting method capable of processing and predicting asynchronous time series information, To more accurately process and predict asynchronous time series information

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Sequence prediction model training method, prediction system, prediction method and medium
  • Sequence prediction model training method, prediction system, prediction method and medium
  • Sequence prediction model training method, prediction system, prediction method and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0065] In the present invention, the generated periodic signal is used to train, verify and test the predictive model of the present invention, and to use the tested model to predict, to illustrate the present invention, the model training process is as follows figure 1 shown.

[0066] For randomly encountered asynchronous time series, a periodic signal is generated according to the following formula:

[0067]

[0068] Among them, X t is the value of the signal at time t, N is the number of random periodic signals, α j , ω j ，b j ，β j is the periodic signal parameter of 0-5 random continuous distribution, ε is the standard normal distribution noise, and η is the noise amplification factor. In this example, N=10, η=0.5, and 10,000 groups of sequences with a length of 11 were generated according to the above random formula. Except for the last two time points, the time interval is fixed at 1, and the time interval of other time points in the sequence is 0 -2 random...

Embodiment 2

[0078] In this embodiment, the present invention provides an application in the field of medical treatment to predict and analyze the therapeutic effect of PD-1 inhibitors on patients with lung cancer.

[0079] This example predicts and analyzes the therapeutic effect of PD-1 inhibitors on patients with lung cancer. figure 2 with image 3 The system block diagram shown shows that the training process of the prediction model is as follows:

[0080] Data collection: The basic clinical information of 99 second-line lung cancer patients before and during immunotherapy with PD-1 inhibitors (including patient gender, age, smoking history, tumor family history, radiotherapy history, pathological type, cancer stage), Multiple CT images taken at different times and expert segmentation and labeling (the radiologist marks the main lesion on the CT according to the medical records for accurate segmentation, and is confirmed and marked by a senior oncologist), laboratory test information...

Embodiment 3

[0110] This embodiment is aimed at the short-term prediction of securities prices in secondary market securities transactions, see Figure 4 , the training process of the prediction model is as follows:

[0111] Collect complete securities transaction information in the market within a specified time period, including transaction price, transaction volume, and transaction time as a training set;

[0112] Normalize the collected data, divide the securities transaction price by 50, and divide the transaction volume by 50, and organize them into a time series X sec ∈R T×2 , corresponding to the transaction time point T trade ∈R T . Add feature 1 at the point in time when each transaction price is greater than the average price of n (n∈(5,10,30,60,180)) days in the past, and less than or equal to the new feature 0, to get feature X comp ∈R T×5 , put X sec and x comp Splicing in the feature dimension to get the feature X ts ∈R T×7 .

[0113] input as X ts and T trade ,...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a sequence prediction model training method, a prediction system, a prediction method and a medium. The model training method comprises the steps of taking asynchronous time sequence information containing true values as a training set, performing preprocessing, and recording time information of asynchronous time sequence features; performing fusion dimension reduction oneach asynchronous time sequence feature in a time dimension; wherein if the feature after dimension reduction is the content to be predicted, the feature is a prediction result; if the feature is notthe content which needs to be predicted, splicing all the asynchronous time series features except the true value after dimension reduction to obtain a total feature, modeling the total feature, and carrying out feature prediction at a certain time point to obtain a prediction result; and calculating loss in combination with the prediction result and the true value, and training a prediction model. Multi-modal data can be fused for modeling, the model expression capability is greatly enhanced compared with a single-modal model, and a more accurate prediction result is obtained by combining historical data.

Description

technical field [0001] The invention relates to an intelligent information prediction system, in particular to a training method for a prediction model of asynchronous time series, a prediction system, a prediction method and a medium, and belongs to the technical field of artificial intelligence. Background technique [0002] In recent years, the trend prediction of time series data through intelligent information analysis technology has played an important role in many industries. Different patents have different improvement contents, such as the Chinese authorized patent CN103500272B (a fuzzy information granulation method for time series trend prediction ), using the fuzzy information granulation method to predict the trend of time series, which mainly involves the improvement of preprocessing of time series data; another example is the Chinese authorized patent CN105095613B (a method and device for forecasting based on sequence data), through the data The method of esta...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06N20/00G06N20/10G06N3/04G06N3/08G06K9/62G06F16/2458G06Q10/04G06Q40/04G16H50/30G06T7/00G06T7/174
CPCG06N20/00G06N20/10G06N3/08G06F16/2474G06F16/2477G06Q10/04G06Q40/04G16H50/30G06T7/0012G06T7/174G06T2207/10081G06T2207/20081G06T2207/20084G06T2207/30096G06N3/044G06N3/045G06F18/213G06F18/253
Inventor 陈嘉骏杨健程葛亮
Owner 点内(上海)生物科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com