發(fā)布時(shí)間：2018-05-09 19:59

  本文選題：遷移學(xué)習(xí) + 域適應(yīng)��； 參考：《南京郵電大學(xué)》2017年碩士論文

【摘要】：隨著大數(shù)據(jù)時(shí)代的到來(lái),人們可以更加容易地獲得大量數(shù)據(jù)。此外,由于機(jī)器學(xué)習(xí)領(lǐng)域不斷的發(fā)展,如何讓計(jì)算機(jī)具有舉一反三的能力,如何使大量數(shù)據(jù)可以更好地發(fā)揮作用,這些問(wèn)題均變得非常實(shí)際和有價(jià)值。為了解決這些問(wèn)題,遷移學(xué)習(xí)被提出并越來(lái)越受到人們的重視。在常規(guī)機(jī)器學(xué)習(xí)中有一個(gè)重要假設(shè),即訓(xùn)練所需的數(shù)據(jù)和目標(biāo)所需的數(shù)據(jù)必須具有相同的分布或者來(lái)自相同的特征空間。在現(xiàn)實(shí)生活中,這一假設(shè)是很難實(shí)現(xiàn)的。具體來(lái)說(shuō),對(duì)于一個(gè)分類(lèi)問(wèn)題,如果訓(xùn)練集的樣本和目標(biāo)集的樣本不具有相同的分布,這就可以理解為源域與目標(biāo)域不具有相同的特征空間。傳統(tǒng)的解決方法是通過(guò)搜集更多的數(shù)據(jù)去模擬目標(biāo)域的分布,但是這樣代價(jià)極大。遷移學(xué)習(xí)是一種有效縮短兩個(gè)域之間“距離”的方法,這樣就可以接近傳統(tǒng)機(jī)器學(xué)習(xí)中的假設(shè),通過(guò)源數(shù)據(jù)訓(xùn)練出適合目標(biāo)數(shù)據(jù)的模型。論文對(duì)遷移學(xué)習(xí)算法做了詳細(xì)的研究和總結(jié)。具體工作如下:(1)對(duì)現(xiàn)有遷移學(xué)習(xí)算法做了全面的總結(jié)和研究,并對(duì)各種方法的性質(zhì)進(jìn)行了比較,對(duì)各種算法適合使用的領(lǐng)域也做了詳細(xì)的闡述。(2)具體研究了遷移成分分析算法、測(cè)地流核算法、子空間對(duì)齊算法、最大獨(dú)立域適應(yīng)算法、信息理論學(xué)習(xí)算法等常用的算法,同時(shí)還研究了基于深度學(xué)習(xí)的遷移學(xué)習(xí)方法。(3)將遷移成分分析算法、測(cè)地流核算法、子空間對(duì)齊算法、最大獨(dú)立域適應(yīng)算法、信息理論學(xué)習(xí)算法以及深度學(xué)習(xí)相關(guān)的遷移學(xué)習(xí)算法運(yùn)用到人臉表情識(shí)別當(dāng)中,并使用不同的數(shù)據(jù)庫(kù)對(duì)遷移學(xué)習(xí)算法在人臉表情識(shí)別中做了實(shí)驗(yàn)和比較。遷移學(xué)習(xí)有效解決了人臉表情識(shí)別當(dāng)中源域與目標(biāo)域不具有相同特征空間時(shí)的分類(lèi)問(wèn)題。(4)對(duì)遷移學(xué)習(xí)中存在的問(wèn)題進(jìn)行了分析,并對(duì)未來(lái)的發(fā)展進(jìn)行了展望。
[Abstract]:With the arrival of big data's era, people can easily access a large number of data. In addition, due to the continuous development in the field of machine learning, how to make the computer have the ability to draw inferences from one another and how to make a large number of data work better, these problems have become very practical and valuable. In order to solve these problems, transfer learning has been put forward and paid more and more attention to. There is an important assumption in conventional machine learning that the training data and the target data must have the same distribution or from the same feature space. In real life, this assumption is difficult to achieve. Specifically, for a classification problem, if the samples of the training set and the target set do not have the same distribution, it can be understood that the source domain and the target domain do not have the same feature space. The traditional solution is to simulate the distribution of target domain by collecting more data, but this is costly. Transfer learning is an effective method to shorten the "distance" between two domains, so that it can approach the hypothesis of traditional machine learning and train a model suitable for target data through source data. In this paper, the transfer learning algorithm is studied and summarized in detail. The specific work is as follows: (1) A comprehensive summary and study of the existing transfer learning algorithms are made, and the properties of the various methods are compared, and the fields in which the algorithms are suitable for use are also elaborated in detail. (2) the migration component analysis algorithms are studied in detail. Geodesic flow accounting method, subspace alignment algorithm, maximum independent domain adaptation algorithm, information theory learning algorithm and other commonly used algorithms. At the same time, the migration component analysis algorithm, geodesic flow accounting method, is also studied. Subspace alignment algorithm, maximum independent domain adaptation algorithm, information theory learning algorithm and transfer learning algorithm related to depth learning are applied to face expression recognition. Different databases are used to compare the transfer learning algorithm in facial expression recognition. Migration learning effectively solves the classification problem in facial expression recognition where the source domain and the target domain do not have the same feature space.
【學(xué)位授予單位】：南京郵電大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類(lèi)號(hào)】：TP391.41;TP181

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王志良,陳鋒軍,薛為民;人臉表情識(shí)別方法綜述[J];計(jì)算機(jī)應(yīng)用與軟件;2003年12期

2 孫蔚;王波;;人臉表情識(shí)別綜述[J];電腦知識(shí)與技術(shù);2012年01期

3 楊梅娟;;人臉表情識(shí)別綜述[J];甘肅科技;2006年04期

4 劉曉e，

本文編號(hào)：1867218