首页|Patent Issued for Training a classification model using labeled training data th at does not overlap with target classifications for the classification model (US PTO 11947632)
Patent Issued for Training a classification model using labeled training data th at does not overlap with target classifications for the classification model (US PTO 11947632)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
News editors obtained the following quote from the background information suppli ed by the inventors:“This disclosure relates generally to training a classifica tion model, and more specifically to training theclassification model using exi sting training data with labels that do not overlap with classes to be outputby the classification model.“Many online systems employ one or more machine learning models to analyze data. Different onlinesystems may use different machine learning models for differen t types of analysis or to analyze differenttypes of data. However, obtaining la beled training data for training one or more machine learning models isdifficul t for many online concierge systems. While data can be manually labeled to form training data for amachine learning model, creation of manually labeled trainin g data is tedious and expensive. Additionally,many manually labeled training da ta sets are generally unable to be reused for different types of analysisif the distribution of data used to generate the training data changes over time.”
BusinessCyborgsEmerging TechnologiesMachine LearningMaplebear Inc.
NETL
NSTL
