Adaptive Learning for Soil Classification in Laser-Induced Breakdown Spectroscopy Streaming
Bibliographic record
Abstract
The application of machine learning (ML) has accelerated the development of laser-induced breakdown spectroscopy (LIBS) in soil analysis. However, analyzing remote LIBS data in real time using ML is challenging due to several factors. Firstly, building robust ML models requires extensive calibration datasets, which are not always possible with limited LIBS experimental data. Secondly, matrix effects can worsen LIBS performance, and changes in sample physical properties or the apparatus can impact the distribution and intensity of emission lines. These issues may lead to concept drift in real-time/online data streaming, causing the relationship between the input and the target spectra to change over time. Consequently, an ML model designed for one LIBS system may not apply to another. To conquer these challenges, we propose a framework based on transfer learning to use limited experimental data and adapt to the emission line variation in the LIBS streaming. A model is first pre-trained using a large labelled source dataset and then fine-tuned with new experimental measurements to classify soil samples. LIBS measurements are conducted with variations in sample properties and experimental parameters to simulate differences in remote LIBS sensors. The collected spectra are fed into the model by chunks, and data evolution is dynamically learned by self-balanced learning to self-adapt to the domain shift. The proposed framework is found effective in improving classification accuracy during data streaming by implementing transfer learning and supporting adaptation compared to the literature. The code of the proposed method is available in the GitHub athttps://github.com/kelci2017/LIBS_streaming.
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 | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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".