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

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.Code available at https://github.com/Rodger-Lau/SynEVO.

The growing amount and diversity of urban data brings in the requirement of model generalization, expansion and evolution for cross-temporal and cross-source domain transfer. It has been demonstrated that the process of acquiring skills in human brain is analogous to machine learning model evolution. This paper built a neuroscience-inspired deep learning framework, SynEVO, for efficient and effective model evolution.SynEVO inherits the most important information sharing scheme, synaptic structure in brain and implement three key mechanisms of brain, progressive curriculum learning, information transfer and expansion, as well as complementary structured learning to realize iterative information aggregation and expansion. These three components can work cooperatively to construct the overall synaptic structure for collective intelligence, and enable model evolution across temporal and source domains. Experiments show the collective intelligence increases the model generalization capacity under both source and temporal shifts by at 0.5% to 42%, and validate the efficient convergency of progressive curriculum learning. The extremely reduced memory cost, i.e., only 21.75% memory cost against SOTA on model training and evolution advances urban computing towards efficient model deployment and sustainable computing paradigm.