Data & Analytics

MIRAGE methodology — robust multi-modal architecture for translating fMRI-to-image models from seen visual decoding to mental imagery reconstruction. Demonstrates that SOTA on seen images doesn't guarantee SOTA on mental imagery and proposes a multi-modal, multi-loss architecture that excels at both. Use when researching: fMRI visual decoding, mental imagery reconstruction, brain decoding generalization, seen-to-imagery transfer, NSD-Imagery dataset, multi-modal brain decoding, vision model g...

Energy-first neural architecture design framework based on biological principles. Systematic validation across vision, text, neuromorphic, and physiological datasets with 2,203 experiments. Activation: energy-first, neural architecture, biological principles, energy-regularized, lambda sweep.

元学习上下文方法实现无需训练的跨被试脑解码。通过上下文学习实现训练无关的跨个体fMRI解码。适用于零样本脑解码、快速脑机接口、个体化神经科学。触发词：元学习脑解码、上下文学习、跨被试、训练无关、零样本。

Meta-Learning Biologically Plausible Plasticity Rules

Multilevel Covariate-Assisted Principal Regression (MCAP) for brain functional connectivity analysis. Handles hierarchically nested neuroimaging data, identifies cluster-specific projections, and models covariance matrix outcomes with subject-level covariates. Use when: analyzing lifespan brain connectivity, multilevel fMRI data, functional connectivity regression, covariance matrix outcomes.

Maximum entropy principle for neural network connectivity — normative framework for understanding how task constraints shape neural connectivity structure without gradient descent.

Multi-Axial Projective Sphere (MAPS) methodology for geometrically visualizing higher d-valued quantum state-space of qudits. Extends Bloch sphere to qudits with n projectional intersecting axes.

Framework for analyzing learning dynamics in low-rank RNNs via overlap space decomposition. Distinguishes loss-visible overlaps (determine activity/output/loss) from loss-invisible overlaps (encode training history). Enables understanding of why functionally equivalent networks learn differently. Activation: low-rank RNN learning, RNN overlap space, loss-visible invisible, RNN gradient descent dynamics, RNN learning theory, Ger Barak RNN.

Identifying structural design principles (local cycles) that shape computational abilities of recurrent neural networks. Found that 2- and 3-cycles strongly enhance computational power, and biologically-inspired interneurons dramatically increase capacity.

Methodology for using explainable AI attribution methods to understand and predict LLM-brain alignment during language processing. Uses gradient-based attribution to quantify word contributions to LLM predictions and predict fMRI data from narrative listening tasks. Activation: LLM-brain alignment, XAI attribution, fMRI prediction, gradient attribution, conductance analysis, language neuroscience

Skill for understanding and applying the research from arXiv:2607.14086 "Leveraging unlabelled data for generalizable neural population decoding"

Exact variance and Fano factor analytical formulae for arbitrary level crossings in stationary Gaussian processes. Extends the Kac-Rice mean crossing rate to capture clustering vs. regularity statistics, critical for neuronal spike train analysis, neural coding reliability, and stochastic neural dynamics. Use when analyzing spike train variability, threshold crossing statistics, or neural coding Fano factors.

Learning Dynamic Stability Landscapes in Synchronization Networks methodology - graph-to-image prediction paradigm for predicting stability landscapes from network topology. Pioneers image-like per-node stability landscapes beyond scalar indices. Applicable to neuroscience, power grids, biological synchronization. Activation: stability landscape, synchronization stability, graph-to-image prediction, dynamic stability, oscillator networks, power grid stability.

Comprehensive survey of machine learning methods for studying latent neural activity dynamics - from state-space models to deep generative models covering single-region dynamics, multi-region communication, and neural manifold geometry.

Large fluctuation theory for open quantum systems — analyzing atypical measurement outcomes in driven dissipative steady states. Shows large-deviation functions develop lines and surfaces with discontinuous derivatives, unlike equilibrium analytic Wigner functions. Provides framework for rare event statistics in non-equilibrium quantum systems. Activation: large fluctuations, open quantum systems, large-deviation, non-equilibrium, driven dissipative, Wigner function, rare events, steady state...

Kuramoto-von Mises时间序列模型用于耦合振荡器的概率建模。无需假设热力学平衡，通过Langevin动力学构造实现非平衡 regime的准确建模，在高采样率下具有闭式代数解。

Kuramoto模型脑网络相位动力学分析方法论。使用振荡器同步框架研究脑网络相位耦合，分析催产素等神经调节物质对脑网络动态的影响。适用于脑网络同步性分析、神经调节研究、网络神经科学。触发词：Kuramoto模型、脑网络、相位耦合、同步性、神经调节、催产素、oxytocin、phase coupling、synchronization、brain network dynamics。

Mean-field chaos 的预测性理论框架。证明随机循环网络的确定性混沌可通过连续历史唯一预测未来,展开功率谱到 Krylov 状态空间暴露潜在确定性组织。区分微观敏感性和预测复杂性。

Koopman-von Neumann (KvN) molecular dynamics methodology for computing Green-Kubo transport coefficients as quantum algorithm readout problems. Formulates classical NVE and NVT dynamics as unitary evolutions on Hilbert spaces, enabling quantum speedup for molecular property estimation with O(log(1/ε)) qubit scaling.

Conserved Kinematic Representations for Zero-Shot Decoding in Handwriting BCIs. Methodology aligning neural activity to imagined kinematics for zero-shot capable ML decoding of unseen characters in BCI systems. Use when: researching brain-computer interfaces, motor cortex representations, zero-shot decoding, handwriting BCIs, kinematic primitives, logographic language neuroprosthetics, compositional motor control. Keywords: zero-shot BCI, kinematic representation, handwriting decoding, motor ...

Geometric analysis of attractor boundaries and storage capacity limits in kernel Hopfield networks trained with Kernel Logistic Regression (KLR). Covers attractor basin geometry, Ridge of Optimization, morphing analysis, SNR vs Cover's theorem, and dynamical stability. Activation: kernel Hopfield, KLR associative memory, attractor basin geometry, Ridge of Optimization, morphing analysis Hopfield, storage capacity Hopfield, SNR analysis Hopfield, Cover's theorem associative memory, crosstalk n...

Kernel Hopfield networks: geometric analysis of attractor boundaries and storage capacity limits. KLR-trained associative memories with P/N ~16 for random sequences and ~20 for structured data. Trigger words: kernel Hopfield, associative memory, KLR, attractor basin, storage capacity, kernel logistic regression.

Just EEG Transformer (JET) — generative EEG framework using conditional flow matching to model neural signals as continuous trajectories, preserving spectral structure, temporal stationarity, and signal statistics. ICML 2026. Reduces TS-FID by >40% on large-scale benchmarks. arXiv:2605.21280

Stochastic Cortical Self-Reconstruction (SCSR) framework for personalized mapping of gray matter atrophy in neurodegenerative disorders. Enables individualized healthy reference estimation directly from observed cortical thickness at vertex level, allowing detection of subtle subject-specific deviations. Evaluates generalization and transferability across populations (UK Biobank to Chinese dataset) with multiple training strategies and reconstruction backbones.