United States Patent	8,577,815
Barajas ,   et al.	November 5, 2013

---

Method and system for concurrent event forecasting

Download: 8,577,815.pdf

[text]

 

AbstractA method and system for characterizing, detecting, and predicting or forecasting multiple target events from a past history of these events includes compressing temporal data streams into self-organizing map (SOM) clusters, and determining trajectories of the temporal streams via the clusters to predict the multiple target events. The system includes an evolutionary multi-objective optimization (EMO) module for processing the temporal data streams, which are obtained from a plurality of heterogeneous domains; a SOM module for characterizing the temporal data streams into self-organizing map clusters; and a target event prediction (TEP) module for generating prediction models of the map clusters. The SOM module employs a vector quantization method that places a set of vectors on a low-dimensional grid in an ordered fashion. The prediction models each include trajectories of the temporal data streams, and the system predicts the multiple target events using the trajectories.

---

Inventors:	Barajas; Leandro G. (Troy, MI), Cho; Youngkwan (Los Angeles, CA), Srinivasa; Narayan (Oak Park, CA)
Applicant:	Name City State Country Type Barajas; Leandro G. Cho; Youngkwan Srinivasa; Narayan Troy Los Angeles Oak Park MI CA CA US US US
Assignee:	GM Global Technology Operations LLC (Detroit, MI)
Family ID:	43899232
Appl. No.:	12/604,606
Filed:	October 23, 2009

---

A method for simultaneously characterizing, detecting, and predicting multiple target events comprising: processing a plurality of temporal data streams of past occurrences of the target events using an evolutionary multi-objective optimization (EMO) module of a controller, wherein the temporal data streams are obtained asynchronously with respect to each other from a plurality of heterogeneous domains, and wherein at least some of the plurality of heterogeneous domains are of an incompatible format with the other heterogeneous domains; characterizing the temporal data streams from the EMO into map clusters via a vector quantization method using a self-organizing map (SOM) module of the controller, wherein the SOM module is adapted to map the temporal data streams from a high-dimensional space to a one-dimensional space, which places vectors one a one-dimension grid in an ordered fashion; generating prediction models of the map clusters using a target event prediction (TEP) module of the controller, wherein the prediction models each include trajectories of the temporal data streams that are represented as a string of events in the one-dimensional space; and estimating a future occurrence of the multiple target events using the trajectories from the TEP module.