User Guide

Preprocessing

SecretFlow provides several common utility functions/classes to change raw features into a representation for more suitable for the downstream pipeline.

DataFrame

Secretflow provides federated data encapsulation in the form of DataFrame. DataFrame is composed of data blocks of multiple parties and supports horizontal or vertical partitioned data.

WeightOfEnvidenceEncoding (WOE)

Secretflow provides WOE Encoding for vertical datasets.

Private Set Intersection(PSI)

SecretFlow SPU now supports ECDH-PSI, KKRT16-PSI, and BC22-PCG-PSI. Please check Private Set Intersection(PSI) for details.

Private Set Intersection(PSI)

Private set intersection is a secure multiparty computation cryptographic technique that allows two parties holding sets to compare encrypted versions of these sets in order to compute the intersection.

MPC Machine Learning

SecretFlow provides a variety of MPC modeling capabilities through the MPC security protocol and HE homomorphic encryption.

Linear Models

a set of methods intended for regression in which the mean of target value is expected to be a linear combination of the features (or a map of linear combinations).

Decision trees

Decision Trees (DTs) are a non-parametric supervised learning method used for classification and regression.

Feature Engineering

Feature Engineering includes Pearson product-moment correlation coefficient, Variance Inflation Factor (VIF) and Hypothesis Testing for linear Regression Coefficients.

Federated Learning

Federated learning is a machine learning technique that trains an algorithm across multiple decentralized edge devices or servers holding local data samples, without exchanging them.

Horizontal Federated Learning

For cases that multi participants share the same feature space but differ in sample ID.

Vertical Federated Learning

For cases that multi participants share the same sample ID space but differ in feature space.

Mix Federated Learning

For cases that parts of participants share the same sample ID space but differ in feature space, where others share the same feature space but differ in sample ID.