Multi-level information extraction based on convolutional neural network and its optimal training method
Bibliographic record
Abstract
This paper focuses on the characteristics of multilevel information extraction, based on the convolutional neural network model (CNN), introduces the multi-scale feature fusion and multilevel feature fusion strategy to study the multilevel information extraction method, and proposes the full convolutional neural network based on the attention mechanism and residual connection to form the multilevel information extraction model.Aiming at the gradient disappearance and saddle point problem of convolutional neural network, an activation gradient (AG) algorithm is proposed to optimize its training, which is improved to a class of activation gradient convolutional neural network (AG-CNN).The practical application effect of the multilevel information extraction model in this paper is verified by the information extraction work of net-pen culture in river-type reservoirs.Compared with the classical models such as UNet and ResUNet, the intersection and integration ratio (IoU), recall rate, precision rate, and F1 score of this paper's model reach the highest 80.28%, 91.02%, 87.18%, and 89.03% among all the models, which possesses a stronger extraction capability.And in the multilevel information extraction experiments on Cifar100 and Caltech256 datasets, when the number of batch training data is greater than 100, the accuracy rate and performance of the experimental group basically remain stable.
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 | 0.001 |
| 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 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| 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".