Deep learning sample labeling method based on big data

Abstract

The invention provides a deep learning sample labeling method based on big data. The method comprises the following steps: receiving user labeling input related to a first group of sample objects in asample library; training a preference prediction model comprising a weight vector, the weight vector comprising a weighted value of each of a plurality of features associated with a sample library, the sample library comprising a first set of sample objects presented to the user, training the weighted value of each feature with the received user annotation input; selecting a second set of sampleobjects to be provided to the user, the second set of sample objects providing more priori knowledge obtained from the user annotation input relative to other unidentified sample objects in the samplelibrary; and according to the trained preference prediction model, pushing a preset number of preference objects to be provided for the user. According to the deep learning sample labeling method based on the big data, the interest and preference information of the user is mined based on the inherent interaction operation of the user, the sample labeling efficiency and the deep learning effect are improved, and therefore accurate content recommendation can be better achieved.

Description

technical field [0001] The present invention relates to machine learning, in particular to a deep learning sample labeling method based on big data. Background technique [0002] In the era of information overload, personality-based recommendations are becoming more and more important. Traditional technologies are based on the analysis of users' explicit feedback data. For example, online education platforms request users to rate objects such as courses and teachers. However, in fact, the amount of explicit feedback data and application scenarios are relatively limited. It takes a lot of time for the user, resulting in a poor experience. In reality, a large amount of user interest information is often hidden in the user's normal interactive operations. When mining user preferences, if the label information is uncertain, it needs to be manually labeled or determined in advance, and then the machine learning model is trained through the complete data set, and then the object...

AI Technical Summary

Problems solved by technology

However, the manual labeling method consumes a lot of human resources, and there are human biases, resulting in low efficiency and poor effect of labeling information determination; thus, it is impossible to achieve accurate recommendations for users

For samples with a high imbalance rate, that is, there are labeled samples of minority and majority classes, the classification of the samples is easy to shift to the majority class.

Images