Optimization of music education and management decision support system based on machine learning

Abstract

This study designed and implemented a music education and management decision support system based on machine learning. By constructing a hierarchical processing framework and a multimodal hybrid model, the LSTM temporal network and the attention mechanism were innovatively integrated for learning behavior analysis. Combined with Bayesian hyperparameter optimization and compound loss function design, the system achieved a skill assessment accuracy of 92.4% and a recommendation hit rate of 85.7% in real-world scenario experiments. The response time was optimized to less than 120ms, verifying the effectiveness of machine learning technology in the dynamic allocation of educational resources. The proposed multi-objective optimization paradigm (simultaneously minimizing system delay and maximizing teaching effect) provides a scalable technical path for intelligent education management.