iMinds

Cerebral arteriovenous malformations (cAVMs) are vascular abnormalities associated with a significant risk of rupture and hemorrhage. Endovascular embolization is a potentially curative treatment for a large set of malformations, but its success and safety depend on cAVM hemodynamics. However, hemodynamics is unpredictable in many cases. Therefore, computational models have become a valuable tool for investigating cAVM hemodynamics, offering insights that are otherwise difficult or impossible to obtain before or during interventions. Moreover, these models could be used to optimize the endovascular procedure pre-operatively. This review provides an overview of the two principal computational approaches used to simulate cAVM hemodynamics and embolization, i.e., lumped parameter models and computational fluid dynamics (CFD). Lumped parameter models simplify the complex vasculature by representing it as an electrical analog circuit. These models have evolved from simple, idealized geometries to sophisticated networks informed by patient-specific data, including the implementation of autoregulation mechanisms. In contrast, CFD models offer spatially resolved simulations of fluid flow by solving governing equations on discretized domains. To model the anatomical complexity of the cAVM nidus, several CFD studies employ porous media implementations, approximating the nidus' vascular network as a sponge-like material. Furthermore, the use of multiphase flow allows for the explicit simulation of embolic agent injections, their solidification process, and their occlusive effect. Both modeling approaches have already shown added value regarding validating clinical observations, estimating rupture risk, and predicting embolization effects. Toward the future, we envision that these models could lead to more effective, personalized treatment for cAVM patients.

We live in a rich and varied acoustic world, which is experienced by individuals or communities as a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">soundscape</i>. Computational auditory scene analysis, disentangling acoustic scenes by detecting and classifying events, focuses on objective attributes of sounds, such as their category and temporal characteristics, ignoring their effects on people, such as the emotions they evoke within a context. To fill this gap, we propose the affective soundscape captioning (ASSC) task, which enables automated soundscape analysis, thus avoiding labour-intensive subjective ratings and surveys in conventional methods. With soundscape captioning, context-aware descriptions are generated for soundscape by capturing the acoustic scenes (ASs), audio events (AEs) information, and the corresponding human affective qualities (AQs). To this end, we propose an automatic soundscape captioner (SoundSCaper) system composed of an acoustic model, i.e. SoundAQnet, and a large language model (LLM). SoundAQnet simultaneously models multi-scale information about ASs, AEs, and perceived AQs, while the LLM describes the soundscape with captions by parsing the information captured with SoundAQnet. SoundSCaper is assessed by two juries of 32 people. In expert evaluation, the average score of SoundSCaper-generated captions is slightly lower than that of two soundscape experts on the evaluation set D1 and the external mixed dataset D2, but not statistically significant. In layperson evaluation, SoundSCaper outperforms soundscape experts in several metrics on datasets D1 and D2. In addition to human evaluation, compared to other automated audio captioning (AAC) systems with and without LLM, SoundSCaper performs better on the ASSC task in several natural language processing (NLP) based metrics. Overall, SoundSCaper performs well in human subjective evaluation and various objective captioning metrics, and the generated captions are comparable to those annotated by soundscape experts. The model, source code, LLM scripts, human assessment data, instructions, and evaluation statistics are all publicly available.

Reservoir Computing (RC) is a lightweight way to model time-dependent data, yet its reliance on static, randomly initialized network architectures often limits performance on challenging real-world problems. We introduce Hebbian Architecture Generation (HAG), an unsupervised rule that grows connections between neurons that frequently activate together-embodying the biological maxim "neurons that fire together wire together." Starting from an almost empty reservoir, HAG progressively sculpts a task-specific wiring. Across a diverse set of classification and forecasting tasks, reservoirs reshaped by HAG are consistently more accurate than traditional Echo State Networks and reservoirs tuned with popular plasticity rules such as Intrinsic Plasticity or Anti-Oja learning. In other words, letting the network rewire itself from data turns a once-static RC model into a flexible, high-performance learner without a single gradient step. By coupling the efficiency of RC with the adaptability of Hebbian plasticity, HAG moves reservoir computing closer to its biological inspiration and shows that structural self-organization is a practical route to robust, task-aware processing of real-world time-series data.

