CTMN-CTIS

CTMN-CTIS: Cognitive-Temporal Mastery Network for Personalized Mathematical Knowledge Prediction Overview CTMN-CTIS is an advanced personalized learning framework that integrates cognitive tracking and temporal modeling to accurately predict students’ mathematical knowledge mastery over time. The framework is built on two core components: CTMN (Cognitive Temporal Mastery Network) — models the cognitive engagement and temporal learning dependencies using recurrent neural networks and attention mechanisms. CTIS (Cognitive-Temporal Integration Strategy) — combines immediate cognitive states with long-term learning trajectories through Bayesian updating and recurrent modeling for robust, adaptive prediction. This approach enhances predictive accuracy compared to traditional methods and serves as a foundation for intelligent tutoring systems and personalized education platforms. ✨ Features Cognitive Temporal Mastery Network (CTMN) Utilizes RNN-based encoder-decoder architecture with multimodal input support (audio, visual, text). Employs attention mechanisms to dynamically weight learner interactions. Models cognitive state evolution and temporal learning progression in real-time. See Figure 1 (p.7) for the CTMN architecture schematic. Cognitive-Temporal Integration Strategy (CTIS) Integrates short-term cognitive response modeling with long-term temporal learning patterns. Combines static and dynamic branches through Neighborhood Cross Attention (NCA) and Dynamic-Static Interaction (DSI) modules. Implements Bayesian cognitive state updating to continuously refine predictions (Fig. 3, p.9). Uses regularization and joint optimization for model stability and adaptability. Multimodal Encoder Architecture Multi-scale convolutional feature extraction with cross-feature self-attention (CFSA). Supports incomplete feature alignment and missing data reconstruction. See Figure 2 (p.8) for multimodal encoder details. 📊 Datasets Dataset Description Purpose Cognitive Tracking in Mathematics Learning Records of learners’ cognitive load and performance over time Cognitive state modeling Temporal Patterns in Math Skill Acquisition Longitudinal data of mathematical skill development Temporal sequence modeling Personalized Learning Trajectories Student-specific performance paths Personalized prediction Cognitive & Temporal Dynamics in Math Mastery Integrated cognitive and temporal records Model evaluation   🚀 Usage Cognitive and temporal state vectors Knowledge mastery probability curves Confidence scores and attention maps Predicted learning trajectory 🧪 Applications Personalized learning systems Real-time mathematical knowledge monitoring Adaptive intelligent tutoring systems Cognitive and temporal learning analytics for education research 🧩 Model Components CTMN — multimodal encoder-decoder, cognitive state modeling (Fig. 1, p.7). CTIS — dynamic-static fusion with Bayesian cognitive tracking (Fig. 3–4, p.9–11). Graphical Propagation Layer — adaptive focus on critical input segments. Attention Mechanisms — temporal weighting and alignment. Policy Gradient Optimization — adaptive feedback loop. 📈 Performance Dataset Accuracy Recall F1 Score AUC Cognitive Tracking in Math 89.78 89.12 88.45 88.89 Temporal Patterns in Skill Acquisition 91.45 90.89 90.23 90.56 Personalized Learning Trajectories 89.34 88.78 88.12 88.45 Cognitive & Temporal Dynamics 91.23 90.67 90.12 90.45 The framework outperforms baseline models such as ResNet, ViT, I3D, BLIP, DenseNet, and MobileNet, as shown in Tables 1–2 (p.14). 🧭 Future Work Extend to other domains beyond mathematics (e.g., language learning, science education). Develop lightweight variants for deployment in low-resource educational settings. Integrate explainable AI components to enhance interpretability for educators. Expand real-time adaptive tutoring functionalities. 📜 License This project is licensed under the MIT License. 🙏 Acknowledgments This work was conducted at Ningbo University of Finance & Economics.Authors: Yifeng Shan, Mengze Zheng.This research was supported by Zhejiang Province Education Science Planning Project and Undergraduate Education Reform Project (2024).