Search

Triadic interactions are general mechanisms by which a node or neuron can regulate directly the link or synapse between other two neurons. The regulation takes place in a familiar way by either depressing or facilitating synaptic transmission. Such interactions are ubiquitous in neural systems, accounting for axo-axonic synapses and tripartite synapses mediated by astrocytes, for instance, and have been related to neuronal and synaptic processes at different time-scales, including short and long-term synaptic plasticity. In the field of network science, triadic interactions have been shown to produce complex spatio-temporal patterns of connectivity. Here, we investigate the emergent behavior of an in silico neural medium constituted by a population of leaky integrate-and-fire neurons with triadic interactions. We observe that, depending on relevant parameters defining triadic interactions, different activity patterns emerge. These include i) a silent phase, ii) a low-activity phase in which complex spatio-temporal patterns of low neuronal firing rate emerge that propagate through the medium, iii) a high-activity phase characterized by complex spatio-temporal patterns of high neuronal firing rate that propagate through the neural medium as waves of high firing activity over a bulk of low activity neurons, and iv) a pseudo-blinking phase in which the neural medium switches between high and low activity states, in a similar fashion to up/down state transitions. Here we analyse in depth the features of such patterns and relate our findings to the recently proposed model of triadic percolation., Comment: 22 pages, 9 figures

Understanding the relationships between geometric structures and semantic concepts is crucial for building accurate models of complex environments. In indoors, certain spatial constraints, such as the relative positioning of planes, remain consistent despite variations in layout. This paper explores how these invariant relationships can be captured in a graph SLAM framework by representing high-level concepts like rooms and walls, linking them to geometric elements like planes through an optimizable factor graph. Several efforts have tackled this issue with add-hoc solutions for each concept generation and with manually-defined factors. This paper proposes a novel method for metric-semantic factor graph generation which includes defining a semantic scene graph, integrating geometric information, and learning the interconnecting factors, all based on Graph Neural Networks (GNNs). An edge classification network (G-GNN) sorts the edges between planes into same room, same wall or none types. The resulting relations are clustered, generating a room or wall for each cluster. A second family of networks (F-GNN) infers the geometrical origin of the new nodes. The definition of the factors employs the same F-GNN used for the metric attribute of the generated nodes. Furthermore, share the new factor graph with the S-Graphs+ algorithm, extending its graph expressiveness and scene representation with the ultimate goal of improving the SLAM performance. The complexity of the environments is increased to N-plane rooms by training the networks on L-shaped rooms. The framework is evaluated in synthetic and simulated scenarios as no real datasets of the required complex layouts are available., Comment: Submitted to ICRA 2025

Having prior knowledge of an environment boosts the localization and mapping accuracy of robots. Several approaches in the literature have utilized architectural plans in this regard. However, almost all of them overlook the deviations between actual as-built environments and as-planned architectural designs, introducing bias in the estimations. To address this issue, we present a novel localization and mapping method denoted as deviations-informed Situational Graphs or diS-Graphs that integrates prior knowledge from architectural plans even in the presence of deviations. It is based on Situational Graphs (S-Graphs) that merge geometric models of the environment with 3D scene graphs into a multi-layered jointly optimizable factor graph. Our diS-Graph extracts information from architectural plans by first modeling them as a hierarchical factor graph, which we will call an Architectural Graph (A-Graph). While the robot explores the real environment, it estimates an S-Graph from its onboard sensors. We then use a novel matching algorithm to register the A-Graph and S-Graph in the same reference, and merge both of them with an explicit model of deviations. Finally, an alternating graph optimization strategy allows simultaneous global localization and mapping, as well as deviation estimation between both the A-Graph and the S-Graph. We perform several experiments in simulated and real datasets in the presence of deviations. On average, our diS-Graphs outperforms the baselines by a margin of approximately 43% in simulated environments and by 7% in real environments, while being able to estimate deviations up to 35 cm and 15 degrees.

