arXiv subjects

Bruce J. West

Publications and source records attributed to Bruce J. West.

At least 19 recordsLinked to original sources

arXiv2026

Complexity synchronization as a diagnostic and control principle for adaptive systems

Adaptive systems can exhibit similar levels of performance while relying on fundamentally different internal modes of coordination. Standard metrics such as average cooperation or payoff indicate whether a system succeeds, but do not reveal how coordination is organized across interacting components or which adaptive variables should be targeted when performance fails. Here we propose complexity synchronization (CS), the synchronization of evolving temporal complexity across coupled variables, as a diagnostic and intervention guiding principle for adaptive systems. We test this idea in an adaptive multi agent system composed of Selfish Algorithm agents interacting in a reduced Predator Prey model with a Prisoners Dilemma like payoff structure. Temporal complexity is quantified using sliding window modified diffusion entropy analysis (MDEA) and detrended fluctuation analysis (DFA). CS is defined as the correlation between the resulting time dependent scaling exponents. In the high-interaction regime, MDEA-based CS increases with cooperative performance, whereas DFA based CS captures a distinct persistence dominated coordination mode. Our results show that CS can reveal functionally relevant subsystems and provide a principled basis for targeted repair. More broadly, CS offers a general diagnostic and engineering framework for understanding and controlling coordination in biological, social, human machine, and other adaptive systems.

nlin.AO↗

arXiv2026

Fractional Order Federated Learning for Battery Electric Vehicle Energy Consumption Modeling

Federated learning on connected electric vehicles (BEVs) faces severe instability due to intermittent connectivity, time-varying client participation, and pronounced client-to-client variation induced by diverse operating conditions. Conventional FedAvg and many advanced methods can suffer from excessive drift and degraded convergence under these realistic constraints. This work introduces Fractional-Order Roughness-Informed Federated Averaging (FO-RI-FedAvg), a lightweight and modular extension of FedAvg that improves stability through two complementary client-side mechanisms: (i) adaptive roughness-informed proximal regularization, which dynamically tunes the pull toward the global model based on local loss-landscape roughness, and (ii) non-integer-order local optimization, which incorporates short-term memory to smooth conflicting update directions. The approach preserves standard FedAvg server aggregation, adds only element-wise operations with amortizable overhead, and allows independent toggling of each component. Experiments on two real-world BEV energy prediction datasets, VED and its extended version eVED, show that FO-RI-FedAvg achieves improved accuracy and more stable convergence compared to strong federated baselines, particularly under reduced client participation.

cs.LG↗

arXiv2026

When Gradient Clipping Becomes a Control Mechanism for Differential Privacy in Deep Learning

Privacy-preserving training on sensitive data commonly relies on differentially private stochastic optimization with gradient clipping and Gaussian noise. The clipping threshold is a critical control knob: if set too small, systematic over-clipping induces optimization bias; if too large, injected noise dominates updates and degrades accuracy. Existing adaptive clipping methods often depend on per-example gradient norm statistics, adding computational overhead and introducing sensitivity to datasets and architectures. We propose a control-driven clipping strategy that adapts the threshold using a lightweight, weight-only spectral diagnostic computed from model parameters. At periodic probe steps, the method analyzes a designated weight matrix via spectral decomposition and estimates a heavy-tailed spectral indicator associated with training stability. This indicator is smoothed over time and fed into a bounded feedback controller that updates the clipping threshold multiplicatively in the log domain. Because the controller uses only parameters produced during privacy-preserving training, the resulting threshold updates are post-processing and do not increase privacy loss beyond that of the underlying DP optimizer under standard composition accounting.

cs.LG↗

arXiv2026

Roughness-Informed Federated Learning

Federated Learning (FL) enables collaborative model training across distributed clients while preserving data privacy, yet faces challenges in non-independent and identically distributed (non-IID) settings due to client drift, which impairs convergence. We propose RI-FedAvg, a novel FL algorithm that mitigates client drift by incorporating a Roughness Index (RI)-based regularization term into the local objective, adaptively penalizing updates based on the fluctuations of local loss landscapes. This paper introduces RI-FedAvg, leveraging the RI to quantify the roughness of high-dimensional loss functions, ensuring robust optimization in heterogeneous settings. We provide a rigorous convergence analysis for non-convex objectives, establishing that RI-FedAvg converges to a stationary point under standard assumptions. Extensive experiments on MNIST, CIFAR-10, and CIFAR-100 demonstrate that RI-FedAvg outperforms state-of-the-art baselines, including FedAvg, FedProx, FedDyn, SCAFFOLD, and DP-FedAvg, achieving higher accuracy and faster convergence in non-IID scenarios. Our results highlight RI-FedAvg's potential to enhance the robustness and efficiency of federated learning in practical, heterogeneous environments.

