arXiv subjects

Davide Bernardi

Publications and source records attributed to Davide Bernardi.

9 recordsLinked to original sources

arXiv2026

A multiscale theory based on metabolic scaling connects forest dynamics to tree-size distributions

Scaling relations linking species size, abundance, and resource availability are among the most robust empirical regularities in ecology. However, a mechanistic explanation for how these community-level laws emerge from ecological processes remains elusive. Here, we address this gap by developing a minimal spatially explicit dynamical framework for forest communities that incorporates seed dispersal, growth limited by local light availability, local competition, and global resource constraints grounded in metabolic scaling principles. By deriving an analytical solution for the tree-size distribution, we show that its stationary state exhibits two distinct power-law regimes whose exponents are controlled by the relative strength of resource and spatial competition. The crossover between these regimes is set by the interplay between seed injection and local resource availability, establishing an explicit link between the scaling exponent of the size distribution and forest condition. Finally, we show that boundary disturbances can break the ecological balance between competing species and induce effects that propagate deeply into the forest bulk, far beyond the single-plant dispersal range. Together, these results provide a unifying dynamical perspective on forest scaling laws with potential applications to a broad range of biological communities.

q-bio.PE↗

arXiv2026

Dispersal diversity buffers species vulnerability to local extinction

Predicting species persistence within ecological communities is a fundamental challenge for both empirical and theoretical ecology. Existing methods span from mechanistic models, whose parameters are difficult to estimate from data, to statistical tools whose context-specific parameters are less interpretable. Here, we present a general framework, grounded in the statistical physics of complex systems, that integrates the key processes governing species survival into a single measurable quantity: the competitive balance. This metric quantifies a focal species' vulnerability to competitive exclusion beyond what is captured by its abundance alone by incorporating the diversity of dispersal strategies and the structure of interspecific interactions within the community. Crucially, it can be inferred from spatial abundance data, thus circumventing the need to estimate species traits or dispersal parameters. Our results reveal that greater heterogeneity in dispersal strategies reduces vulnerability to competitive exclusion for a given abundance. Although we validate the framework using tropical and temperate forest data, it can be applied to a range of different ecosystems, providing a systemic and interpretable tool for assessing a context-dependent species vulnerability that accounts for its interactions with the entire community.

q-bio.PE↗

arXiv2026

Habitat heterogeneity and dispersal network structure as drivers of metacommunity dynamics

Spatial structure and species interactions jointly shape the dynamics and biodiversity of ecological systems, yet most theoretical models either neglect spatial heterogeneity or sacrifice analytical tractability. Here, we provide a unified microscopic, mechanistic framework for deriving effective metapopulation and metacommunity models from individual-based ecological dynamics on arbitrary dispersal networks. The resulting coarse-grained description features an effective dispersal kernel that encodes both microscopic dynamical parameters and network topology. Based on this framework, we demonstrate exact analytical results for species persistence in both homogeneous and heterogeneous landscapes, including a generalization of the classical concept of metapopulation capacity to non-uniform local extinction rates. Incorporating stochasticity arising from finite carrying capacities, we obtain a reduced one-dimensional description that reveals universal finite-size scaling laws for extinction times and fluctuations. Extending the approach to multiple competing species, we prove that in homogeneous environments monodominance can be avoided only in a fine-tuned, marginally stable coexistence state, and that the classic metapopulation capacity gives only a necessary but not sufficient condition for persistence. We demonstrate that heterogeneous habitats can support stable coexistence, but only above a critical level of heterogeneity. Finally, we outline how additional ecological processes can be systematically incorporated within the same formalism. Together, these results provide analytical benchmarks and a general route for constructing spatially explicit ecological theories based on an interpretable underlying mechanistic foundation.

q-bio.PE↗

arXiv2025

Finite size scaling of survival statistics in metapopulation models

Spatial metapopulation models are fundamental to theoretical ecology, enabling to study how landscape structure influences global species dynamics. Traditional models, including recent generalizations, often rely on the deterministic limit of stochastic processes, assuming large population sizes. However, stochasticity - arising from dispersal events and population fluctuations - profoundly shapes ecological dynamics. In this work, we extend the classical metapopulation framework to account for finite populations, examining the impact of stochasticity on species persistence and dynamics. Specifically, we analyze how the limited capacity of local habitats influences survival, deriving analytical expressions for the finite-size scaling of the survival probability near the critical transition between survival and extinction. Crucially, we demonstrate that the deterministic metapopulation capacity plays a fundamental role in the statistics of survival probability and extinction time moments. These results provide a robust foundation for integrating demographic stochasticity into classical metapopulation models and their extensions.

q-bio.PE↗

arXiv2025

A novel metric for species vulnerability and coexistence in spatially-extended ecosystems

We develop a theoretical framework to understand the persistence and coexistence of competitive species in a spatially explicit metacommunity model with a heterogeneous dispersal kernel. Our analysis, based on methods from the physics of disordered systems and non-Gaussian dynamical mean field theory, reveals that species coexistence is governed by a single key parameter, which we term competitive balance. From competitive balance, we derive a novel metric to quantitatively assess the vulnerability of a species, showing that abundance alone is not sufficient to determine it. Rather, a species' vulnerability crucially depends on the state of the metacommunity as a whole. We test our theory by analyzing two distinct tropical forest datasets, finding excellent agreement with our theoretical predictions. A key step in our analysis is the introduction of a new quantity - the competitive score - which disentangles the abundance distribution and enables us to circumvent the challenge of estimating both the colonization kernel and the joint abundance distribution. Our findings provide novel and fundamental insights into the ecosystem-level trade-offs underlying macroecological patterns and introduce a robust approach for estimating extinction risks.

