Research map / Probabilistic and scientific modeling / Dynamical systems
Functional connectivity modeling
Infers interactions between brain regions with latent factor and multi-timescale models.
50 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Dynamical systems, keyword cluster “brain · activity · neurons”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
11 papers in 2023–24 and 21 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 4.9% to 5.5%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · A state-space model of cross-region dynamic connectivity in MEG/EEG (NeurIPS 2016 · 16 citations)
- 2017 · Cross-Spectral Factor Analysis (NeurIPS 2017 · 14 citations)
- 2018 · Parameter-Free Centralized Multi-Task Learning for Characterizing Developmental Sex Differences in Resting State Functional Connectivity (AAAI 2018 · 7 citations)
- 2019 · Manifold-regression to predict from MEG/EEG brain signals without source modeling (NeurIPS 2019 · 25 citations)
- 2020 · Uncovering the Topology of Time-Varying fMRI Data using Cubical Persistence (NeurIPS 2020 · 34 citations)
- 2021 · CCNLab: A Benchmarking Framework for Computational Cognitive Neuroscience (NeurIPS 2021 · 3 citations)
- 2023 · An information-theoretic quantification of the content of communication between brain regions (NeurIPS 2023 · 19 citations)
- 2024 · BrainLM: A foundation model for brain activity recordings (ICLR 2024 · 45 citations)
- 2025 · Accurate Identification of Communication Between Multiple Interacting Neural Populations (ICML 2025 · 0 citations)
- 2026 · Uncovering Latent Communication Patterns in Brain Networks via Adaptive Flow Routing (ICML 2026 · 0 citations)
Earliest papers here
- A state-space model of cross-region dynamic connectivity in MEG/EEG (NeurIPS 2016 · 16 citations)
- Generalized Correspondence-LDA Models (GC-LDA) for Identifying Functional Regions in the Brain (NeurIPS 2016 · 13 citations)
- Cross-Spectral Factor Analysis (NeurIPS 2017 · 14 citations)
What has been tried
32 papers here report 281 measured changes (a component added, removed or swapped, compared with a baseline): helped 64%, no clear effect 9%, hurt 24%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add pretraining | UKB future-state forecasting · MSE | 0.695 -> 0.648 | Helped | BrainLM: A foundation model for brain activity recordings (ICLR 2024 · 45 citations) |
| use adaptive ascending (SALK) instead of Local fusion | SEED-VIG · ACC | 81.93 -> 85.15 | Helped | A Local-Ascending-Global Learning Strategy for Brain-Computer Interface (AAAI 2024 · 20 citations) |
| Add the multi-scale spectral module to the temporal-spatial module | Resting-state inter-subject · Average Pearson R | 0.151 -> 0.473 | Helped | NeuroBOLT: Resting-state EEG-to-fMRI Synthesis with Multi-dimensional Feature Mapping (NeurIPS 2024 · 9 citations) |
| use matrix trace cost rather than log-determinant cost | EEG-EMG-Fusion inter-subject 3-class · accuracy | 0.985±0.003 -> 0.994±0.002 | Helped | Learning Cortico-Muscular Dependence through Orthonormal Decomposition of Density Ratios (NeurIPS 2024 · 1 citation) |
| Automatically infer unobserved inputs instead of manually specifying all stimulus signals | Memory network · effectome cosine similarity | Helped | Accurate Identification of Communication Between Multiple Interacting Neural Populations (ICML 2025 · 0 citations) | |
| Add behavior-supervised first-phase latent learning | WFCI 1 · behavior MSE | 0.5413 ± 0.0185 -> 0.4955 ± 0.0254 | Helped | Dynamical Modeling of Behaviorally Relevant Spatiotemporal Patterns in Neural Imaging Data (ICML 2025 · 0 citations) |
| Use P=1 instead of P≥2 | Synthetic Data · test log-likelihood | Hurt | Learning Time-Varying Multi-Region Brain Communications via Scalable Markovian Gaussian Processes (ICML 2025 · 0 citations) | |
| Use all BTH scales S1–S5 rather than only a single scale | evaluated scale configurations · performance | Helped | THD-BAR: Topology Hierarchical Derived Brain Autoregressive Modeling for EEG Generic Representations (NeurIPS 2025 · 0 citations) | |
| Use consistency loss rather than regularization loss alongside reconstruction loss | Synthetic dataset · DFE | Helped | Inpainting the Neural Picture: Inferring Unrecorded Brain Area Dynamics from Multi-Animal Datasets (NeurIPS 2025 · 0 citations) | |
| jointly identify ensembles and dynamics instead of separating the two steps | Synthetic data · correlation with ground-truth components | Helped | CREIMBO: Cross-Regional Ensemble Interactions in Multi-view Brain Observations (ICLR 2025 · 0 citations) | |
| replace the Transformer encoder with a bidirectional RNN | GLM Simulation · test negative log-likelihood | 3.008 × 10^7 -> 3.032 × 10^7 | Hurt | Identifying interactions across brain areas while accounting for individual-neuron dynamics with a Transformer-based variational autoencoder (NeurIPS 2025) |
| Set the number of states to 2 instead of 1 | Synthetic data · test log-likelihood | Helped | Multi-Region Markovian Gaussian Process: An Efficient Method to Discover Directional Communications Across Multiple Brain Regions (ICML 2024) |
What reviewers will expect
Datasets and baselines named in the 17 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- mp-srSLDS · 2 of 17 papers
- mDLAG · 2 of 17 papers
- LFADS · 2 of 17 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 32 papers; from OpenAlex and the papers' affiliations).
- University of North Carolina at Chapel Hill · 3 papers
- Georgia Institute of Technology · 3 papers
- Yale University · 2 papers
- Stanford University · 2 papers
- University of North Carolina, Chapel Hill · 2 papers
- University of Washington · 2 papers
- Carnegie Mellon University · 2 papers
- Imperial College London · 2 papers
Use it
Download BibTeX of 37 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Latent models of neural activity (70 papers)
- Recurrent circuit models of inference (70 papers)
- Low-rank and diffusion neural dynamics (53 papers)
- Synaptic plasticity learning rules (72 papers)
Elsewhere in the topic
- Physics-informed dynamics learning (127 papers)
- Latent dynamics models (111 papers)
- State-space inference for stochastic systems (104 papers)
- Switching and law-discovery dynamics (81 papers)
- Latent-space control and planning (78 papers)
- Data assimilation from sparse observations (58 papers)
- Stochastic population dynamics (53 papers)
- Stable contractive dynamics (42 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).