吉大正元隐私计算安全评测

隐私计算评测从数据增强、数据评测、算法评测等多角度出发，面向金融行业数据安全与隐私保护应用场景，针对隐私计算平台面临的数据与算法的数据偏见、算法偏见、内生脆弱性等安全威胁，提出面向金融行业的隐私计算评测解决方案，构建联邦学习可信安全评测框架体系，在充分发挥数据价值，又不影响用户数据隐私与安全的基础上，实现数据与算法的可信安全评测。

Privacy computing evaluation is conducted from multiple perspectives including data augmentation, data evaluation, and algorithm evaluation. Targeting the application scenarios of data security and privacy protection in the financial industry, and addressing security threats such as data biases, algorithm biases, and endogenous vulnerabilities of data and algorithms encountered by privacy computing platforms, this work proposes a financial industry-oriented privacy computing evaluation solution and constructs a trusted security evaluation framework for federated learning. It enables trusted security evaluation of data and algorithms while fully unlocking the value of data without compromising user data privacy and security.

搜集汇总

背景与挑战

背景概述

该数据集提供面向金融行业的隐私计算安全评测解决方案，通过数据增强、算法评测等方法构建联邦学习可信安全框架，检测数据偏见和算法脆弱性，保障数据隐私安全。适用于隐私计算评测和大模型评测场景。