SynEVO: A neuro-inspired spatiotemporal evolutional framework for cross-domain adaptation

2025-05-21

AI Summary

Existing spatiotemporal learners exhibit weak cross-domain transferability and require retraining for each source domain. Method: We propose an evolvable brain-inspired spatiotemporal network integrating curriculum-based sample reordering, an elastic shared memory container, and a task-agnostic feature extractor to enable model growth and disentanglement of shared versus domain-specific representations. We theoretically derive the information-theoretic bound of cross-domain collective intelligence for the first time and design a dynamic self-adaptive coupling mechanism alongside a novel discrepancy metric to support continual multi-domain evolution and knowledge aggregation. Contribution/Results: Experiments demonstrate that our framework achieves up to a 42% improvement in cross-domain generalization performance. It establishes a novel NeuroAI paradigm for spatiotemporal modeling grounded in neuromorphic computing principles and disentangled representation learning.

Technology Category

• Machine Learning: Evolutionary Learning
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Models for Web evolution
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Discovering regularities from spatiotemporal systems can benefit various scientific and social planning. Current spatiotemporal learners usually train an independent model from a specific source data that leads to limited transferability among sources, where even correlated tasks requires new design and training. The key towards increasing cross-domain knowledge is to enable collective intelligence and model evolution. In this paper, inspired by neuroscience theories, we theoretically derive the increased information boundary via learning cross-domain collective intelligence and propose a Synaptic EVOlutional spatiotemporal network, SynEVO, where SynEVO breaks the model independence and enables cross-domain knowledge to be shared and aggregated. Specifically, we first re-order the sample groups to imitate the human curriculum learning, and devise two complementary learners, elastic common container and task-independent extractor to allow model growth and task-wise commonality and personality disentanglement. Then an adaptive dynamic coupler with a new difference metric determines whether the new sample group should be incorporated into common container to achieve model evolution under various domains. Experiments show that SynEVO improves the generalization capacity by at most 42% under cross-domain scenarios and SynEVO provides a paradigm of NeuroAI for knowledge transfer and adaptation.

Problem

• Enhancing cross-domain knowledge transfer in spatiotemporal systems
• Breaking model independence for shared cross-domain intelligence
• Improving generalization capacity via neuro-inspired evolutionary learning

Innovation

• Neuro-inspired synaptic evolutional spatiotemporal network
• Curriculum learning with elastic common container
• Adaptive dynamic coupler for model evolution