Pan-Canadian Predictive Modeling of Lithium–Cesium–Tantalum Pegmatites with Deep Learning and Natural Language Processing
Bibliographic record
Abstract
Abstract The discovery of new lithium resources is essential because lithium plays a vital role in the manufacturing of green technology. Along with brines and volcano–sedimentary deposits, approximately a one-third share of global lithium resources is associated with lithium-cesium-tantalum (LCT) pegmatites, with Canada hosting numerous examples. This research applied generative adversarial networks, natural language processing, and convolutional neural networks to generate mineral prospectivity models and support exploration targeting for Canadian LCT pegmatites. Geoscientific text data included within public bedrock geology maps and natural language processing were used to convert conceptual targeting criteria into evidence layers that complement more traditional, geophysical and geochronological data used for mineral prospectivity modeling (MPM). A multilayer architecture of convolutional neural networks, including an attention mechanism, was designed for data modeling. This architecture was trained and validated using variable synthetically generated class labels, input image sizes, and hyperparameters, resulting in an ensemble of 1000 models. The uncertainty of the ensemble was analyzed using a risk–return analysis, yielding a bivariate choropleth risk–return plot that facilitates the interpretation of prospectivity models for downstream applications. This was further complemented by employing post hoc interpretability algorithms to translate the black-box nature of neural networks into comprehensible content. The low-risk and high return class of our prospectivity models reduces the search space for discovering LCT pegmatites by 88%, delineating 99% of known LCT pegmatites in Canada. The results of this study suggest that our workflow (i.e., combining synthetic data generation, natural language processing, convolutional neural networks, and uncertainty propagation for MPM) facilitates decision-making for regional-scale lithium exploration and could also be applied to other mineral systems.
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) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
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".