To establish general principles of optimal control of active matter, we study the elementary problem of moving an active particle by a trap with minimum work input. We show analytically that (open-loop) optimal protocols are not affected by activity, but work fluctuations are always increased. For closed-loop protocols, which rely on initial measurements of the self-propulsion, the average work has a minimum for a finite persistence time. Using these insights, we derive an optimal cyclic active information engine, which is found to have a higher precision and information efficiency when operated with a run-and-tumble particle than for an active Ornstein-Uhlenbeck particle and, we argue, than for any other type of active particle., Comment: 7 pages, 3 figures

As part of an ongoing worldwide effort to comprehend and monitor insect biodiversity, this paper presents the BIOSCAN-5M Insect dataset to the machine learning community and establish several benchmark tasks. BIOSCAN-5M is a comprehensive dataset containing multi-modal information for over 5 million insect specimens, and it significantly expands existing image-based biological datasets by including taxonomic labels, raw nucleotide barcode sequences, assigned barcode index numbers, geographical, and size information. We propose three benchmark experiments to demonstrate the impact of the multi-modal data types on the classification and clustering accuracy. First, we pretrain a masked language model on the DNA barcode sequences of the BIOSCAN-5M dataset, and demonstrate the impact of using this large reference library on species- and genus-level classification performance. Second, we propose a zero-shot transfer learning task applied to images and DNA barcodes to cluster feature embeddings obtained from self-supervised learning, to investigate whether meaningful clusters can be derived from these representation embeddings. Third, we benchmark multi-modality by performing contrastive learning on DNA barcodes, image data, and taxonomic information. This yields a general shared embedding space enabling taxonomic classification using multiple types of information and modalities. The code repository of the BIOSCAN-5M Insect dataset is available at https://github.com/bioscan-ml/BIOSCAN-5M.

Background: Despite the recent advances in computational fluid dynamics (CFD) techniques applied to blood flow within the left atrium (LA), the relationship between atrial geometry, flow patterns, and blood stasis within the left atrial appendage (LAA) remains unclear. A better understanding of this relationship would have important clinical implications, as thrombi originating in the LAA are a common cause of stroke in patients with atrial fibrillation (AF). Aim: To identify the most representative atrial flow patterns on a patient-specific basis and study their influence on LAA blood stasis by varying the flow split ratio and some common atrial modeling assumptions. Methods: Three recent techniques were applied to nine patient-specific computational fluid dynamics (CFD) models of patients with AF: a kinematic atrial model to isolate the influence of wall motion because of AF, projection on a universal LAA coordinate system, and quantification of stagnant blood volume (SBV). Results: We identified three different atrial flow patterns based on the position of the center of the main circulatory flow. The results also illustrate how atrial flow patterns are highly affected by the flow split ratio, increasing the SBV within the LAA. As the flow split ratio is determined by the patient's lying position, the results suggest that the most frequent position adopted while sleeping may have implications for the medium- and long-term risks of stroke., Comment: Accepted manuscript, 21 pages, 10 figures, 5 tables

Boolean networks can be viewed as functions on the set of binary strings of a given length, described via logical rules. They were introduced as dynamic models into biology, in particular as logical models of intracellular regulatory networks involving genes, proteins, and metabolites. Since genes can have several modes of action depending on their expression levels, binary variables are often not sufficiently rich, requiring the use of multivalued networks instead. In this paper, we explore the multivalued generalization of Boolean networks by writing the standard $(\wedge, \vee, \lnot)$ operations on $\{0, 1\}$ in terms of the operations $(\odot, \oplus, \neg)$ on $\big\{0,\frac{1}{m}, \frac{2}{m}, \dots, \frac{m-1}{m}, 1\big\}$ from multivalued logic. We recall the basic theory of this mathematical framework, and give a novel algorithm for computing the fixed points that in many cases has essentially the same complexity as in the binary case. Our approach provides a biologically intuitive representation of the network. Furthermore, it uses tools to compute lattice points in rational polytopes, tapping a rich area of algebraic combinatorics as a source for combinatorial algorithms for network analysis. An implementation of the algorithm is provided., Comment: 25 pages. Several improvements throughout the text, including a new title, missing references, and new Theorem 32

