— A short overview of temporal graphs in the area of artificial intelligence, Temporal Neural Networks
Work in progress….. Still trying to understand
Introduction
Temporal graphs are used in advancing artificial intelligence to enable the modeling of dynamic systems where interactions evolve over time. These models are widely used across various domains, such as social networks, recommender systems, and epidemic modeling, to capture temporal dependencies and uncover insights that static graphs cannot provide.
The traditional research on computational temporal graph problems, is not always directly applicable, but metrics and graph structure theory is used a lot.
Expand on this
Applications in AI
- Social Network Analysis: Temporal graphs help understand how influence spreads over time, detect communities, and model user interactions.
- Recommender Systems: They enable dynamic user-item modeling, accounting for temporal behaviors to enhance recommendations.
- Traffic and Mobility Analysis: Modeling temporal road networks or public transit systems to predict traffic patterns and optimize routing.
- Health Informatics: Temporal graphs are applied to model patient health trajectories and disease progression in temporal biomedical networks.
“Temporal” Networks used in AI
Discrete-Time Models
Continuous-Time Models
Hybrid Models
Causality-Aware Models
Dynamic Graph Representation Learning Models
Probabilistic Models
Temporal Neural Networks (TNN)
TNNs are general-purpose neural architectures designed to model temporal data and fit into the Continuous-Time Models and Dynamic Graph Representation Learning Models categories. They often focus on sequences or event data and are typically paired with temporal graphs for specific use cases.
- Applications:
- Continuous-time predictions (e.g., time-series forecasting, sequential event prediction).
- Incorporating temporal dependencies into graph tasks like link prediction or node classification.