Optimized Conditional Random Field (CRF)-based algorithm capable of finding optimal solutions despite hidden or incomplete fields in data
Tony Jebara, Ph.D.
Specialized to conditional likelihood and thus able to handle hidden or incomplete informationImproved efficiency means that companies can train on larger datasets in less timeAbility to handle missing information also means training can occur in more realistic settings and across broader datasetsAllows for improved application in parallelization and cloud computing implementationHighly applicable to fields that are expanding rapidly (e.g., Genomics, HER)Can be applied to a very wide variety of problemsPatent information:Patent Pending (US 20120317060)Tech Ventures Reference: IR M11-115
Information retrieval, data mining, natural language processing, web mining, and many other applications that require an efficient and versatile discriminative probabilistic modelGenomics, transcriptomics, proteomics, electronic health records, and biomedical informatics in general are all fields that could greatly benefit from this technologyCompanies moving into cloud computing
Tony Jebara, Ph.D.
美国