q-bio.PE↗

arXiv2024

A Survey on Knowledge Editing of Neural Networks

Deep neural networks are becoming increasingly pervasive in academia and industry, matching and surpassing human performance on a wide variety of fields and related tasks. However, just as humans, even the largest artificial neural networks make mistakes, and once-correct predictions can become invalid as the world progresses in time. Augmenting datasets with samples that account for mistakes or up-to-date information has become a common workaround in practical applications. However, the well-known phenomenon of catastrophic forgetting poses a challenge in achieving precise changes in the implicitly memorized knowledge of neural network parameters, often requiring a full model re-training to achieve desired behaviors. That is expensive, unreliable, and incompatible with the current trend of large self-supervised pre-training, making it necessary to find more efficient and effective methods for adapting neural network models to changing data. To address this need, knowledge editing is emerging as a novel area of research that aims to enable reliable, data-efficient, and fast changes to a pre-trained target model, without affecting model behaviors on previously learned tasks. In this survey, we provide a brief review of this recent artificial intelligence field of research. We first introduce the problem of editing neural networks, formalize it in a common framework and differentiate it from more notorious branches of research such as continuous learning. Next, we provide a review of the most relevant knowledge editing approaches and datasets proposed so far, grouping works under four different families: regularization techniques, meta-learning, direct model editing, and architectural strategies. Finally, we outline some intersections with other fields of research and potential directions for future works.

cs.LG↗

arXiv2024

Spatially disordered environments stabilize competitive metacommunities

Metapopulation models have been instrumental in demonstrating the ecological impact of landscape structure on the survival of a focal species in complex environments. However, extensions to multiple species with arbitrary dispersal networks often rely on phenomenological assumptions limiting their scope. Here, we develop a multilayer network model of competitive dispersing metacommunities to investigate how spatially structured environments impact species coexistence and ecosystem stability. We show that homogeneous environments always lead to monodominance unless all species' fitness parameters are in an exact trade-off. However, this precise fine-tuning does not guarantee coexistence in generic heterogeneous environments. By introducing general spatial disorder in the model, we solve it exactly in the mean-field limit, finding that stable coexistence becomes possible in the presence of strong disorder. Crucially, coexistence is supported by the spontaneous localization of species through the emergence of ecological niches. Our results remain qualitatively valid in arbitrary dispersal networks, where topological features can improve species coexistence. Finally, we employ our model to study how correlated disorder promotes spatial ecological patterns in realistic terrestrial and riverine landscapes. Our work provides a novel framework to understand how landscape structure enables coexistence in metacommunities by acting as the substrate for ecological interactions.

q-bio.PE↗

arXiv2022

Alexa Teacher Model: Pretraining and Distilling Multi-Billion-Parameter Encoders for Natural Language Understanding Systems

We present results from a large-scale experiment on pretraining encoders with non-embedding parameter counts ranging from 700M to 9.3B, their subsequent distillation into smaller models ranging from 17M-170M parameters, and their application to the Natural Language Understanding (NLU) component of a virtual assistant system. Though we train using 70% spoken-form data, our teacher models perform comparably to XLM-R and mT5 when evaluated on the written-form Cross-lingual Natural Language Inference (XNLI) corpus. We perform a second stage of pretraining on our teacher models using in-domain data from our system, improving error rates by 3.86% relative for intent classification and 7.01% relative for slot filling. We find that even a 170M-parameter model distilled from our Stage 2 teacher model has 2.88% better intent classification and 7.69% better slot filling error rates when compared to the 2.3B-parameter teacher trained only on public data (Stage 1), emphasizing the importance of in-domain data for pretraining. When evaluated offline using labeled NLU data, our 17M-parameter Stage 2 distilled model outperforms both XLM-R Base (85M params) and DistillBERT (42M params) by 4.23% to 6.14%, respectively. Finally, we present results from a full virtual assistant experimentation platform, where we find that models trained using our pretraining and distillation pipeline outperform models distilled from 85M-parameter teachers by 3.74%-4.91% on an automatic measurement of full-system user dissatisfaction.

cs.CL↗

arXiv2018

Fast Flux Service Network Detection via Data Mining on Passive DNS Traffic

In the last decade, the use of fast flux technique has become established as a common practice to organise botnets in Fast Flux Service Networks (FFSNs), which are platforms able to sustain illegal online services with very high availability. In this paper, we report on an effective fast flux detection algorithm based on the passive analysis of the Domain Name System (DNS) traffic of a corporate network. The proposed method is based on the near-real-time identification of different metrics that measure a wide range of fast flux key features; the metrics are combined via a simple but effective mathematical and data mining approach. The proposed solution has been evaluated in a one-month experiment over an enterprise network, with the injection of pcaps associated with different malware campaigns, that leverage FFSNs and cover a wide variety of attack scenarios. An in-depth analysis of a list of fast flux domains confirmed the reliability of the metrics used in the proposed algorithm and allowed for the identification of many IPs that turned out to be part of two notorious FFSNs, namely Dark Cloud and SandiFlux, to the description of which we therefore contribute. All the fast flux domains were detected with a very low false positive rate; a comparison of performance indicators with previous works show a remarkable improvement.

cs.CR↗