一种基于注意力权重分配与自适应残差学习的基因组预测模型
By using a deep learning model based on attention weight allocation and adaptive residual learning, the problem of insufficient global dependency capture in high-dimensional sparse SNP data by traditional genome prediction models is solved, achieving higher-precision genome prediction and supporting biological breeding and genomics research.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG UNIV
- Filing Date
- 2025-04-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing genome prediction models struggle to effectively capture global dependencies and complex traits among genes when processing high-dimensional sparse SNP data. Traditional models have limited prediction accuracy, and the application of attention mechanisms in genome selection has not been fully explored.
A deep learning model based on attention weight allocation and adaptive residual learning is adopted. The SNP site weights are implicitly calculated through a multi-layer fully connected network and combined with adaptive residual connections to construct a genome prediction model that captures global dependencies and preserves the original information.
It significantly improves the accuracy and efficiency of genome prediction, enhances the precision of predicting complex traits, and provides stronger support for biological breeding and genomics research.
Smart Images
Figure CN120544667B_ABST