[Paper Review] A Comprehensive Survey of Incentive Mechanism for Federated Learning

Rongfei Zeng, Chao Zeng+3 more|arXiv (Cornell University)|Jun 27, 2021

Privacy-Preserving Technologies in Data52 references59 citations

TL;DR

A systematic survey of incentive mechanisms for federated learning, detailing problem formulation, taxonomy, main techniques (Shapley value, Stackelberg games, auctions, contracts, RL, blockchain), and future directions.

ABSTRACT

Federated learning utilizes various resources provided by participants to collaboratively train a global model, which potentially address the data privacy issue of machine learning. In such promising paradigm, the performance will be deteriorated without sufficient training data and other resources in the learning process. Thus, it is quite crucial to inspire more participants to contribute their valuable resources with some payments for federated learning. In this paper, we present a comprehensive survey of incentive schemes for federate learning. Specifically, we identify the incentive problem in federated learning and then provide a taxonomy for various schemes. Subsequently, we summarize the existing incentive mechanisms in terms of the main techniques, such as Stackelberg game, auction, contract theory, Shapley value, reinforcement learning, blockchain. By reviewing and comparing some impressive results, we figure out three directions for the future study.

Motivation & Objective

• Identify the incentive problem in federated learning and its goal to improve FL performance.
• Provide a comprehensive taxonomy of incentive schemes in FL across settings, phases, and techniques.
• Summarize existing mechanisms by main techniques and sub-problems, with emphasis on contribution evaluation, node selection, and payments.
• Highlight assumptions, advantages, and limitations to guide future research in FL incentives.

Proposed method

• Define the incentive mechanism for FL with multi-dimensional contributions and model owner payments.
• Classify incentive schemes by application setting, FL phase, main techniques, sub-problems, and information symmetry.
• Review mechanisms using Shapley value, Stackelberg games, auctions, contract theory, reinforcement learning, blockchain, and other approaches.
• Discuss properties like IC, IR, fairness, PE, CR, BB, and introduce Performance Improvement (PI) as key for FL.
• Provide examples, discuss computational/privacy challenges, and compare representative works.

Experimental results

Research questions

• RQ1 What constitutes an effective incentive mechanism in Federated Learning and how does it impact FL performance?
• RQ2 How are incentive schemes in FL constructed across cross-device and cross-silo settings, and which techniques are most effective in different sub-problems (contribution evaluation, node selection, payment allocation)?
• RQ3 What assumptions (information symmetry) do existing schemes rely on, and what are their implications for practicality and robustness?
• RQ4 What are the key future directions and open challenges in designing FL incentive mechanisms?],
• RQ5 key_findings':['Shapley value is widely used for contribution evaluation but is computationally expensive, with various approximation methods proposed.
• RQ6 Stackelberg games and auctions are commonly used for leader-follower payment and resource allocation problems in FL.
• RQ7 Contract theory, reputation systems, and blockchain are employed to address information asymmetry, trust, and robustness in FL incentives.
• RQ8 Reinforcement learning offers a mechanism for dynamic strategy adaptation under incomplete information.
• RQ9 Prototype schemes show performance improvements in FL training time and model accuracy when well-designed incentives are used.
• RQ10 Future research should focus on training performance, MEC/5G/IoT constraints, and cross-silo FL incentives.

Key findings

• - Shapley value: contribution measurement and profit sharing, with high computational cost and privacy concerns.
• - Stackelberg games: leader-follower dynamics for resource payments and training time, with incomplete information handled via DRL in some works.
• - Auctions: multi-dimensional bids for data, computation, and communication resources, often achieving social welfare gains; NP-hardness addressed via greedy/approximate methods.
• - Contracts: information-asymmetry mitigation through menu-based designs and monotonicity constraints.
• - RL and blockchain: enable adaptive strategies and robust, transparent incentive mechanisms.
• - Cross-silo incentives are less explored compared with cross-device settings.”],
• table_headers:[],