一個(gè)完整的AI系統(tǒng)不僅僅包括處理數(shù)據(jù)的計(jì)算部分，還包括保存數(shù)據(jù)的存儲(chǔ)部分和傳輸數(shù)據(jù)的網(wǎng)絡(luò)部分。只有各部分互相協(xié)調(diào)一致、消除了系統(tǒng)的短板，整個(gè)系統(tǒng)才能發(fā)揮最大的效能。英特爾?公司推出了各種各樣的硬件產(chǎn)品，這些產(chǎn)品共同構(gòu)建用于不同場(chǎng)景的高效能AI系統(tǒng)。

An AI system includes not only computing part for data processing, also storage for data saving and network for data transferring. The AI system cannot achieve maximum efficiency unless each part works coordinately to eliminate bottleneck. Intel? has various products that can construct high efficiency AI systems for different usage modes.

AI技術(shù)的發(fā)展離不開(kāi)軟件層面的優(yōu)化，該系列課程介紹英特爾?在基礎(chǔ)加速庫(kù)，編程語(yǔ)言和特定深度學(xué)習(xí)開(kāi)源框架所做的優(yōu)化工作。通過(guò)軟件層面的優(yōu)化，將底層硬件的加速性能完全發(fā)揮出來(lái)，為客戶(hù)提供更加強(qiáng)大的計(jì)算力。

Software optimization is very crucial to AI technology development. This session includes optimization on MKL&MKL-DNN, programming language and specific deep learning framework. It will fully unleash hardware accelerate feature and provide more powerful compute to customers.

深度學(xué)習(xí)與機(jī)器學(xué)習(xí)是一個(gè)實(shí)踐性非常強(qiáng)的領(lǐng)域。本分會(huì)面向開(kāi)發(fā)者分享英特爾?在深度學(xué)習(xí)與機(jī)器學(xué)習(xí)領(lǐng)域的實(shí)戰(zhàn)經(jīng)驗(yàn)，包括如何在人工智能軟件框架中實(shí)現(xiàn)高性能的應(yīng)用開(kāi)發(fā)，如何構(gòu)建大規(guī)模的人工智能訓(xùn)練集群，同時(shí)也將介紹英特爾?架構(gòu)數(shù)據(jù)中心平臺(tái)在具體應(yīng)用案例中的應(yīng)用效果及技巧

Deep Learning and Machine Learning are practical. This track is to share Intel's experience in deep learning and machine learning practices on Intel ? Architecture, including how to develop high-performance AI application with frameworks; ?how to build large-scale Deep Learning training clusters, as well as the best known methods from AI case studies.