【摘要】：層次多標(biāo)簽分類方法,依據(jù)標(biāo)簽之間的相關(guān)性組織成層次結(jié)構(gòu),并將這種層次結(jié)構(gòu)作為一種監(jiān)督信息,從而更好地解決多標(biāo)簽分類問(wèn)題.在層次多標(biāo)簽分類問(wèn)題中常用的方法有兩種,一種可稱為損失無(wú)關(guān)方法,另一種可稱為損失敏感方法.對(duì)于損失敏感方法,常用的損失函數(shù)有HMC-loss,該損失函數(shù)可對(duì)假正和假負(fù)兩種錯(cuò)誤給予不同的權(quán)重,并將層次信息添加到損失函數(shù)當(dāng)中.當(dāng)利用HMC-loss預(yù)測(cè)時(shí),盡管得到的損失值是理想的,但實(shí)際預(yù)測(cè)的標(biāo)簽數(shù)卻遠(yuǎn)多于真實(shí)的標(biāo)簽數(shù).另外,層次信息的引入會(huì)對(duì)標(biāo)簽結(jié)點(diǎn)的決策順序產(chǎn)生不利影響.針對(duì)這些問(wèn)題,首先提出改進(jìn)的損失函數(shù)IMH-loss,其次使用貝葉斯決策理論,提出了一種貝葉斯風(fēng)險(xiǎn)隨決策過(guò)程可變的層次多標(biāo)簽分類方法.在真實(shí)數(shù)據(jù)集上的實(shí)驗(yàn)結(jié)果表明,該方法在保證召回率的同時(shí),提升了標(biāo)簽預(yù)測(cè)精度.
[Abstract]:The hierarchical multi-label classification method is organized into a hierarchical structure according to the correlation between labels, and the hierarchical structure is regarded as a kind of supervisory information to solve the problem of multi-label classification better. There are two commonly used methods in hierarchical multi-label classification, one is loss-independent and the other is loss-sensitive. For loss-sensitive methods, the commonly used loss function is HMC-loss.This loss function can give different weights to false positive and false negative errors, and add hierarchical information to the loss function. When using HMC-loss prediction, although the loss value obtained is ideal, the actual number of tags predicted is much more than the actual number of tags. In addition, the introduction of hierarchical information will adversely affect the decision order of label nodes. To solve these problems, an improved loss function (IMH-lossing) is proposed, and then a hierarchical multi-label classification method of Bayesian risk variable with the decision process is proposed by using Bayesian decision theory. The experimental results on real data sets show that the proposed method not only guarantees the recall rate, but also improves the label prediction accuracy.
【作者單位】： 山西大學(xué)計(jì)算機(jī)與信息技術(shù)學(xué)院;山西大學(xué)計(jì)算智能與中文信息處理教育部重點(diǎn)實(shí)驗(yàn)室;
【基金】：國(guó)家自然科學(xué)基金(61632011,61272095,61432011,U1435212,61573231,61672331)
【分類號(hào)】：TP301.6

【相似文獻(xiàn)】

相關(guān)期刊論文 前3條

1 文春勇;朱信忠;徐慧英;趙建民;;基于最小風(fēng)險(xiǎn)的貝葉斯決策理論相關(guān)反饋方法[J];計(jì)算機(jī)應(yīng)用研究;2009年03期

2 李小光;;混合損失函數(shù)支持向量回歸機(jī)的性能研究[J];西北大學(xué)學(xué)報(bào)(自然科學(xué)版);2011年02期

3 路緒清;唐杰;李涓子;蔡月茹;;基于關(guān)鍵詞抽取的hypertext自動(dòng)建立方法[J];計(jì)算機(jī)科學(xué);2005年02期

相關(guān)會(huì)議論文 前2條

1 謝世斌;劉萬(wàn)春;朱玉文;;基于貝葉斯決策理論和主成分分析的人臉識(shí)別[A];第三屆全國(guó)數(shù)字成像技術(shù)及相關(guān)材料發(fā)展與應(yīng)用學(xué)術(shù)研討會(huì)論文摘要集[C];2004年

2 吳佳金;楊志豪;林原;林鴻飛;;基于改進(jìn)Pairwise損失函數(shù)的排序?qū)W習(xí)方法[A];第六屆全國(guó)信息檢索學(xué)術(shù)會(huì)議論文集[C];2010年

，

本文編號(hào)：2168908