Sadamori Kojaku, Filippo Radicchi, Yong-Yeol Ahn & Santo Fortunato 

Nature Communications volume 15, Article number: 9446 (2024)

Recent advances in machine learning research have produced powerful neural graph embedding methods, which learn useful, low-dimensional vector representations of network data. These neural methods for graph embedding excel in graph machine learning tasks and are now widely adopted. However, how and why these methods work—particularly how network structure gets encoded in the embedding—remain largely unexplained. Here, we show that node2vec—shallow, linear neural network—encodes communities into separable clusters better than random partitioning down to the information-theoretic detectability limit for the stochastic block models. We show that this is due to the equivalence between the embedding learned by node2vec and the spectral embedding via the eigenvectors of the symmetric normalized Laplacian matrix. Numerical simulations demonstrate that node2vec is capable of learning communities on sparse graphs generated by the stochastic blockmodel, as well as on sparse degree-heterogeneous networks. Our results highlight the features of graph neural networks that enable them to separate communities in the embedding space.

Read the full article at: www.nature.com

Gordana Dodig-Crnkovic 

The development of naturalistic approaches to complexity of life continues a lineage of thought from Prigogine’s thermodynamics to contemporary complexity science. The paper highlights the central themes of self-organization, emergence, and the interplay between physical, informational, and biological processes. Prigogine’s concept of dissipative structures and irreversibility provided a foundation for understanding complexity in physical systems, which later expanded into biology through Kauffman’s models of creativity and evolution. Margulis’s endosymbiosis theory illuminate the cooperative dynamics underpinning life’s complexity, while Walker’s work integrates thermodynamics and information theory to bridge the gap between chemistry and biology through multiscale interactions and adaptive dynamics. By synthesizing these perspectives, this article situates life as an emergent phenomenon shaped by interactions across scales, proposing a unified framework for understanding complexity in the natural world.

Read the full article at: www.preprints.org

Fabian Dvorak, Urs Fischbacher, Katrin Schmelz

The Economic Journal, ueae085,

We provide systematic insights on strategic conformist—as well as anti-conformist—behaviour in situations where people are evaluated, i.e., where an individual has to be selected for reward (e.g., promotion) or punishment (e.g., layoffs). To affect the probability of being selected, people may attempt to fit in or stand out in order to affect the chances of being noticed or liked by the evaluator. We investigate such strategic incentives for conformity or anti-conformity experimentally in three different domains: facts, taste and creativity. To distinguish conformity and anti-conformity from independence, we introduce a new experimental design that allows us to predict participants’ independent choices based on transitivity. We find that the prospect of punishment increases conformity, while the prospect of reward reduces it. Anti-conformity emerges in the prospect of reward, but only under specific circumstances. Similarity-based selection (i.e., homophily) is much more important for the evaluators’ decisions than salience. We also employ a theoretical approach to illustrate strategic key mechanisms of our experimental setting.

Read the full article at: academic.oup.com

Alexis Pietak, Michael Levin

Regulatory networks such as gene regulatory networks (GRNs) are critically important for efforts in biomedicine and synthetic biology. They have classically been viewed as mechanistic, “clockwork-like” systems, assumed to require direct changes to network topology via genetic modification to effect significant, stable changes in their output functions. This perspective limits therapeutic approaches, suggesting a need for alternative conceptual framing. Here we show how regulatory networks can behave as analog computational agents to perform sophisticated information processing, driven by patterns of stimulus inputs, without a change in network topology. We introduce and develop a new conceptual and computational framework for working with regulatory networks called the Regulatory Network Machine (RNM). Given a regulatory network model, our RNM framework enables the construction of detailed maps that embody the “software-like” nature of a regulatory network, providing easy identification of the specific interventions necessary to achieve desired outcomes. We demonstrate the use of our RNM framework to gain insights into important biological examples including yeast osmoadaptation, PI3K/AKT/mTor cross-signaling cascades, and embryonic stem cell differentiation. Importantly, we show how system-level outcomes can be induced in a biological system without requiring genetic rewiring. Our RNM approach also elucidates system factors that support the innate computational capabilities of regulatory networks, and ascertains the interventions that provide the most control for the least amount of effort. Ultimately, we hope to use insights gained from our RNM framework to expand the horizons of biomedicine, providing an effective avenue to move beyond “single-factor, single treatment” and “one-constant-dose” biomedical paradigms.

Read the full article at: www.researchgate.net