Transfer learning design method and system based on domain adaptation under multi-example multi-label framework

Abstract

The invention discloses a transfer learning design method and system based on domain adaptation under a multi-example multi-label framework. According to the invention, multi-example multi-label learning and transfer learning are unified into one framework, source domain data samples and target domain data samples are effectively utilized for correlation statistics, and the source domain samples can be effectively used in the learning of target domain tasks; the characteristics of a source domain data sample set and a target domain data set in RKHS are utilized, a two-step domain adaptation process formed by sample weighting and a sample selection mechanism based on clustering is utilized, so that the learning of target tasks has enough training samples weighted and selected from the source domain set; and a miFV algorithm is utilized to convert multi-examples into a single example, and the calculation cost problem of domain adaptation is solved.

Description

technical field [0001] The invention relates to a design method and system based on domain adaptation transfer learning under a multi-instance multi-marker framework, and belongs to the technical field of machine learning. Background technique [0002] Traditional supervised learning assumes that samples have only one kind of semantic information or category labels, but such assumptions are not consistent with many learning tasks in reality. For example, in a document classification task, a news report about smog may involve multiple semantic concepts such as "environment", "population", and "economy" at the same time; in an image recognition task, a photo of a city scene may simultaneously Contains semantic objects such as "building", "street", and "sunset". The extension of class labels makes it difficult to achieve good results in traditional supervised learning frameworks that only consider explicit and single semantics. Therefore, each training sample is assigned a se...

AI Technical Summary

Problems solved by technology

The present invention can well solve the problem of training samples, thereby effectively expanding the application range of multi-instance multi-label learning; the present invention completes the unification of multi-instance multi-label learning and migration learning, and solves the problem of sample collection under large-scale data The problem of different distributions; the present invention relaxes the assumption of traditional multi-instance multi-label supervised learning, improves the learning efficiency of multi-instance multi-label algorithm, and improves the accuracy of classification

Images