While the mechanical behavior of noble nanoporous metals has been the subject of numerous studies, less is known about their recently developed refractory-based counterparts. Here we report on the mechanical properties, deformation mechanisms and topological changes of nanoporous tantalum, a prototypical refractory metal, by means of atomistic simulations of compression tests. An open-source multi-cpu and gpu-capable software is presented and used for the generation of computational samples. The stress strain curves show a non-linear elastic response, with early yielding. The plastic regime is first characterized by a linear hardening followed by an exponential hardening at large strains, associated with a high degree of densification. Plasticity is dominated by dislocation activity, with twinning and vacancy formation appearing as complementary deformation mechanisms. In order to study the mechanical response from a topological perspective, we track the evolution of the genus throughout the tests, finding direct correlations with each regime of the stress strain curves. The results are in agreement with previous studies of plasticity in nanoporous metals and highlight the importance of using topological metrics, for gaining insights into complex aspects of the deformation of nanoporous metals.

This article focuses on the prediction of the vibration frequency response of handheld probes. A novel approach that involves machine learning and readily available data from probes was explored. Vibration probes are efficient and affordable devices that provide information about testing airborne sound insulation in building acoustics. However, fixing a probe to a vibrating surface downshifts sensor resonances and underestimates levels. Therefore, the calibration response of the sensor included in a probe differs from the frequency response of that same probe. Simulation techniques of complex mechanical systems may describe this issue, but they include hardly obtainable parameters, ultimately restricting the model. Thus, this study discusses an alternative method, which comprises different parts. Firstly, the vibration frequency responses of 85 probes were measured and labelled according to six features. Then, Linear Regression, Decision Tree Regression and Artificial Neural Networks algorithms were analysed. It was revealed that decision tree regression is the more appropriate technique for this data. The best decision tree models, in terms of scores and model structure, were fine-tuned. Eventually, the final suggested model employs only four out of the six original features. A trade-off solution that involved a simple structure, an interpretable model and accurate predictions was accomplished. It showed a maximum average deviation from test measurements ranging from 0.6 dB in low-frequency to 3 dB in high-frequency while remaining at a low computational load. This research developed an original and reliable prediction tool that provides the vibration frequency response of handheld probes.

In this paper we introduce two conceptual algorithms for minimising abstract convex functions. Both algorithms rely on solving a proximal-type subproblem with an abstract Bregman distance based proximal term. We prove their convergence when the set of abstract linear functions forms a linear space. This latter assumption can be relaxed to only require the set of abstract linear functions to be closed under the sum, which is a classical assumption in abstract convexity. We provide numerical examples on the minimisation of nonconvex functions with the presented algorithms., Comment: 14 pages, 6 figures

Motivated by the question of how biological systems maintain homeostasis in changing environments, Shinar and Feinberg introduced in 2010 the concept of absolute concentration robustness (ACR). A biochemical system exhibits ACR in some species if the steady-state value of that species does not depend on initial conditions. Thus, a system with ACR can maintain a constant level of one species even as the environment changes. Despite a great deal of interest in ACR in recent years, the following basic question remains open: How can we determine quickly whether a given biochemical system has ACR? Although various approaches to this problem have been proposed, we show that they are incomplete. Accordingly, we present new methods for deciding ACR, which harness computational algebra. We illustrate our results on several biochemical signaling networks., Comment: Proof in Section 5 edited, in response to reviewer comments

Triadic interactions are higher-order interactions that occur when a set of nodes affects the interaction between two other nodes. Examples of triadic interactions are present in the brain when glia modulate the synaptic signals among neuron pairs or when interneuron axon-axonic synapses enable presynaptic inhibition and facilitation, and in ecosystems when one or more species can affect the interaction among two other species. On random graphs, triadic percolation has been recently shown to turn percolation into a fully-fledged dynamical process in which the size of the giant component undergoes a route to chaos. However, in many real cases, triadic interactions are local and occur on spatially embedded networks. Here we show that triadic interactions in spatial networks induce a very complex spatio-temporal modulation of the giant component which gives rise to triadic percolation patterns with significantly different topology. We classify the observed patterns (stripes, octopus, and small clusters) with topological data analysis and we assess their information content (entropy and complexity). Moreover, we illustrate the multistability of the dynamics of the triadic percolation patterns and we provide a comprehensive phase diagram of the model. These results open new perspectives in percolation as they demonstrate that in presence of spatial triadic interactions, the giant component can acquire a time-varying topology. Hence, this work provides a theoretical framework that can be applied to model realistic scenarios in which the giant component is time-dependent as in neuroscience., Comment: 59 pages, 11 figures

