本文選題：深度學(xué)習(xí)　切入點(diǎn)：人工智能　出處：《物理》2017年09期 　論文類型：期刊論文

【摘要】：深度學(xué)習(xí)是一類通過多層信息抽象來學(xué)習(xí)復(fù)雜數(shù)據(jù)內(nèi)在表示關(guān)系的機(jī)器學(xué)習(xí)算法。近年來,深度學(xué)習(xí)算法在物體識(shí)別和定位、語音識(shí)別等人工智能領(lǐng)域,取得了飛躍性進(jìn)展。文章將首先介紹深度學(xué)習(xí)算法的基本原理及其在高能物理計(jì)算中應(yīng)用的主要?jiǎng)訖C(jī)。然后結(jié)合實(shí)例綜述卷積神經(jīng)網(wǎng)絡(luò)、遞歸神經(jīng)網(wǎng)絡(luò)和對(duì)抗生成網(wǎng)絡(luò)等深度學(xué)習(xí)算法模型的應(yīng)用。最后,文章將介紹深度學(xué)習(xí)與現(xiàn)有高能物理計(jì)算環(huán)境結(jié)合的現(xiàn)狀、問題及一些思考。
[Abstract]:Depth learning is a kind of machine learning algorithm which can learn the internal representation of complex data by multi-layer information abstraction. In recent years, depth learning algorithm has been applied in artificial intelligence fields such as object recognition and location, speech recognition and so on. This paper first introduces the basic principle of depth learning algorithm and the main motivation of its application in high energy physics computation, and then summarizes the convolution neural network with examples. The application of depth learning algorithm models such as recurrent neural networks and confrontation generating networks. Finally, this paper will introduce the current situation, problems and some thoughts about the combination of depth learning with the existing high energy physics computing environment.
【作者單位】： 中國科學(xué)院高能物理研究所計(jì)算中心;
【分類號(hào)】：O572;TP18
，

本文編號(hào)：1602626