Modeling the mechanical behavior of human tissues, particularly tumor tissues, poses significant challenges due to the difficulty in acquiring samples. In this study, we performed a total of ten measurements on five freshly excised peritoneal metastasis samples, alongside ten measurements from two healthy colon samples, to develop mechanical models using the standard linear solid (SLS) model and its generalized forms. The peritoneal metastasis samples included colon cancer (2 samples), ovarian cancer (1 sample), and rectal cancer (2 samples). An ex vivo static indentation test was conducted to assess the stress relaxation behavior of both tumor and healthy tissues using a step indentation protocol. A novel cross-validation approach was employed for model selection, based on mean square error (MSE) values. Due to the irregularity and complexity of tumor tissues, 80% of the tumor measurements required more complex models with additional parameters compared to the healthy colon tissues. The five-element double Maxwell-Wiechert (DMW) arm model was suitable for describing the mechanical behavior of all healthy colon tissue measurements. In contrast, the seven-element triple Maxwell-Wiechert (TMW) arm model best described 80% of the tumor tissue measurements, while the DMW model was adequate for the remaining 20%. Further histopathological analysis of the tissue samples may help elucidate the relationship between biological composition and mechanical properties.

Drawing parallels between Deep Artificial Neural Networks (DNNs) and biological systems can aid in understanding complex biological mechanisms that are difficult to disentangle. Temporal processing, an extensively researched topic, is one such example that lacks a coherent understanding of its underlying mechanisms. In this study, we investigate temporal processing in a Deep Reinforcement Learning (DRL) agent performing an interval timing task and explore potential biological counterparts to its emergent behavior. The agent was successfully trained to perform a duration production task, which involved marking successive occurrences of a target interval while viewing a video sequence. Analysis of the agent’s internal states revealed oscillatory neural activations, a ubiquitous pattern in biological systems. Interestingly, the agent’s actions were predominantly influenced by neurons exhibiting these oscillations with high amplitudes and frequencies corresponding to the target interval. Parallels are drawn between the agent’s time-keeping strategy and the Striatal Beat Frequency (SBF) model, a biologically plausible model of interval timing. Furthermore, the agent maintained its oscillatory representations and task performance when tested on different video sequences (including a blank video). Thus, once learned, the agent internalized its timekeeping mechanism and showed minimal reliance on its environment to perform the timing task. A hypothesis about the resemblance between this emergent behavior and certain aspects of the evolution of biological processes like circadian rhythms, has been discussed. This study aims to contribute to recent research efforts of utilizing DNNs to understand biological systems, with a particular emphasis on temporal processing.

ABSTRACTWe present a novel approach to help decision‐makers efficiently identify preferred solutions from the Pareto set of a multi‐objective optimization problem. Our method uses a Bayesian model to estimate the decision‐maker's utility function based on pairwise comparisons. Aided by this model, a principled elicitation strategy selects queries interactively to balance exploration and exploitation, guiding the discovery of high‐utility solutions. The approach is flexible: it can be used interactively or a posteriori after estimating the Pareto front through standard multi‐objective optimization techniques. Additionally, at the end of the elicitation phase, it generates a reduced menu of high‐quality solutions, simplifying the decision‐making process. Through experiments on test problems with up to nine objectives, our method demonstrates superior performance in finding high‐utility solutions with a small number of queries. We also provide an open‐source implementation of our method to support its adoption by the broader community.

The visual appeal of flowers in retail and aesthetic settings is significantly influenced by lighting. Although previous research has examined how different light sources impact flower colour perception, the effect of specific colour rendering metrics remains underexplored. This study conducted a visual experiment to identify the optimal spectrum for presenting natural flowers more attractively. Ten illuminants with a correlated colour temperature around 3500 K were selected using k -means clustering and generated using a multi-channel tuneable light source. Between 6 and 14 observers participated in a paired comparison visual experiment. Results suggest a positive correlation between the ANSI/IES TM 30-24 (TM-30) local chroma shift ( R csh, j ) and the visual attractiveness of red and green flowers. No significant correlation was observed with the TM-30 colour fidelity index ( R f ), gamut area index ( R g ), the CIE colour rendering index (CRI R a ) or memory colour rendering index. The most attractive spectra for red flowers exhibited high R csh,1 , whereas the most attractive spectra for green plants exhibited high R csh,6 . A spectrum with high values of both R csh,1 and R csh,6 was identified as optimal for enhancing the visual attractiveness of red flowers and green plants. However, no optimal spectrum was found for yellow flowers.

High-performance computing (HPC) clusters are essential for training large-scale AI models, yet they often suffer from severe underutilization due to network bottlenecks. This paper investigates the critical role of job placement in multi-tenant AI clusters and its impact on network performance. We propose a flow-level system model that jointly considers job placement, network topology, and routing strategy to evaluate link loads and congestion. By analyzing optimal, random and state-of-the-art placement strategies across modern network topologies, we demonstrate that placement decisions significantly influence network efficiency. Our results show that job placement cannot be ignored even under optimal routing, and that existing placement strategies are dependent on the routing strategy. This work underscores the importance of prioritizing job placement, as suboptimal placements can lead to significant performance degradation in AI workloads on HPC infrastructure.