Incentive Mechanism Design in Federated Reinforcement Learning With Uniform and Non-Uniform Data Bounds
Bibliographic record
Abstract
In this paper, we examine the contract design for integrating federated learning (FL) and reinforcement learning (RL). A key feature of federated reinforcement learning (FRL) is its dynamic data size, which poses challenges in incentivizing consistent user participation. Unlike traditional contracts where the contract is the same for all global training round, our design specifies the required amount of data along with the corresponding reward for each global training round. We introduce a data bound reflecting the data accumulation rate and study two scenarios: uniform data bound (same for all users) and non-uniform data bound (vary by user). For uniform data bounds, we show that monotonicity holds, simplifying individual rationality (IR) and incentive compatibility (IC) constraints. However, for non-uniform data bounds, monotonicity is violated. To address this, we propose an algorithm for non-uniform data bounds, reducing the complexity of contract design. To design the contract, we maximize the server's utility, defined as the accuracy minus the total payment, subject to individual rationality (IR) and incentive compatibility (IC), while considering data bounds. Through extensive simulations, we have validated IR and IC constraints for all types of users in every global training round. We observe that 1) the server increases data requirements as data accumulates for each type, 2) payments grow with user effort, 3) our dynamic contract model significantly outperforms existing static contract, and 4) the proposed algorithm effectively handles the complexities of non-uniform data bounds, making it a robust solution for dynamic contract design.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.009 | 0.019 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".