Generative models are used as an alternative data augmentation technique to alleviate the data scarcity problem faced in the medical imaging field. Diffusion models have gathered special attention due to their innovative generation approach, the high quality of the generated images and their relatively less complex training process compared with Generative Adversarial Networks. Still, the implementation of such models in the medical domain remains at early stages. In this work, we propose exploring the use of diffusion models for the generation of high quality full-field digital mammograms using state-of-the-art conditional diffusion pipelines. Additionally, we propose using stable diffusion models for the inpainting of synthetic lesions on healthy mammograms. We introduce MAM-E, a pipeline of generative models for high quality mammography synthesis controlled by a text prompt and capable of generating synthetic lesions on specific regions of the breast. Finally, we provide quantitative and qualitative assessment of the generated images and easy-to-use graphical user interfaces for mammography synthesis., Comment: 10 pages + 2 pages of references, 6 figures

Understanding biodiversity is a global challenge, in which DNA barcodes - short snippets of DNA that cluster by species - play a pivotal role. In particular, invertebrates, a highly diverse and under-explored group, pose unique taxonomic complexities. We explore machine learning approaches, comparing supervised CNNs, fine-tuned foundation models, and a DNA barcode-specific masking strategy across datasets of varying complexity. While simpler datasets and tasks favor supervised CNNs or fine-tuned transformers, challenging species-level identification demands a paradigm shift towards self-supervised pretraining. We propose BarcodeBERT, the first self-supervised method for general biodiversity analysis, leveraging a 1.5 M invertebrate DNA barcode reference library. This work highlights how dataset specifics and coverage impact model selection, and underscores the role of self-supervised pretraining in achieving high-accuracy DNA barcode-based identification at the species and genus level. Indeed, without the fine-tuning step, BarcodeBERT pretrained on a large DNA barcode dataset outperforms DNABERT and DNABERT-2 on multiple downstream classification tasks. The code repository is available at https://github.com/Kari-Genomics-Lab/BarcodeBERT, Comment: Main text: 5 pages, Total: 9 pages, 2 figures, accepted at the 4th Workshop on Self-Supervised Learning: Theory and Practice (NeurIPS 2023)

The role of type Ia supernovae (SN Ia), mainly the Delay Time Distributions (DTDs) determined by the binary systems, and the yields of elements created by different explosion mechanisms, are studied by using the MulChem chemical evolution model applied to our Galaxy. We explored 15 DTDs and 12 tables of elemental yields produced by different SN Ia explosion mechanisms, doing a total of 180 models. Chemical abundances for $\alpha$-elements (O, Mg, Si, Ca) and Fe derived from these models are compared with recent solar region observational data of $\alpha$-elements over Fe relative abundances, [X/Fe], as a function of [Fe/H] and age. A multidimensional maximum likelihood analysis shows that 52 models are able to fit all these datasets simultaneously, considering the 1$\sigma$ confidence level. The combination of STROLG1 DTD and LN20181 SN Ia yields provides the best fit. The exponential model with very prompt events is a possible DTD, but a combination of several channels is more probable. The SN Ia yields that include MCh or Near MCh correspond to 39 (75%) of the 52 best models. Regarding the DTD, 31 (60%) of the 52 most probable models correspond to the SD scenario, while the remaining 21 (40%) are based on the DD scenario. Our results also show that the relatively large dispersion of the observational data may be explained by the stellar migration from other radial regions, and/or perhaps a combination of DTDs and explosion channels., Comment: 26 pages, 28 figures. Accepted for publication in MNRAS