《電子技術(shù)應(yīng)用》
您所在的位置:首頁(yè) > 通信與網(wǎng)絡(luò) > 設(shè)計(jì)應(yīng)用 > 聯(lián)邦學(xué)習(xí)在金融數(shù)據(jù)安全領(lǐng)域的研究與應(yīng)用
聯(lián)邦學(xué)習(xí)在金融數(shù)據(jù)安全領(lǐng)域的研究與應(yīng)用
信息技術(shù)與網(wǎng)絡(luò)安全 1期
張海濤  
(五礦國(guó)際信托有限公司,,北京100027)
摘要: 近年來(lái),,金融領(lǐng)域明文數(shù)據(jù)流通所引起的數(shù)據(jù)泄露問(wèn)題日漸突出,傳統(tǒng)的跨機(jī)構(gòu)數(shù)據(jù)融合的機(jī)器學(xué)習(xí)方式面臨著新的問(wèn)題與挑戰(zhàn),。因此,,立足于金融數(shù)據(jù)安全領(lǐng)域,,從用戶隱私和數(shù)據(jù)安全角度出發(fā),,概述聯(lián)邦學(xué)習(xí)理論并深入分析其目前在金融行業(yè)的應(yīng)用現(xiàn)狀,,指出現(xiàn)有的聯(lián)邦學(xué)習(xí)還存在通信效率低,、數(shù)據(jù)異構(gòu)性突出等問(wèn)題,。最后提出健全聯(lián)邦學(xué)習(xí)標(biāo)準(zhǔn)體系,、時(shí)刻關(guān)注監(jiān)管要求等建議,為推動(dòng)聯(lián)邦學(xué)習(xí)在金融數(shù)據(jù)安全領(lǐng)域中的合法應(yīng)用提供參考性意見。
中圖分類號(hào): TP391
文獻(xiàn)標(biāo)識(shí)碼: A
DOI: 10.19358/j.issn.2096-5133.2022.01.001
引用格式: 張海濤. 聯(lián)邦學(xué)習(xí)在金融數(shù)據(jù)安全領(lǐng)域的研究與應(yīng)用[J].信息技術(shù)與網(wǎng)絡(luò)安全,,2022,,41(1):3-9.
Research and application of federated learning in the field of financial data security
Zhang Haitao
(Minmetals International Trust Co.,Ltd.,,Beijing 100027,China)
Abstract: Abstract: In recent years, the problem of data leakage caused by the circulation of plaintext data in the financial field has become increasingly prominent. The traditional machine learning method of inter agency data fusion faces new problems and challenges. Therefore, based on the field of financial data security, from the perspective of user privacy and data security, this paper summarizes the federated learning theory, deeply analyzes its current application status in the financial industry, and points out that the existing federated learning still has some problems, such as low communication efficiency and prominent data heterogeneity. Finally,it puts forward suggestions on improving the federated learning standard system and paying attention to regulatory requirements at all times, so as to provide reference opinions for promoting the legal application of federated learning in the field of financial data security.
Key words : federated learning,;financial data security,;data privacy,;credit card fraud

0 引言

2020年4月,,中共中央、國(guó)務(wù)院印發(fā)了《關(guān)于構(gòu)建更加完善的要素市場(chǎng)化配置體制機(jī)制的意見》,,明確指出在當(dāng)今數(shù)字經(jīng)濟(jì)化時(shí)代,,數(shù)據(jù)是至關(guān)重要的一種新型生產(chǎn)要素。但是,,隨著數(shù)據(jù)賦能研究的不斷深入,,隱私保護(hù)和數(shù)據(jù)泄露等問(wèn)題日益突出,。如2018年3月,,超5 000萬(wàn)Facebook用戶信息被政治數(shù)據(jù)公司“劍橋分析”獲取并利用,,2018年11月,匯豐銀行(HSBC Bank)部分客戶財(cái)務(wù)狀況和個(gè)人信息被泄露,。金融作為數(shù)據(jù)密集型行業(yè),,對(duì)數(shù)據(jù)安全、隱私保護(hù)以及監(jiān)管科技等有著更高的要求,。實(shí)現(xiàn)數(shù)據(jù)的多方協(xié)同和授權(quán)共享,得到更優(yōu)的模型和決策,,是當(dāng)前人工智能賦能金融科技的一個(gè)重大挑戰(zhàn)[1]。Google于2016年提出聯(lián)邦學(xué)習(xí)(Federated Learning)概念為這一困境帶來(lái)了新的思路與解決辦法,。目前,,聯(lián)邦學(xué)習(xí)技術(shù)已經(jīng)在金融科技領(lǐng)域的智能營(yíng)銷,、反欺詐,、信用卡評(píng)分、產(chǎn)品推薦等多個(gè)業(yè)務(wù)場(chǎng)景中得到了具體應(yīng)用。





本文詳細(xì)內(nèi)容請(qǐng)下載:http://forexkbc.com/resource/share/2000003929





作者信息:

張海濤

(五礦國(guó)際信托有限公司,,北京100027)




此內(nèi)容為AET網(wǎng)站原創(chuàng),,未經(jīng)授權(quán)禁止轉(zhuǎn)載,。