Birrul Walidain

Mr. Birrul Walidain

Department of Physics, Universitas Syiah Kuala, Banda Aceh, Aceh, Indonesia
orcid

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2026 in IOP Conference Series: Earth and Environmental Science
Verified
Conference Paper
Open access
Birrul WalidainBirrul Walidain
Birrul Walidain
NINasrullah Idris
Nasrullah Idris
KSKhairun Saddami
Khairun Saddami
NYNatasya Yuzza
Natasya Yuzza
RMRara Mitaphonna
Rara Mitaphonna
Accurate elemental identification in Laser-Induced Breakdown Spectroscopy (LIBS) is hampered by dense line overlap, matrix effects, and the quadratic computational cost of standard self-attention on high-resolution spectra. We investigate an Informer-based deep learning model for qualitative multi-element LIBS analysis, trained entirely on a physics-based synthetic library generated from Saha-Boltzmann Local Thermal Equilibrium (LTE) plasma modelling. The Informer encoder employs ProbSparse self-attention, reducing the complexity to O ( L log L ) and is optimized with a Multi-Label Focal Loss to handle class imbalance across 17 elemental labels. On 7,500 simulated test spectra, a compact 2-layer Informer attains a macro F 1 score of 0.64 (precision 0.59, recall 0.77), outperforming a deeper 3-layer variant that exhibited overfitting. As a proof of concept on experimental data, the trained model was applied to LIBS spectra of an Aceh Traditional Women’s Medicine and successfully identified Na, Mg, Ca, and Mn in agreement with manual line identification using NIST Atomic Spectra Database. These results demonstrate that an Informer-based, simulation-trained LIBS model can transfer from synthetic to real spectra without empirical calibration standards and has practical potential as a rapid support tool for multi-element assessment of complex herbal preparations.