Fine-Grained Traceability for Transparent ML Pipelines

TL;DR

FG-Trac is a novel, model-agnostic framework that provides verifiable, fine-grained traceability of individual data samples throughout machine learning pipelines, enhancing transparency and accountability.

Contribution

It introduces a comprehensive mechanism for sample-level traceability, integrating cryptographic commitments and contribution scoring without altering existing models.

Findings

Preserves predictive performance while enabling traceability

Provides verifiable evidence of sample usage and propagation

Works with diverse ML pipeline architectures

Abstract

Modern machine learning systems are increasingly realised as multistage pipelines, yet existing transparency mechanisms typically operate at a model level: they describe what a system is and why it behaves as it does, but not how individual data samples are operationally recorded, tracked, and verified as they traverse the pipeline. This absence of verifiable, sample-level traceability leaves practitioners and users unable to determine whether a specific sample was used, when it was processed, or whether the corresponding records remain intact over time. We introduce FG-Trac, a model-agnostic framework that establishes verifiable, fine-grained sample-level traceability throughout machine learning pipelines. FG-Trac defines an explicit mechanism for capturing and verifying sample lifecycle events across preprocessing and training, computes contribution scores explicitly grounded in training checkpoints, and anchors these traces to tamper-evident cryptographic commitments. The framework integrates without modifying model architectures or training objectives, reconstructing complete and auditable data-usage histories with practical computational overhead. Experiments on a canonical convolutional neural network and a multimodal graph learning pipeline demonstrate that FG-Trac preserves predictive performance while enabling machine learning systems to furnish verifiable evidence of how individual samples were used and propagated during model execution.