cs.LG↗

arXiv2026

Statistical Roughness-Informed Machine Unlearning

Machine unlearning aims to remove the influence of a designated forget set from a trained model while preserving utility on the retained data. In modern deep networks, approximate unlearning frequently fails under large or adversarial deletions due to pronounced layer-wise heterogeneity: some layers exhibit stable, well-regularized representations while others are brittle, undertrained, or overfit, so naive update allocation can trigger catastrophic forgetting or unstable dynamics. We propose Statistical-Roughness Adaptive Gradient Unlearning (SRAGU), a mechanism-first unlearning algorithm that reallocates unlearning updates using layer-wise statistical roughness operationalized via heavy-tailed spectral diagnostics of layer weight matrices. Starting from an Adaptive Gradient Unlearning (AGU) sensitivity signal computed on the forget set, SRAGU estimates a WeightWatcher-style heavy-tailed exponent for each layer, maps it to a bounded spectral stability weight, and uses this stability signal to spectrally reweight the AGU sensitivities before applying the same minibatch update form. This concentrates unlearning motion in spectrally stable layers while damping updates in unstable or overfit layers, improving stability under hard deletions. We evaluate unlearning via behavioral alignment to a gold retrained reference model trained from scratch on the retained data, using empirical prediction-divergence and KL-to-gold proxies on a forget-focused query set; we additionally report membership inference auditing as a complementary leakage signal, treating forget-set points as should-be-forgotten members during evaluation.

cs.LG↗

arXiv2024

Complexity synchronization analysis of neurophysiological data: Theory and methods

We apply modified diffusion entropy analysis (MDEA) to assess multifractal dimensions of ON time series (ONTS) and complexity synchronization (CS) analysis to infer information transfer among ONs that are part of a network of organ networks (NoONs). The purpose of this paper is to advance the validation, standardization, and repeatability of MDEA and CS analysis of heterogeneous neurophysiological time series data. Results from processing these datasets show that the complexity of brain, heart, and lung ONTS significantly co-vary over time during cognitive task performance but that certain principles, guidelines, and strategies for the application of MDEA analysis need consideration.

q-bio.NC↗

arXiv2024

Complexity Control

We introduce a dynamic model for complexity control (CC) between systems, represented by time series characterized by different temporal complexity measures, as indicated by their respective inverse power law (IPL) indices. Given the apparent straightforward character of the model and the generality of the result, we formulate a hypothesis based on the closeness of the scaling measures of the model to the empirical complexity measures of the human brain. CC is a proper model for describing the recent experimental results, such as the rehabilitation in walking arm in arm and the complexity synchronization effect. The CC effect can lead to the design of mutual-adaptive signals to restore the misaligned complexity of maladjusted organ networks or, on the other hand, to disrupt the complexity of a malicious system and lower its intelligent behavior.

nlin.AO↗

arXiv2024

Cell Motility in Cancer, Crucial Events, Criticality, and Lévy Walks

The analysis of glioblastoma (GB) cell locomotion and its modeling inspired by Levy random walks is presented herein. We study such walks occurring on a two-dimensional plane where the walk is similar to the motion of a bird flying with a constant velocity, but with random changes of direction in time. The intelligence of the bird is signaled by the instantaneous changes of flying direction, which become invisible in the time series obtained by projecting the 2D walk either on the x axis or the y axis. We establish that the projected 1D time series share the statistical complexity of time series frequently used to monitor physiological processes, shedding light on the role of crucial events (CE-s) in pathophysiology. Such CE-s are signified by abrupt changes of flying direction which are invisible in the 1D physiological time series. We establish a connection between the complex scaling index δgenerated by the CE-s through μ_{R} = 2 - δ, where μ_{R} is the inverse power law index of the probability density function of the time interval between consecutive failures of the process of interest. We argue that the identification of empirical indices along with their theoretical relations afford important measures to control cancer.

nlin.AO↗

arXiv2023

Complexity Synchronization in Emergent Intelligence

In this work, we use a simple multi-agent-based model (MABM), implementing selfish algorithm (SA) agents, to create an adaptive environment and show, using modified diffusion entropy analysis (MDEA), that the mutual-adaptive interaction between the parts of such a network manifests complexity synchronization (CS). CS has been experimentally shown to exist among organ-networks (ONs) of the brain (neurophysiology), lungs (respiration), and heart (cardiovascular reactivity) and to be explained theoretically as a synchronization of the multifractal scaling parameters characterizing each time series. Herein, we find the same kind of CS in the emergent intelligence (i.e., without macroscopic control and based on self-interest) between two groups of agents playing an anti-coordination game, thereby suggesting the potential for the same CS in real-world social phenomena and in human-machine interactions.

