Research map / Probabilistic and scientific modeling / Dynamical systems
Identifying Connectivity Distributions from Neural Dynamics Using Flows
Timothy Doyeon Kim, Ulises Pereira Obilinovic, Yiliu Wang, Eric Todd SheaBrown, Uygar Sümbül
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add maximum-entropy flow over connectivity (Connector) on top of LINT with matched conditional covariance S | LINT-trained lrRNN · Quadstable attractor, silencing one cell type | flow-field MSE to ground truth: 0.0894 -> 0.0097 | Improved |
| add Connector to LINT for perturbed latent trajectories | LINT-trained lrRNN · Quadstable attractor, silencing one cell type | latent trajectory R2: 0.695 -> 0.929 | Improved |
| add Connector to LINT for recovering p(n) | LINT-trained lrRNN · Quadstable attractor | optimal transport distance W: 6.980 -> 0.056 | Improved |
| set Connector conditional covariance S different from ground-truth conditional covariance (ablation) | Connector · Synthetic attractors | connectivity distribution recovery | Hurt |
| use Connector-inferred connectivity for clustering instead of LINT connectivity matrix | lrRNN connectivity clustering · Quadstable attractor cell-type recovery | recovery of four cell types | Improved |
| use Connector instead of raw lrRNN connectivity for clustering | lrRNN distilled from FINDR · Rat frontal cortex decision-making | interpretability of cell types | Improved |
| use the proposed dissimilarity measure instead of Valente et al. effective-connectivity measure | connectivity dissimilarity · QA, BA, LC, RA synthetic datasets | lowest dissimilarity at true match | Improved |
| increase number of sampled neurons K with ridge coefficient c = 10^-2 (ablation) | N-hat estimate · Synthetic datasets | MSE of N-hat to ground-truth E[n|m] | Hurt |
| use small ridge coefficient (c = 10^-4 or 10^-6) with large K (ablation) | N-hat estimate · Synthetic datasets | MSE of N-hat to ground-truth E[n|m] | Improved |
| replace lrRNN with more general latent variable models (ablation) | connectivity inference · Context-dependent decision-making RNN | identifiability of connectivity | Hurt |
Closest papers by meaning
- Flow-field inference from neural data using deep recurrent networks (ICML 2025 · similarity 0.88)
- Learning Reveals Invisible Structure in Low-Rank RNNs (NeurIPS 2026 · similarity 0.88)
- Self-Supervised Discovery of Neural Circuits in Spatially Patterned Neural Responses with Graph Neural Networks (NeurIPS 2025 · similarity 0.87)
- How connectivity structure shapes rich and lazy learning in neural circuits (ICLR 2024 · similarity 0.87)
- Optimal Recurrent Network Topologies for Dynamical Systems Reconstruction (ICML 2024 · similarity 0.87)
- Partial observation can induce mechanistic mismatches in data-constrained models of neural dynamics (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
