This is the authors' abstract. We don't add key points for this paper.
Innovation and entrepreneurship (IE) abilities are essential for student success in today’s competitive and continuously changing market. Higher education institutions are increasingly emphasizing IE development; however, present assessment systems frequently lack objectivity, adaptability, and the ability to integrate multiple student data sources, thereby restricting accurate evaluation. Traditional methods focus on subjective judgments or single-source measurements, which do not fully represent the multifaceted aspects of students’ innovative and entrepreneurial abilities. Furthermore, conventional models frequently underperform in predicting outcomes for unknown student characteristics. The Deep Residual-Dynamic Backtracking Search Optimization (Deep Residual-DBTSO) model combines ResNet-based deep learning with a DBTSO algorithm for hyperparameter optimization. Min-Max normalization and one-hot encoding are used to preprocess multi-source student data such as academic performance, project outputs, engagement in innovative activities, social interactions, and internet conduct. DBTSO dynamically optimizes important parameters like learning rate, dropout, and network depth to improve model generalization and avoid overfitting. The improved model captures complex semantic and behavioral links, allowing for reliable evaluation and prediction of students’ IE capabilities. Experimental findings reveal that Deep Residual-DBTSO outperforms conventional models with 96.85% accuracy, 95.90% sensitivity, 93.25% specificity, and 90.5% MCC. The approach captures complicated correlations across several student attributes, resulting in reliable, consistent, and generalizable forecasts of IE skills. The Deep Residual-DBTSO framework is an effective tool for educators, administrators, and policymakers, enabling focused interventions, tailored learning pathways, and creating an innovative academic environment.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Dai et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: