arXiv subjects

Qian-Yuan Tang

Publications and source records attributed to Qian-Yuan Tang.

8 recordsLinked to original sources

arXiv2026

Beyond variance: sensitivity-based dimensions in brain networks underlie individual differences in cognitive ability

Explaining individual differences in cognitive abilities requires both identifying brain parameters that vary across individuals and understanding how brain networks are recruited for specific tasks. Typically, task performance relies on the integration and segregation of functional subnetworks, often captured by parameters like regional excitability and connectivity. Yet, the high dimensionality of these parameters hinders pinpointing their functional relevance. Here, we apply stiff-sloppy analysis to human brain data, revealing that certain subtle parameter combinations ("stiff dimensions") powerfully influence neural activity during task processing, whereas others ("sloppy dimensions") vary more extensively but exert minimal impact. Using a pairwise maximum entropy model of task fMRI, we show that even small deviations in stiff dimensions-derived through Fisher Information Matrix analysis-govern the dynamic interplay of segregation and integration between the default mode network (DMN) and a working memory network (WMN). Crucially, separating a 0-back task (vigilant attention) from a 2-back task (working memory updating) uncovers partially distinct stiff dimensions predicting performance in each condition, along with a global DMN-WMN segregation shared across both tasks. Altogether, stiff-sloppy analysis challenges the conventional focus on large parameter variability by highlighting these subtle yet functionally decisive parameter combinations.

q-bio.NC↗

arXiv2026

Data-driven inference of brain dynamical states from the r-spectrum of correlation matrices

We present a data-driven framework to characterize large-scale brain dynamical states directly from correlation matrices at the single-subject level. By treating correlation thresholding as a percolation-like probe of connectivity, the approach tracks multiple cluster- and network-level observables and identifies a characteristic percolation threshold, rc, at which these signatures converge. We use $r_c$ as an operational and physically interpretable descriptor of large-scale brain dynamical state. Applied to resting-state fMRI data from a large cohort of healthy individuals (N = 996), the method yields stable, subject-specific estimates that covary systematically with established dynamical indicators such as temporal autocorrelations. Numerical simulations of a whole-brain model with a known critical regime further show that $r_c$ tracks changes in collective dynamics under controlled variations of excitability. By replacing arbitrary threshold selection with a criterion intrinsic to correlation structure, the r-spectra provides a physically grounded approach for comparing brain dynamical states across individuals.

q-bio.NC↗

arXiv2025

Finite-time scaling on low-dimensional map bifurcations

Recent work has introduced the concept of finite-time scaling to characterize bifurcation diagrams at finite times in deterministic discrete dynamical systems, drawing an analogy with finite-size scaling used to study critical behavior in finite systems. In this work, we extend the finite-time scaling approach in several key directions. First, we present numerical results for 1D maps exhibiting period-doubling bifurcations and discontinuous transitions, analyzing selected paradigmatic examples. We then define two observables, the finite-time susceptibility and the finite-time Lyapunov exponent, that also display consistent scaling near bifurcation points. The method is further generalized to special cases of 2D maps including the 2D Chialvo map, capturing its bifurcation between a fixed point and a periodic orbit, while accounting for discontinuities and asymmetric periodic orbits. These results underscore fundamental connections between temporal and spatial observables in complex systems, suggesting new avenues for studying complex dynamical behavior.

cond-mat.dis-nn↗

arXiv2025

Behavior of the scaling correlation functions under severe subsampling

Scale-invariance is a ubiquitous observation in the dynamics of large distributed complex systems. The computation of its scaling exponents, which provide clues on its origin, is often hampered by the limited available sampling data, making an appropriate mathematical description a challenge. This work investigates the behavior of correlation functions in fractal systems under conditions of severe subsampling. Analytical and numerical results reveal a striking robustness: the correlation functions continue to capture the expected scaling exponents despite substantial data reduction. This behavior is demonstrated numerically for the random 2-D Cantor set and the Sierpinski gasket, both consistent with exact analytical predictions. Similar robustness is observed in 1-D time series both synthetic and experimental, as well as in high-resolution images of a neuronal structure. Overall, these findings are broadly relevant for the structural characterization of biological systems under realistic sampling constraints.

