The field of predictive modeling for complex systems is rapidly advancing, with a focus on developing innovative methods and techniques to improve forecast accuracy and interpretability. Recent developments have highlighted the potential of deep learning architectures, such as those integrating attention mechanisms and physics-informed neural networks, to enhance predictive capabilities for complex phenomena like sea ice dynamics and ocean front forecasting. Additionally, the use of neural operators and Fourier neural operators has shown promise in simulating and predicting the evolution of interfaces and chaotic systems. Noteworthy papers include those proposing IceMamba, a deep learning architecture for seasonal sea ice forecasting, and CTP, a hybrid CNN-Transformer-PINN model for ocean front prediction, which have demonstrated state-of-the-art performance in their respective domains.
Advances in Predictive Modeling for Complex Systems
Sources
Seasonal Forecasting of Pan-Arctic Sea Ice with State Space Model
CTP: A hybrid CNN-Transformer-PINN model for ocean front forecasting
Predicting The Evolution of Interfaces with Fourier Neural Operators
Panda: A pretrained forecast model for universal representation of chaotic dynamics
Hypothesis on the Functional Advantages of the Selection-Broadcast Cycle Structure: Global Workspace Theory and Dealing with a Real-Time World
Predicting Dynamical Systems across Environments via Diffusive Model Weight Generation
Learning High-dimensional Ionic Model Dynamics Using Fourier Neural Operators
Stochastic Fractional Neural Operators: A Symmetrized Approach to Modeling Turbulence in Complex Fluid Dynamics
Fourier-Invertible Neural Encoder (FINE) for Homogeneous Flows
SynEVO: A neuro-inspired spatiotemporal evolutional framework for cross-domain adaptation
Neural Field Equations with random data
FlowMixer: A Constrained Neural Architecture for Interpretable Spatiotemporal Forecasting
Dynamic Reservoir Computing with Physical Neuromorphic Networks