BraggNet: Complex Photonic Integrated Circuit Reconstruction Using Deep Learning
Bibliographic record
Abstract
We propose a deep learning model to reconstruct physical designs of complex coupled photonic systems, such as waveguide Bragg gratings, from their spectral responses for inverse design and fabrication diagnosis. Traditional reconstructing algorithms demand considerable computing resources at every query. Conversely, machine learning algorithms use most of the computing resources during the training process and provide effortless and orders-of-magnitude faster analysis in response to queries. This approach is demonstrated using silicon photonic grating-assisted, contra-directional couplers consisting of thousands of Bragg periods. The contra-directional couplers are modeled as coupled cavities, for which a transfer matrix model is used to generate a synthetic dataset comprising a strategic design parameter space. The free-form, architecture-independent model allows to include any geometries to the design parameter space. Upon proper training, the model achieves 1.4% mean absolute percentage error on device reconstruction and thus proves suitable for inverse design applications. To further show its potential for assessment of fabricated devices, another dataset is generated to emulate the fabrication conditions of a nominal design hindered by fabrication imperfections. The model is shown to reconstruct devices from experimental measurements with greater than 600-fold improvement in speed compared to the classical layer-peeling algorithm. This proves promising for data-driven processes required by Industry 4.0.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| 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 | 0.001 |
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".