physics.comp-ph↗

arXiv2023

On the Overlooked Structure of Stochastic Gradients

Stochastic gradients closely relate to both optimization and generalization of deep neural networks (DNNs). Some works attempted to explain the success of stochastic optimization for deep learning by the arguably heavy-tail properties of gradient noise, while other works presented theoretical and empirical evidence against the heavy-tail hypothesis on gradient noise. Unfortunately, formal statistical tests for analyzing the structure and heavy tails of stochastic gradients in deep learning are still under-explored. In this paper, we mainly make two contributions. First, we conduct formal statistical tests on the distribution of stochastic gradients and gradient noise across both parameters and iterations. Our statistical tests reveal that dimension-wise gradients usually exhibit power-law heavy tails, while iteration-wise gradients and stochastic gradient noise caused by minibatch training usually do not exhibit power-law heavy tails. Second, we further discover that the covariance spectra of stochastic gradients have the power-law structures overlooked by previous studies and present its theoretical implications for training of DNNs. While previous studies believed that the anisotropic structure of stochastic gradients matters to deep learning, they did not expect the gradient covariance can have such an elegant mathematical structure. Our work challenges the existing belief and provides novel insights on the structure of stochastic gradients in deep learning.

cs.LG↗

arXiv2022

On the Power-Law Hessian Spectrums in Deep Learning

It is well-known that the Hessian of deep loss landscape matters to optimization, generalization, and even robustness of deep learning. Recent works empirically discovered that the Hessian spectrum in deep learning has a two-component structure that consists of a small number of large eigenvalues and a large number of nearly-zero eigenvalues. However, the theoretical mechanism or the mathematical behind the Hessian spectrum is still largely under-explored. To the best of our knowledge, we are the first to demonstrate that the Hessian spectrums of well-trained deep neural networks exhibit simple power-law structures. Inspired by the statistical physical theories and the spectral analysis of natural proteins, we provide a maximum-entropy theoretical interpretation for explaining why the power-law structure exist and suggest a spectral parallel between protein evolution and training of deep neural networks. By conducing extensive experiments, we further use the power-law spectral framework as a useful tool to explore multiple novel behaviors of deep learning.

cs.LG↗

arXiv2020

Calibrated Intervention and Containment of the COVID-19 Pandemic

Within a short period of time, COVID-19 grew into a world-wide pandemic. Transmission by pre-symptomatic and asymptomatic viral carriers rendered intervention and containment of the disease extremely challenging. Based on reported infection case studies, we construct an epidemiological model that focuses on transmission around the symptom onset. The model is calibrated against incubation period and pairwise transmission statistics during the initial outbreaks of the pandemic outside Wuhan with minimal non-pharmaceutical interventions. Mathematical treatment of the model yields explicit expressions for the size of latent and pre-symptomatic subpopulations during the exponential growth phase, with the local epidemic growth rate as input. We then explore reduction of the basic reproduction number R_0 through specific disease control measures such as contact tracing, testing, social distancing, wearing masks and sheltering in place. When these measures are implemented in combination, their effects on R_0 multiply. We also compare our model behaviour to the first wave of the COVID-19 spreading in various affected regions and highlight generic and less generic features of the pandemic development.

q-bio.PE↗

arXiv2016

Critical fluctuations in proteins native states

We study a large data set of protein structure ensembles of very diverse sizes determined by nuclear magnetic resonance. By examining the distance-dependent correlations in the displacement of residues pairs and conducting finite size scaling analysis it was found that the correlations and susceptibility behave as in systems near a critical point implying that, at the native state, the motion of each amino acid residue is felt by every other residue up to the size of the protein molecule. Furthermore certain protein's shapes corresponding to maximum susceptibility were found to be more probable than others. Overall the results suggest that the protein's native state is critical, implying that despite being posed near the minimum of the energy landscape, they still preserve their dynamic flexibility.

physics.bio-ph↗