數據監管越來越嚴，我們還有什麼機會？

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"數據已經成爲當今社會不可或缺的生產要素。據"},{"type":"link","attrs":{"href":"https:\/\/theconversation.com\/the-worlds-data-explained-how-much-were-producing-and-where-its-all-stored-159964","title":null,"type":null},"content":[{"type":"text","marks":[{"type":"underline"}],"text":"IDC"}]},{"type":"text","text":"統計，2018 年，全球總共創建、捕獲、複製和消費了33 ZB （即33 萬億千兆字節）的數據量，2020 年增長到了59ZB，預計到2025 年將達到175ZB。隨着數據越來越多，人們越來越重視數據安全、隱私保護等問題，隱私計算因此得到了很大的關注。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"根據KPMG《隱私計算行業研究報告》，國內隱私計算市場規模將快速發展，三年後技術服務營收有望觸達100-200億人民幣的空間，甚至將撬動千億級的數據平臺運營收入空間。那當前隱私計算髮展情況如何，未來又有哪些機會呢？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"數據合作帶來更多收益"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"數據孤島導致沉睡的數據資產無法流通，難以實現商業價值。在"},{"type":"link","attrs":{"href":"https:\/\/www.clustar.ai\/","title":"xxx","type":null},"content":[{"type":"text","text":"星雲Clustar"}]},{"type":"text","text":" CEO 陳沫看來，數據合作帶來的收益遠大於要爲此付出的成本，安全的數據合作可以帶來更多收益。隱私計算就是要連接孤島，產生“1+1>2”的效果。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隱私計算目前落地應用相對成熟的領域是金融。金融行業是數據價值轉化率最高、並且對隱私要求也最高的行業之一。相對於其他行業，金融尤其是銀行業，在數據安全和隱私問題上考慮得要更早些。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"過去十年，我國積累的數據百分之七八十都在互聯網。相比之下，銀行擁有的數據維度是比較少的，這導致銀行做營銷時難以精準地找到匹配人羣。互聯網金融的興起與其擁有的海量數據也不無關係。而具有豐富的風控標籤數據的第三方機構提供數據樣本的審批週期較長、模型效果難優化等問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"星雲Clustar創始人陳凱將目前銀行在數據上面臨的主要問題歸納爲以下三點："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"數據來源問題。之前通過第三方SDK獲取數據的方式被禁止，數據收集方式會受影響。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"同質化嚴重。爲了合規，廠商不能直接出售明文數據，基本以提供解決方案爲主，這使銀行無法得知數據來源，失去了自由度和可解釋性。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"有嚴重的滯後性。用戶消費習慣、財務狀況等是不斷變化的，這可能導致僅僅幾個月前做的評估就失效了，但風控等級可能無法及時跟進這些變化。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隱私計算可以幫助銀行彌補數據方面的弱勢。銀行通過隱私計算平臺獲取大量數據的計算結果，在不直接接觸數據的情況下，發揮出數據價值，使其業務形態得到巨大自由度。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"據陳凱舉例介紹道，隱私計算技術提供商星雲Clustar依託海量數據源爲某股份制銀行實現了精準、可控的閃電貸營銷業務，在數據“可用不可見”的前提下，保證了銀行客戶的隱私和數據安全，同時優化了銀行業務流程。據悉，該業務的當月營銷轉化率達3.5倍以上，聯邦模型AUC達到0.73，營銷精準率和客戶轉化率均有明顯提升。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/c5\/e4\/c542d29355aa0132920c28eeb990abe4.png","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","text":"隱私計算在廣告投放示意圖"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"除了精準營銷，隱私計算在金融領域的場景也在不斷擴展，如信貸風控、保險定價等等。而除了金融，醫療、教育、保險等行業是缺數據的，這些都是隱私計算的機會。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"當然，隱私計算的價值不只體現在傳統行業，互聯網企業內部也可以實現價值共贏。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“比如螞蟻和騰訊數據不通，他們不會知道彼此數據的具體價值，這表明大廠們也會有數據認知上的缺失。另外，隨着市場差異化越來越大、產生的數據越來越多，大廠們在新環境下未必可以直接看見市場、理解新的業務場景。”陳沫說道。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"越來越多的行業開始使用隱私計算技術進行數據保護。面對不同行業各具特色的需求，陳凱表示，不同行業的數據底座90%以上是可以複用的。也就是說，底層的基礎平臺是跨行業的，但越深入行業，隱私計算平臺上的應用會越需要行業的支持。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"監管趨嚴催生更多機會"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"除了技術，監管也是當前隱私保護的另一個重要抓手。