Neural-model optimization improves tested silver-carp oil extraction
Source check: Source checked. Not independent scientific confirmation.
Finding: An ANN-guided search selected extraction conditions for silver-carp viscera oil, followed by physical confirmation. The measured yield was 79.44%, compared with 76.68% at a response-surface-selected setting. This is a small process-optimization experiment; separate encapsulation improvements were not shown to be AI-selected.
Technology: Multilayer perceptron with genetic-algorithm search over its learned response surface
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- A learned nonlinear extraction-response model actively supplied the search objective for a newly tested process setting, rather than only fitting previously measured yields.
- Test and result
- The ANN learned from 17 response-surface-design experiments, with 70/15/15 training/test/validation allocation and a three-input, ten-hidden-node architecture selected through trials. Its learned response served as the genetic-search fitness function. The selected setting predicted 82.45% extraction; follow-up measurements yielded 79.44%, versus 76.68% for the response-surface setting, with source-reported significant difference.
- Limitation / still unknown
- Seventeen training-design runs produce very small evaluation subsets, and architecture choice by trial/error limits generalization claims. The physical optimized-setting measurement is distinct from the reported high fit correlation; its 79.44% yield is below the 82.45% prediction. The result concerns one byproduct/process system rather than universal extraction efficiency. Gelatin/coacervation optimization, capsule stability and simulated digestion are separate conventional downstream experiments, so they are not credited as additional AI-guided discoveries.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12271628/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Original ANN/search construction,17-experiment comparison/validation Methods and represented extraction Results actually read; later microencapsulation claims not attributed to learned optimization..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.