Multi-Task Deep Learning for Surface Metrology

TL;DR

This paper introduces a deep learning framework for surface metrology that predicts surface texture parameters and their uncertainties, demonstrating high accuracy and calibration across tactile and optical measurement systems.

Contribution

The work presents a reproducible multi-task deep learning approach for surface parameter prediction and uncertainty quantification, integrating calibration methods for reliable measurement intervals.

Findings

High regression accuracy for surface parameters (R2 > 0.98)

Effective uncertainty modeling with calibrated intervals

Naive multi-output models underperform compared to single-target models

Abstract

A reproducible deep learning framework is presented for surface metrology to predict surface texture parameters together with their reported standard uncertainties. Using a multi-instrument dataset spanning tactile and optical systems, measurement system type classification is addressed alongside coordinated regression of Ra, Rz, RONt and their uncertainty targets (Ra_uncert, Rz_uncert, RONt_uncert). Uncertainty is modelled via quantile and heteroscedastic heads with post-hoc conformal calibration to yield calibrated intervals. On a held-out set, high fidelity was achieved by single-target regressors (R2: Ra 0.9824, Rz 0.9847, RONt 0.9918), with two uncertainty targets also well modelled (Ra_uncert 0.9899, Rz_uncert 0.9955); RONt_uncert remained difficult (R2 0.4934). The classifier reached 92.85% accuracy and probability calibration was essentially unchanged after temperature scaling (ECE 0.00504 -> 0.00503 on the test split). Negative transfer was observed for naive multi-output trunks, with single-target models performing better. These results provide calibrated predictions suitable to inform instrument selection and acceptance decisions in metrological workflows.