本文選題：結(jié)構(gòu)化稀疏學(xué)習(xí) + 算法　； 參考：《Frontiers of Information Technology & Electronic Engineering》2017年04期

【摘要】：稀疏學(xué)習(xí)由于其簡約特性和計(jì)算優(yōu)勢而獲得了越來越多的關(guān)注,在具有稀疏性的條件下,許多計(jì)算問題可以在實(shí)踐中得到有效的處理。而結(jié)構(gòu)化稀疏學(xué)習(xí)則進(jìn)一步將結(jié)構(gòu)信息進(jìn)行編碼,在多個研究領(lǐng)域取得成功。隨著各類型結(jié)構(gòu)的發(fā)現(xiàn),人們相繼提出了各種結(jié)構(gòu)化正則函數(shù)。這些正則函數(shù)通過利用特定的結(jié)構(gòu)信息極大提高了稀疏學(xué)習(xí)算法的性能。在本文中,我們從想法、形式化、算法和應(yīng)用等方面系統(tǒng)的回顧了結(jié)構(gòu)化稀疏學(xué)習(xí)。我們將這些算法置于最小化損失函數(shù)和懲罰函數(shù)的統(tǒng)一框架中,總結(jié)了算法的開源軟件實(shí)現(xiàn),并比較了典型優(yōu)化算法解決結(jié)構(gòu)化稀疏學(xué)習(xí)問題時的計(jì)算復(fù)雜度。在實(shí)驗(yàn)中,我們給出了無監(jiān)督學(xué)習(xí)在結(jié)構(gòu)化信號恢復(fù)和層次化圖像重建中的應(yīng)用,以及具有圖結(jié)構(gòu)引導(dǎo)的邏輯回歸的在監(jiān)督學(xué)習(xí)中的應(yīng)用。
[Abstract]:Sparse learning has attracted more and more attention due to its simplicity and computational advantages. Under the condition of sparsity, many computing problems can be effectively dealt with in practice. Structured sparse learning further encodes structural information, which is successful in many research fields. With the discovery of various types of structures, various structured regular functions have been proposed one after another. These regular functions greatly improve the performance of sparse learning algorithms by using specific structural information. In this paper, we systematically review structured sparse learning from ideas, formalization, algorithms and applications. We put these algorithms under the unified framework of minimization loss function and penalty function, summarize the open source software implementation of the algorithm, and compare the computational complexity of the typical optimization algorithm in solving the structured sparse learning problem. In the experiment, we give the application of unsupervised learning in structured signal restoration and hierarchical image reconstruction, and the application of logical regression with graph structure guidance in supervised learning.
【作者單位】： College
【基金】：Project supported by the National Natural Science Foundation of China(No.61303264)
【分類號】：TP181

【相似文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前4條

1 袁洪芬;k-超正則函數(shù)及其相關(guān)函數(shù)的性質(zhì)[D];河北師范大學(xué);2006年

2 苑文法;有界正則函數(shù)導(dǎo)數(shù)及系數(shù)估計(jì)[D];西北工業(yè)大學(xué);2002年

3 李小伶;Clifford分析中的k-正則函數(shù)[D];河北師范大學(xué);2008年

4 曹南斌;Clifford分析中無界域上雙正則函數(shù)的Plemelj公式[D];河北師范大學(xué);2002年

，

本文編號：2073029