Fragments Adjacency Prediction: A Contour Based Approach Using Transformer Models with Rotary Positional Encoding
Bibliographic record
Abstract
This paper introduces an automated deep learning method for predicting adjacency relationships between 2D fragments, with potential applications in archaeology and other domains requiring fragment pair identification. Such tasks often require the identification of fragment matching within highly irregular, eroded datasets, where conventional techniques struggle with scalability and generalization. To address these computational challenges, we propose a transformer-based neural network architecture that leverages contour-based features and integrates Rotary Positional Encoding (RoPE) to enhance spatial relationship modeling. Unlike traditional approaches that rely on exhaustive pairwise comparisons or similarity matrices, our method directly predicts adjacency relationships, significantly improving computational efficiency. Given the scarcity of real-world datasets, we developed a synthetic data generation framework capable of simulating diverse fragment patterns to enhance model robustness and mitigate overfitting. Experimental evaluations, by achieving an accuracy of 82%, demonstrate that accurate adjacency predictions can be achieved using minimal input features, highlighting the effectiveness of contour-based representations. While our model has been validated on synthetic datasets, its performance can be further refined through fine-tuning with real-world archaeological and forensic fragment data. This work lays the foundation for future advancements incorporating additional modalities such as texture and color, ultimately contributing to more reliable and scalable fragment matching systems for practical applications.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".