Hulu智能算法技術架構介紹_第1頁
Hulu智能算法技術架構介紹_第2頁
Hulu智能算法技術架構介紹_第3頁
Hulu智能算法技術架構介紹_第4頁
Hulu智能算法技術架構介紹_第5頁
已閱讀5頁,還剩22頁未讀 繼續免費閱讀

下載本文檔

版權說明:本文檔由用戶提供并上傳,收益歸屬內容提供方,若內容存在侵權,請進行舉報或認領

文檔簡介

1、算法無處不在Make Algorithms UbiquitousHulu智能算法技術架構介紹Machine Learning Algorithms in Industry 機器學習算法企業應用場景ML Algorithms in Hulu 無處不在的機器學習算法在HuluDemocratize ML In Hulu Hulu ML算法案例研究AI Platform In Hulu Hulu人工智能平臺Future Opportunities 展望未來Background 行業背景ML centric businessface recognition self-driving carmedica

2、l image analysis?consumer businessonline entertainment online gametravel / tourist application e-commerceAIis everywhereML Algorithm 機器學習算法的技術定義ProgramAlgorithmvsvsa collection of instructionsthe logic to solve a class of problemsDATA + MODELMachine Learning Algorithmthe method to learn the logicTre

3、nd ML算法應用的趨勢垂直= 水平特型= 通用 專家= 平民化specific scenarios 特定場景predictiontargetingrecommendationubiquitous 無處不在Machine Learning Algorithms in Industry 機器學習算法企業應用場景ML Algorithms in Hulu 無處不在的機器學習算法在HuluDemocratize ML In Hulu Hulu ML算法案例研究AI Platform In Hulu Hulu人工智能平臺Future Opportunities 展望未來HuluThe Premier

4、Digital Video CompanyBest Quality Content Video-On-Demand and Live BroadcastingML Algorithms in HuluML算法在Hulu的應用ViewersContentAdvertisingAlgorithms for Viewers 面向用戶的算法cover story smart start recommendationreasonlayout managementAlgorithms for Viewers 面向用戶的算法onboardingcold start convert-to-payHome Pa

5、gecover story smart startrecommendation reason layout managementadsauto-play QoS to QoEAlgorithms for Content and Advertising 面向內容和廣告的算法ContentAdvertisingcontent valuation video understandingcontent based video encodingrich targeting inventory predictionMachine Learning Algorithms in Industry 機器學習算法

6、企業應用場景ML Algorithms in Hulu 無處不在的機器學習算法在HuluDemocratize ML In Hulu Hulu ML 算法案例研究垂直 Reco水平 Video Content Understanding水平到垂直 Contextual + AdsAI Platform In Hulu Hulu人工智能平臺Future Opportunities 展望未來Recommendations: StoreShelfItem-based Collaborative FilteringMatrix FactorizationDeep Neural NetworkDNN-b

7、ased Relevance AlgorithmAuto-Play, Time Series ModelTimelineWatchWatchWatchWatchWatchWatchSearchSearchBrowseBrowseBrowseWatch BrowseWatchSearchWatch WatchWatchNew Show?Try it outBingeRecurrent Neural NetworkVideo Content Understanding: From Inside 從內部理解視頻內容Video derived meta-data 元數據抽取和發現Ads break,

8、end credit, .BreaksVideo summary, .ArtsActor, car, animal, .ObjectsCooking, wedding, EventsCeremony(0.91) Wedding (0.82) Bride(0.72) Event(0.38) Groom(0.36) Woman(0.31) Meal(0.28) Marriage(0.27) Dress(0.25)Wedding reception(0.18)Video Content Understanding: From Inside 從內部理解視頻內容Live Channel Thumbnai

9、l Preview直播頻道預覽Contextual Ads: 視頻和廣告的結合Deliver semantic related or visual similar ads to viewers during playback視覺連貫的廣告語義相關的廣告Use video derived data to match ads and contentLabels of Target ScenesScene-Level Auto-TaggingMatchingAds ProviderHulu Content PairsAds Deliveryscore:0.86Contextual Ads: 視頻和廣

10、告的結合視覺連貫的廣告語義相關的廣告Contextual Ads: 視頻和廣告的結合視覺連貫的廣告語義相關的廣告Machine Learning Algorithms in Industry 機器學習算法企業應用場景ML Algorithms in Hulu 無處不在的機器學習算法在HuluDemocratize ML In Hulu Hulu ML算法案例研究AI Platform In Hulu Hulu人工智能平臺Future Opportunities 展望未來Network / DevOps / Big DataInfrastructureDataFeatureModelApplic

11、ationGPU ClusterFrame StorageUser Data Static / DynamicContent DataAudio / Video / Text, MetadataShared FeaturesShared ML ModelsAll ML ApplicationsAI Platform 人工智能平臺Application Specific ML ModelsAds Data Audio / Video / TextFeatureData InfrastructureModelApplicationAI Platform 人工智能平臺Convert-to-PayOn

12、boardingRecommendationContent ValuationAuto-PlayCold StartQoS to QoEAd TargetingApplication Specific ML ModelsUser Conversion ModelUser Engagement ModelImage UnderstandingUserUserKnowledge GraphContentContent PersonaEmbeddingSegmentsEmbeddingUser DataAuContent Dataxt,Ads Data Static / Dynamicdio / Video / TeAudio / Video / TextMetadataNetwork / DevOps / Big DataGPU ClusterFrame StorageComputational Mode 計算模式Offlinedays/hoursNearlineminutesOnlinemillisecondsmodeldataMachine Learning Algorithms in Industry 機器學習算法企業應用場景ML Algorithms in Hulu 無處不在的機

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯系上傳者。文件的所有權益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網頁內容里面會有圖紙預覽,若沒有圖紙預覽就沒有圖紙。
  • 4. 未經權益所有人同意不得將文件中的內容挪作商業或盈利用途。
  • 5. 人人文庫網僅提供信息存儲空間,僅對用戶上傳內容的表現方式做保護處理,對用戶上傳分享的文檔內容本身不做任何修改或編輯,并不能對任何下載內容負責。
  • 6. 下載文件中如有侵權或不適當內容,請與我們聯系,我們立即糾正。
  • 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。

評論

0/150

提交評論