arXiv · 2026 · Preprint

Connectome-to-Function: Conditional Generative Latent Representations for Reservoir Computing

Zhuolin Yu, Xingyu Liu, Yuanhao Jia, Yunhang Xiao, Hairuo Xue, Feihan Sun, Guozhang Chen

A conditional generative latent model compresses sparse connectomes to link synaptic graph motifs directly with reservoir computing capacities.

The scientific question

How do specific synaptic connectivity motifs govern distinct computational capacities?

Brain connectomes are sparse, high-dimensional, and sensitive to local variation, making hand-crafted descriptors insufficient for both circuit generation and functional prediction.

Connectomes encoded in a structural latent space related to reservoir computation.
Figure 1 from the preprint ↗

The core idea

Conditional Latent Connectome Encoding

The framework encodes connectome graphs into a compact latent space conditioned on node-level attributes, reconstructing observed wiring and synthesizing candidate circuits for reservoir computing evaluation.

What we found

Task-Specific Structural Signatures

The latent space captures functional variation across reservoir-computing experiments, and interpretability analysis shows that in the studied examples, memory performance is associated with reciprocal recurrent connectivity whereas prediction and classification are more strongly associated with recurrent network spectral properties.

Continue the question

Related reading

Explore our research

Cite this work

Zhuolin Yu, Xingyu Liu, Yuanhao Jia, Yunhang Xiao, Hairuo Xue, Feihan Sun, Guozhang Chen. Connectome-to-Function: Conditional Generative Latent Representations for Reservoir Computing. arXiv, 2026. doi:10.48550/arXiv.2609.06093.