





摘" 要: 金融數(shù)據(jù)具備非線性、高維度的特點(diǎn),同時(shí)對安全性有較高的要求。文中結(jié)合區(qū)塊鏈技術(shù)和模糊聚類算法,提出一種面向區(qū)域互聯(lián)網(wǎng)金融的異常數(shù)據(jù)分析模型,該模型由異常數(shù)據(jù)分析算法和隱私保護(hù)算法組成。異常數(shù)據(jù)分析算法針對模糊均值聚類算法處理高維非線性數(shù)據(jù)能力弱的缺點(diǎn),使用深度信念網(wǎng)絡(luò)進(jìn)行改進(jìn),進(jìn)而提升模型的數(shù)據(jù)處理能力。隱私保護(hù)使用差分隱私保護(hù)算法,在不利用背景知識的前提下完成數(shù)據(jù)的保護(hù),同時(shí)保證了數(shù)據(jù)的可用性。在實(shí)驗(yàn)測試中,將所提模糊聚類算法與常用的主流K?Means算法、DPC算法進(jìn)行了對比,結(jié)果表明:所提算法的性能在所有對比算法中最優(yōu);與此同時(shí),加入隱私保護(hù)算法后對聚類結(jié)果的影響保持在0.021以內(nèi),充分證明了該算法性能的優(yōu)越性。
關(guān)鍵詞: 模糊聚類算法; 區(qū)塊鏈技術(shù); 異常數(shù)據(jù)識別; 深度信念網(wǎng)絡(luò); 差分隱私保護(hù)算法; 區(qū)域數(shù)據(jù)分析
中圖分類號: TN919.5?34; TP391" " " " " " " " "文獻(xiàn)標(biāo)識碼: A" " " " " " " " " " " 文章編號: 1004?373X(2025)06?0052?05
Research on regional big data analysis technology based on blockchain
and fuzzy clustering algorithm
HE Ying
(Chengyi College, Jimei University, Xiamen 361000, China)
Abstract: Financial data has the characteristics of nonlinearity and high dimensionality, while also requiring high security. By combining with blockchain technology and fuzzy clustering algorithm, an anomaly data analysis model for regional internet finance is proposed. This model is composed of anomaly data analysis algorithm and privacy protection algorithm. In allusion to the weakness of the fuzzy mean clustering algorithm in processing high?dimensional nonlinear data, the anomaly data analysis algorithm can use the deep belief networks to improve the data processing capabilities of the model. In the privacy protection, the differential privacy protection algorithm can be used to protect data without utilizing background knowledge, while ensuring data availability. In the experimental testing, the proposed fuzzy clustering algorithm was compared with commonly used mainstream K?Means and DPC algorithms. The results show that the performance of the proposed algorithm is the best among all comparison algorithms. At the same time, the impact of incorporating privacy protection algorithm on clustering results can remain within 0.021, fully demonstrating the superiority of the performance of this algorithm.
Keywords: fuzzy clustering algorithm; blockchain technology; nonlinear data identification; deep belief network; differential privacy protection algorithm; regional data analysis
0" 引" 言
隨著金融數(shù)據(jù)形態(tài)的不斷演進(jìn),傳統(tǒng)的金融運(yùn)營模式也發(fā)生了變化,以互聯(lián)網(wǎng)金融服務(wù)新模式為主體的用戶量急劇增加。互聯(lián)網(wǎng)金融與傳統(tǒng)金融服務(wù)存在著巨大差異,在對區(qū)域互聯(lián)網(wǎng)金融數(shù)據(jù)的監(jiān)管方面,我國仍處于初級發(fā)展階段,存在一定的信用風(fēng)險(xiǎn)問題。因此,對金融數(shù)據(jù)進(jìn)行風(fēng)險(xiǎn)識別并保護(hù)數(shù)據(jù)的隱私安全成為了重要的研究方向。
區(qū)域互聯(lián)網(wǎng)金融數(shù)據(jù)的特點(diǎn)是高維度與非線性[1?2],其對數(shù)據(jù)進(jìn)行風(fēng)險(xiǎn)識別是通過聚類模型找到數(shù)據(jù)群中的異常點(diǎn)并進(jìn)行相對應(yīng)的處理,同時(shí)在數(shù)據(jù)處理過程中還需要保證數(shù)據(jù)的隱私性,避免數(shù)據(jù)泄露現(xiàn)象的發(fā)生。……