awesome-vertical-federated-learning

A curated list of advancements in Vertical Federated Learning (VFL), frameworks and libraries.

Table of Contents

• Publications in Top-tier Conferences
  • Surveys on VFL
  • VFL benchmarks (benchmarks with VFL tasks)
  • VFL algorithms
  • VFL privacy
  • VFL metrics / feature importance estimation
• VFL Datasets (or datasets that are used in benchmarks)
• Frameworks and Libraries with VFL support
  • FATE
  • FedML
  • Falcon

Publications in Top-tier Conferences (or influential)

Surveys on VFL

Type	Title	Year	Conference / Journal	Description
VFL	Vertical Federated Learning: Concepts, Advances and Challenges	2023	Arxiv
General	Towards Open Federated Learning Platforms: Survey and Vision from Technical and Legal Perspectives	2024	Arxiv

VFL benchmarks (benchmarks with VFL tasks)

Bench Type	Title	Year	Conference	Code	Algorithms
VFL	Stalactite: Toolbox for Fast Prototyping of Vertical Federated Learning Systems	2024	RecSys	Code	---
VFL	VertiBench: Advancing Feature Distribution Diversity in Vertical Federated Learning Benchmarks	2024	ICLR	Code Website	GAL, C-VFL, SecureBoost, Pivot, FedTree, FedOnce
VFL	VFLAIR: A Research Library and Benchmark for Vertical Federated Learning	2024	ICLR	Code	---
VFL	FedAds: A Benchmark for Privacy-Preserving CVR Estimation with Vertical Federated Learning	2023	SIGIR	Code	---
General	The OARF Benchmark Suite: Characterization and Implications for Federated Learning Systems	2022	ACM Transactions on Intelligent Systems and Technology	Code	---
General	Fedml: A research library and benchmark for federated machine learning	2020	arxiv	Code	---

VFL algorithms

Algorithm	Model	Category	Title	Code	Year	Conference / Journal
AL	Any	Ensemble-based	Assisted learning: A framework for multiorganization learning	-	2020	Neurips
GAL	Any	Ensemble-based	Gal: Gradient assisted learning for decentralized multi-organization collaborations	Code	2022	Neurips
SplitNN	NN	Split-based	Split learning for health: Distributed deep learning without sharing raw patient data	-	2018	Arxiv
C-VFL	NN	Split-based	Compressed-VFL: Communication-efficient learning with vertically partitioned data	Code	2022	ICML
BlindFL	NN	Split-based	Vertical federated machine learning without peeking into your data	-	2022	SIGMOD
FedOnce	NN	Split-based	Practical vertical federated learning with unsupervised representation learning	Code	2022	IEEE Transactions on Big Data
SecureBoost	GBDT	Split-based	Secureboost: A lossless federated learning framework	-	2021	IEEE Intelligent Systems
Pivot	GBDT	Split-based	Privacy preserving vertical federated learning for tree-based models	Code	2020	VLDB
FedTree	GBDT	Split-based	Fedtree: A federated learning system for trees	Code	2023	MLSyS
VF2Boost	GBDT	Split-based	Vf2boost: Very fast vertical federated gradient boosting for cross-enterprise learning	-	2021	SIGMOD
OpBoost	GBDT	Split-based	OpBoost: A Vertical Federated Tree Boosting Framework Based on Order-Preserving Desensitization		2023	VLDB
Fed-Forest	RF	Split-based	Federated forest	-	2020	IEEE Transactions on Big Data

VFL privacy

Title	Year	Conference / Journal	Description
Privacy Matters: Vertical Federated Linear Contextual Bandits for Privacy Protected Recommendation	2023	KDD	---
A Unified Solution for Privacy and Communication Efficiency in Vertical Federated Learning	2023	Neurips	---
Differentially Private Vertical Federated Clustering	2023	VLDB	---

VFL metrics / feature importance estimation

Title	Year	Conference / Journal	Description
Fair and Efficient Contribution Valuation for Vertical Federated Learning	2024	ICLR	Clients' contribution valuation metric - vertical federated Shapley value (VerFedSV)

VFL Datasets (or datasets that are used in benchmarks)

Type	Dataset	Modality	Link	Benchmark	# parties	# samples	# features	# classes
VFL-native	NUS-WIDE	Image	Link	VertiBench, VFLAIR	5	269,648	64 / 144 / 73 / 128 / 225	2
VFL-native	Satellite	Image	Link	VertiBench	16	3,927	13-channel 158x158	4
VFL-native	Vehicle	Acoustic, Seismic	Link	VertiBench	2	78,823	50 / 50	3
VFL-native	FedAds	Table	Link	FedAds	2	11,300,000	16 / 7	-
Centralized	covtype	Table	Link	VertiBench	-	581,012	54	7
Centralized	msd	Table	Link	VertiBench	-	463,715	90	-
Centralized	realsim	Table	Link	VertiBench	-	72,309	20,958	2
Centralized	gisette	Table	Link	VertiBench	-	60,000	5,000	2
Centralized	epsilon	Table	Link	VertiBench, FedAds	-	400,000	2,000	2
Centralized	letter	Table	Link	VertiBench	-	15,000	16	26
Centralized	radar	Table	Link	VertiBench	-	15,000	16	26
Centralized	MNIST	Image	Link	VertiBench, VFLAIR	-	325,834	174	7
Centralized	CIFAR10	Image	Link	VertiBench, VFLAIR	-	60,000	1,024	10
Centralized	CIFAR100	Image	Link	VFLAIR	-	60,000	1,024	100
Centralized	Breast Cancer	Table	Link	VFLAIR	-	569	32	2
Centralized	Pima Indians Diabetes	Table	Link	VFLAIR	-	768	9	2
Centralized	Breast histopathology images	Image	Link	FedAds
Centralized	Yahoo answers dataset	Text	Link	FedAds
Centralized	Give Me Some Credit	Tabilar	link	FedAds
Centralized	Avazu	Tabilar	link	FedAds	-	45,006,432	23	2

Frameworks and Libraries with VFL support

FATE

github paper

VFL-related (hetero in FATE terminology) features:

• privacy-preserving strategies: SSHE and FedPass

Stalactite

paper

FedML

Implements a bunch of practical algorithms in horizontal and vertical FL settings

Falcon

https://github.com/nusdbsystem/falcon