nlin.AO↗

arXiv2023

Unveiling Pseudo-Crucial Events in Noise-Induced Phase Transitions

Noise-induced phase transitions are common in various complex systems, from physics to biology. In this article, we investigate the emergence of crucial events in noise-induced phase transition processes and their potential significance for understanding complexity in such systems. We utilize the first-passage time technique and coordinate transformations to study the dynamics of the system and identify crucial events. Furthermore, we employ Diffusion Entropy Analysis, a powerful statistical tool, to characterize the complexity of the system and quantify the information content of the identified events. Our results show that the emergence of crucial events is closely related to the complexity of the system and can provide insight into its behavior. This approach may have applications in diverse fields, such as climate modeling, financial markets, and biological systems, where understanding the emergence of crucial events is of great importance.

physics.data-an↗

arXiv2022

Complexity Synchronization

The observational ubiquity of inverse power law spectra (IPL) in complex phenomena entails theory for dynamic fractal phenomena capturing their fractal dimension, dynamics, and statistics. These and other properties are consequences of the complexity resulting from nonlinear dynamic networks collectively summarized for biomedical phenomena as the Network Effect (NE) or focused more narrowly as Network Physiology. Herein we address the measurable consequences of the NE on time series generated by different parts of the brain, heart, and lung organ networks, which are directly related to their inter-network and intra-network interactions. Moreover, these same physiologic organ networks have been shown to generate crucial event (CE) time series, and herein are shown, using modified diffusion entropy analysis (MDEA), to have scaling indices with quasiperiodic changes in complexity, as measured by scaling indices, over time. Such time series are generated by different parts of the brain, heart, and lung organ networks, and the results do not depend on the underlying coherence properties of the associated time series but demonstrate a generalized synchronization of complexity. This high order synchrony among the scaling indices of EEG (brain), ECG (heart), and respiratory time series is governed by the quantitative interdependence of the multifractal behavior of the various physiological organs' network dynamics. This consequence of the NE opens the door for an entirely general characterization of the dynamics of complex networks in terms of complexity synchronization (CS) independently of the scientific, engineering, or technological context.

nlin.AO↗

arXiv2020

Selfish Algorithm and Emergence of Collective Intelligence

We propose a model for demonstrating spontaneous emergence of collective intelligent behavior from selfish individual agents. Agents' behavior is modeled using our proposed selfish algorithm ($SA$) with three learning mechanisms: reinforced learning ($SAL$), trust ($SAT$) and connection ($SAC$). Each of these mechanisms provides a distinctly different way an agent can increase the individual benefit accrued through playing the prisoner's dilemma game ($PDG$) with other agents. The $SA$ provides a generalization of the self-organized temporal criticality ($SOTC$) model and shows that self-interested individuals can simultaneously produce maximum social benefit from their decisions. The mechanisms in the $SA$ are self-tuned by the internal dynamics and without having a pre-established network structure. Our results demonstrate emergence of mutual cooperation, emergence of dynamic networks, and adaptation and resilience of social systems after perturbations. The implications and applications of the $SA$ are discussed.

nlin.AO↗

arXiv2019

Intelligence of small groups

Dunbar hypothesized that $150$ is the maximal number of people with whom one can maintain stable social relationships. We explain this effect as being a consequence of a process of self-organization between $N$ units leading their social system to the edge of phase transition, usually termed criticality. Criticality generates events, with an inter-event time interval distribution characterized by an inverse power law (IPL) index $μ_{S}<2$. These events break ergodicity and we refer to them as crucial events. The group makes decisions and the time persistence of each decision is given by another IPL distribution with IPL index $μ_{R}$, which is different from $μ_{S}$ if $N\neq 150$. We prove that when the number of interacting individuals is equal to $150$, these two different IPL indexes become identical, with the effect of generating the Kardar Parisi Zhang (KPZ) scaling $δ=1/3$. We argue this to be an enhanced form of intelligence, which generates efficient information transmission within the group. We prove the inflrmation transmission efficiency is maximal when $N=150$, the Dunbar number.

