AuditFL makes federated learning updates auditable and economically accountable
AuditFL: A Contract-Backed Clearing Layer for Open Federated Learning
Distributed, Parallel, and Cluster ComputingMachine Learning
Summary
Federated learning lets many clients train a shared model without sharing their data, but typically relies on a trusted coordinator. The authors present AuditFL, a system that records key decisions and transactions on a blockchain to make the training process transparent and auditable. AuditFL keeps heavy tasks off the blockchain but tracks economic commitments and update identifiers to ensure fairness and accountability. Their tests with many clients show the approach maintains model accuracy even under adverse conditions.
What this means in practice
- •For federated learning platform developers: Create transparent and tamper-evident client update tracking and reward systems using blockchain contracts in federated learning deployments.
- •For distributed machine learning engineers: Implement auditable coordination layers that economically constrain aggregators and support audit trails in open federated learning scenarios.
Authors
Alexandros Balla, Spyros Lalis
Abstract
Federated learning (FL) keeps raw data local, but most deployments still rely on a trusted coordinator for client admission, checkpoint distribution, update collection, aggregation, and rewards. This paper presents AuditFL, a contract-backed clearing layer for auditable FL in open settings. AuditFL keeps training and aggregation off chain, stores model artifacts in an assumed reliable off-chain artifact store, and records only content identifiers and economic state on chain. Clients reserve update tickets with deposits, publish update identifiers, and later claim refunds or rewards; the task coordinator publishes global checkpoints and finalizes rounds through deterministic settlement rules. The design does not remove the aggregator, but makes its on-chain round decisions attributable, replayable, auditable, and economically constrained. We implement AuditFL as both a deterministic state-machine prototype and a Solidity contract on a local Ethereum-compatible Anvil chain. On full MNIST with Dirichlet non-IID splits, 12 rounds, and up to 200 clients, AuditFL preserves the Plain FL learning trajectory while its overhead follows closed-form control-plane models. At 200 clients, the contract run issues 5,026 transactions, stores 2,413 off-chain artifacts (144.9 MiB), and consumes 829.7M EVM gas, dominated by per-client reserve and publish calls. Controlled dropout, invalid-update, and artifact-unavailability stress tests reduce inclusion nearly linearly but still finalize rounds, with final-accuracy changes remaining within a few tenths of an accuracy percentage point even under high injected failure rates.