Towards Retraining-Free Rna Modification Prediction with Incremental Learning

• DOI: 10.2139/ssrn.4538477
• Publication Date: 2023-08-15T10:23:40Z
• Authors: Jianbo Qiao; Junru Jin; Haoqing Yu; Leyi Wei
• Keywords: Retraining

RNA modifications are important for deciphering the function of cells and their regulatory mechanisms. In recent years, researchers have developed many deep learning methods to identify specific modifications. However, these require model retraining each new modification cannot progressively newly identified To address this challenge, we propose an innovative incremental framework that incorporates multiple methods. Our experimental results confirm efficacy strategies in addressing challenge. By uniquely targeting 10 types a class setting, our exhibits superior performance. Notably, it can be extended category methylation predictions without need with previous data, improving computational efficiency. Through accumulation knowledge, is able evolve continuously learn differences across methylation, mitigating problem catastrophic forgetting during training. Overall, provides various alternatives enhance prediction novel illuminates potential tacking numerous genome data.