Predicate |
Object |
assignee |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentassignee/MD5_d6a6f422b091ba12ea61d4adbf1b0e8e |
classificationCPCInventive |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06F16-9024 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N3-045 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N3-08 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06N5-041 http://rdf.ncbi.nlm.nih.gov/pubchem/patentcpc/G06F16-2474 |
classificationIPCInventive |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06F16-901 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N5-04 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06F16-2458 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-08 http://rdf.ncbi.nlm.nih.gov/pubchem/patentipc/G06N3-04 |
filingDate |
2019-12-27-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
inventor |
http://rdf.ncbi.nlm.nih.gov/pubchem/patentinventor/MD5_b1669215f878f2c3c8b5a61588fa3ab0 http://rdf.ncbi.nlm.nih.gov/pubchem/patentinventor/MD5_796b4ffb004a570fb0971bf1a87f2b5e |
publicationDate |
2020-04-28-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
publicationNumber |
CN-111079931-A |
titleOfInvention |
A state-space probabilistic multi-time series prediction method based on graph neural network |
abstract |
The invention discloses a state space probabilistic multiple time series prediction method based on graph neural network, comprising: (1) acquiring multiple time series and preprocessing the time series to construct a training set and constructing a graph structure at the same time; (2) Construct a generative model for generating prior distribution and time series of hidden states based on graph neural network and multilayer perceptron, and construct an inference network for generating approximate posterior distribution of hidden states based on graph neural network and recurrent neural network; (3) Construct a loss function according to the prior distribution and approximate posterior distribution of the hidden state, and optimize the parameters of the generative model and the inference network with the goal of maximizing the loss function; (4) When applying, use the inference network to obtain the sequence to be predicted. The hidden state estimation at the latest moment is used to calculate the prior distribution of the hidden state based on the hidden state estimation at the latest moment by using the generative model, and then the estimated value of the time series observation is obtained according to the prior distribution of the hidden state. |
isCitedBy |
http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112783940-B http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112783940-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112465054-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114915444-B http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114915444-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112085866-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/WO-2022165602-A1 http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114519405-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-113159409-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-115640337-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-114519405-B http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112330079-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112966193-B http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-112966193-A http://rdf.ncbi.nlm.nih.gov/pubchem/patent/CN-113379156-A |
priorityDate |
2019-12-27-04:00^^<http://www.w3.org/2001/XMLSchema#date> |
type |
http://data.epo.org/linked-data/def/patent/Publication |