nlin.AO↗

arXiv2019

Complexity Matching and Requisite Variety

Complexity matching characterizes the role of information in interactions between systems and can be traced back to the 1957 Introduction to Cybernetics by Ross Ashby. We argue that complexity can be expressed in terms of crucial events, which are generated by the processes of spontaneous self-organization. Complex processes, ranging from biological to sociological, must satisfy the homeodynamic condition and host crucial events that have recently been shown to drive the information transport between complex systems. We adopt a phenomenological approach, based on the subordination to periodicity that makes it possible to combine homeodynamics and self-organization induced crucial events. The complexity of crucial events is defined by the waiting-time probability density function (PDF) of the intervals between consecutive crucial events, which have an inverse power law (IPL) PDF $ψ(τ)\propto 1/(τ)^{μ}$ with $1<μ<3$. We show that the action of crucial events has an effect compatible with the shared notion of complexity-induced entropy reduction, while making the synchronization between systems sharing the same complexity different from chaos synchronization. We establish the coupling between two temporally complex systems using a phenomenological approach inspired by models of swarm cognition and prove that complexity matching, namely sharing the same IPL index $μ$, facilitates the transport of information, generating perfect synchronization. This new form of complexity matching is expected to contribute significantly to progress in understanding and improving biofeedback therapies.

nlin.AO↗

arXiv2017

Signatures of Parkinson's disease in complexity of low-frequency fluctuations of postural sway velocity

Posturography is routinely used to qualitatively assess one of the cardinal symptoms of Parkinson's disease -- postural instability. While most measures of balance control are derived from displacement of the center of pressure there is evidence that such control is more likely to be velocity-based. We performed static posturographic tests (eyes open and eyes closed) during quiet standing in narrow stance for n=30 patients with Parkinson's disease (PDP) in the ON state and compared the results with those of n=30 age-matched senior controls (HSC) and n=60 young controls (HYC). We used differentiator filters to generate time series of low-frequency fluctuations of sway velocity and calculated their Lempel-Ziv complexity (LZC). With eyes closed, the mediolateral LZC of HSC 0.21 (0.02) was significantly higher than those of HYC 0.19 (0.02) and PDP 0.18 (002). Thus aging and PD have opposite effects on mediolateral LZC which strongly differentiates between HSC and PDP (92% sensitivity and 87% specificity).

q-bio.NC↗

arXiv2017

Complexity Measures of Music

We present a technique to search for the presence of crucial events in music, based on the analysis of the music volume. Earlier work on this issue was based on the assumption that crucial events correspond to the change of music notes, with the interesting result that the complexity index of the crucial events is mu ~ 2, which is the same inverse power-law index of the dynamics of the brain. The search technique analyzes music volume and confirms the results of the earlier work, thereby contributing to the explanation as to why the brain is sensitive to music, through the phenomenon of complexity matching. Complexity matching has recently been interpreted as the transfer of multifractality from one complex network to another. For this reason we also examine the mulifractality of music, with the observation that the multifractal spectrum of a computer performance is significantly narrower than the multifractal spectrum of a human performance of the same musical score. We conjecture that although crucial events are demonstrably important for information transmission, they alone are not suficient to define musicality, which is more adequately measured by the multifractality spectrum.

physics.soc-ph↗

arXiv2017

On the Dynamical Foundation of Multifractality

The crucial aspect of this demonstration is the discovery of renewal events, hidden in the computed dynamics of a multifractal metronome, which enables the replacement of the phenomenon of strong anticipation with a time delayed cross-correlation between the driven and the driving metronome. We establish that the phenomenon of complexity matching, which is the theme of an increasing number of research groups, has two distinct measures. One measure is the sensitivity of a complex system to environmental multi-fractality; another is the level of information transfer, between two complex networks at criticality. The cross-correlation function is evaluated in the ergodic long-time limit, but its delayed maximal value is the signature of information transfer occurring in the non ergodic short-time regime. It is shown that a more complex system transfers its multifractality to a less complex system while the reverse case is not possible.

nlin.AO↗

arXiv2017

Crucial events, randomness and multi-fractality in heartbeats

We study the connection between multi-fractality and crucial events. Multi-fractality is frequently used as a measure of physiological variability. Crucial events are known to play a fundamental role in the transport of information between complex networks. To establish a connection we focus on the special case of heartbeat time series and on the search for a diagnostic prescription to distinguish healthy from pathologic subjects. Over the last twenty years two apparently different diagnostic techniques have been established: the first is based on the observation that the multi-fractal spectrum of healthy patients is broader than the multi-fractal spectrum of pathologic subjects; the second is based on the observation that heartbeat dynamics are a superposition of crucial and Poisson events, with pathologic patients hosting Poisson events with larger probability than the healthy patients. In this paper, we prove that increasing the percentage of Poisson events hosted by heartbeats has the effect of making their multi-fractal spectrum narrower, thereby establishing that the two different diagnostic techniques are compatible with one another and, at the same time, establishing a dynamic interpretation of multi-fractal processes that has been previously overlooked.

nlin.AO↗