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2015年，《國家安全法》正式將數據安全納入國家安全範疇，此後相關法規一直陸陸續續出臺。今年6月，《中華人民共和國數據安全法》正式通過，並將於今年9月1日起施行。據零壹智庫統計，近5年來，國家、地方省市以及各行業監管部門關於數據安全、網絡安全已至少頒佈52部相關法律法規。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不僅如此，針對擁有海量數據的互聯網行業數據安全整治也一直在進行。19年底，光大銀行、微店、房天下等在內的100款違規採集個人信息的App被要求下架整改，此後更是加大了整治力度。僅在今年7月份，國家網信辦就下架了“滴滴企業版”等25款存在嚴重違法違規收集使用個人信息問題的App、工信部將之前通報後仍未完成整改的48款App全部做了下架處理等。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在陳凱看來，行業法律法規的出臺對整個隱私計算市場的冷啓動有極大的幫助。所有與數據相關、但之前沒有考慮到隱私的業務，都有可能成爲隱私計算的市場。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“爲了合規，各企業必須要採取行動，對於之前存在很多有爭議的事情，企業間逐漸達成了共識，新的數據合作也隨之產生。更重要的是，在行業還沒有全面落地的情況下，各類法規的頒佈爲數據流通的底座提供了規範，這對整個行業未來長期的有序發展是非常有利的。”陳凱表示。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"據陳沫介紹，隱私計算市場的機會體現在兩方面：存量市場和增量市場。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"存量市場的機會主要體現在對目前已經初步落地的金融、醫療、政務等領域進行深耕和拓展，在保證數據安全、數據隱私的情況下，讓存量大數據變成可用大數據，充分挖掘出大數據的價值。而增量市場更大、更有開拓性，當然也更充滿不可預測性，市場的期待值也更高。如今很少有產品跟數據毫無關係，無論做合規還是做增益，都蘊藏着巨大的機會。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“用隱私計算賦能行業、賦能社會，雖然聽起來很大，但這卻是真實存在的需求。”陳凱說道。據悉，今年星雲的收入同比增長了5倍，其一些合作伙伴甚至實現了10倍速增長。陳凱透露，今年數據安全相關的企業數量可能已經增加了幾十倍，除了垂直的創業企業，其他跟數據相關的企業，如大數據、安全領域的企業開始進入隱私計算行業。同時，大廠們也已經進場。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"但是，機會背後也意味着挑戰。如何通過技術手段達到法規規定的標準？怎樣做纔是合規的？這些都是隱私計算行業正在面對的問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“第三方數據安全技術提供商連接了數據提供方和需求方，輸出符合法規的數據，可以讓數據產生更大的價值，爲企業賦能。但是參與數據處理、存儲等，還是單純做一個技術輸出平臺，這是星雲目前在思考的事情。”陳凱說道。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"算力制約發展"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隱私計算融合了人工智能、密碼學、區塊鏈及計算芯片等一系列軟硬件技術。在實踐中，隱私保護要用複雜的密碼學進行計算，雖然一定程度上保證了數據安全，但這背後需要付出一項很高的代價：算力。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隱私保護的底層算法可以分爲兩類：一類是基於密碼學理論的算法，另一類是基於計算機硬件。基於密碼學理論的有同態加密、祕密共享等方式，而基於硬件創建的可信環境可以讓數據在獨立的環境中運行，安全性更高。不過值得注意的是，可信環境的建立往往需要對硬件廠商有充分的信任，比如使用英特爾的SGX芯片，則意味着使用者對英特爾的充分信任。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"然而無論使用上述哪種方式，在對海量數據進行加密或做隨機化處理時，都需要進行大量的多方案卷積層計算，模型越複雜，需要的算力越高。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“當前的數據處理速度其實非常慢。”陳凱表示，“相比明文計算，隱私計算在現有CPU計算上的時間還要增加三個數量級。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"限制行業發展的不止算力，大量的數據傳輸對網絡通訊也是很大的挑戰，尤其是像祕密共享這樣對通信要求很高的技術。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"祕密共享是一種分佈式的密碼技術，有大量的重複計算和數據交互，訓練模型需要不斷更新迭代。大量的通信需求、大量容器的交互等都導致應用祕密共享的成本很高，只能支撐像調研這樣的小範圍應用，難以在整個互聯網上實現。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"算力和通信問題制約了隱私計算行業的發展。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"爲解決算力問題，星雲當前給出的解決方案是融合CPU\/GPU\/FPGA服務器、FATE和SDN，推出了多解決方案的智能一體機，通過硬件設計提升單點算力，加速數據加密、解密以及密態計算速度。官方數據顯示，其密態計算效率提升400%、降低延遲300%、降低功耗了70%。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“我們應該知道每個技術擅長什麼、不擅長什麼，基於這個認知再做技術選型。”陳凱說道，“隨着業界對加密技術、芯片等研究的深入，還有量子計算機發展帶來的算力提升，隱私計算的很多問題都會被解決。當然，新的問題也會隨之產生，但那是在很久之後了。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]}]}