Block Model Research Articles

Exploration and future trends on spatial correlation of green innovation efficiency in strategic emerging industries under the digital economy: A social network analysis

With digital technological change and the increasing frequency of interregional innovation links, the spatial correlation and diversity of strategic emerging industries' green innovation efficiency (SEI-GIE) need to be explored in depth. This paper innovatively constructs the SEI-GIE input-output index system under digital economy. The proposed grey model FINGBM(1,1) with ω-order accumulation and weighted initial value optimization realizes effective prediction of 7 input-output indicators of 30 provinces in China from 2021 to 2025. Super-SBM-DEA, gravity model, and social network analysis are applied to explore spatial network structure's dynamic process of SEI-GIE from 12th to 14th Five-Year-Plan period (2011–2025). Empirical results show that (1) Under the effect of digital economy, the SEI-GIE in China generally shows a U-shaped fluctuation trend, in which the growth trend in the central region is obvious, and the western region shows significant fluctuations. (2) The spatial correlation network of SEI-GIE presents a complex and stable center-periphery circle. Particularly, the overall increase in network efficiency highlights the strong small-world characteristics. (3) Beijing, Shanghai, Zhejiang and Jiangsu have always been in the leading core position, with strong influence and control; And Tianjin's core position in the network will decline. Additionally, Guangxi and Chongqing have great potential, but Guangdong needs to strengthen its radiation effect. (4) Block model shows that plate-I (Beijing, Tianjin) receive spatial spillovers from others, while plates-III,IV have significant spillover effects. This study provides theoretical reference for policymakers from a network perspective to promote development of China's SEI-GIE.

A generalized stochastic block model for overlapping community detection

Over the past two decades, community detection has been extensively explored. Yet, the problem of identifying overlapping communities has not been fully solved. In this paper, we introduce a novel approach, called the generalized stochastic block model, to address this issue by allowing nodes to belong to multiple communities. This approach extends the traditional representation of nodal community assignment from a single community label to a label vector, with each element indicating the membership of a node in a specific community. We develop a Markov chain Monte Carlo algorithm to tackle the model. Through numerical experiments conducted on synthetic and empirical networks, we demonstrate the efficacy of the proposed framework in accurately detecting overlapping communities.

The latent cognitive structures of social networks

Abstract When people are asked to recall their social networks, theoretical and empirical work tells us that they rely on shortcuts, or heuristics. Cognitive social structures (CSSs) are multilayer social networks where each layer corresponds to an individual’s perception of the network. With multiple perceptions of the same network, CSSs contain rich information about how these heuristics manifest, motivating the question, Can we identify people who share the same heuristics? In this work, we propose a method for identifying cognitive structure across multiple network perceptions, analogous to how community detection aims to identify social structure in a network. To simultaneously model the joint latent social and cognitive structure, we study CSSs as three-dimensional tensors, employing low-rank nonnegative Tucker decompositions (NNTuck) to approximate the CSS—a procedure closely related to estimating a multilayer stochastic block model (SBM) from such data. We propose the resulting latent cognitive space as an operationalization of the sociological theory of social cognition by identifying individuals who share relational schema. In addition to modeling cognitively independent, dependent, and redundant networks, we propose a specific model instance and related statistical test for testing when there is social-cognitive agreement in a network: when the social and cognitive structures are equivalent. We use our approach to analyze four different CSSs and give insights into the latent cognitive structures of those networks.

Construction of new hydrophobic δ-MnO2 flower-like/kaolin/(3-aminopropyl) triethoxysilane/metal mesh membrane for oil/water separation: modeling of fouling process

A composite material consisting of δ-MnO2 flower-like/kaolin was modified with (3-aminopropyl) triethoxysilane (APTES) and utilized for separating sunflower oil from water using a metal mesh membrane. The δ-MnO2 flower-like/kaolin/APTES/metal mesh membrane was characterized using various techniques such as X-ray diffraction analysis, Fourier-transform infrared spectroscopy, scanning electron microscopy, energy dispersive spectroscopy mapping, Brunauer–Emmett–Teller theory/Barrett–Joyner–Halenda, atomic force microscopy, and contact angle. Based on the outcomes of the conducted experiments, it was observed that the maximum porosity was associated with the δ-MnO2 flower-like/kaolin/APTES 15 (wt.%)/metal mesh (MKA(15 wt.%)M) membrane, and the porosity was calculated to be 0.58%. The blocking filtration model was employed, and a complete blocking model was obtained with a value of n = 2.15 for MKA(15 wt.%)M membrane. The response surface methodology based on Box–Behnken design was utilized to investigate the effect of parameters such as weight percentage of APTES (wt.%), mass of δ-MnO2 flower-like/kaolin/APTES (MKA) (g), volume of oil in water (mL), and temperature (°C) on the flux (L/m2·h) and rejection (%). Under optimal conditions, weight percentage of 13.32 (wt.%) of APTES, concentration of 0.2 (g) of MKA, volume of 37.10 (mL) of oil in water, and a temperature of 59.06 (°C) were obtained. These conditions resulted in a flux of 5977 (L/m2·h) and a rejection rate of 99.99% for 9 s. The kinetic studies indicated that the pseudo-second-order model had the highest correlation coefficient of 0.9991, thus displaying the most agreement with the experimental data.

Ensuring Student Mental Wellbeing whilst introducing Block Mode Intensive Learning and Teaching

As De Montfort University, a UK-based University, introduced block delivery across the undergraduate portfolio the University sought to maintain a continued focus on, and commitment to, an institution-wide approach to embedding mental wellbeing. With mental wellbeing relevant to the whole university community and recognising the power of transformational change to impact wellbeing, the university cross-examined student feedback to understand and ensure mental wellbeing. An intensive block model was introduced at De Montfort University (DMU) from the start of the academic session 2022-23. Undergraduate student experiences and learning from the initial months, focused on those in the Faculty of Computing, Engineering and Media, are presented alongside an examination of the methods and impact of embedding mental wellbeing in this new intensive mode; ensuring the continued institutional approach to embedding mental wellbeing in the curriculum, providing inclusive support for all learners. The learning and impact identified from initial experiences indicate areas of strength and areas for development, enhancement and growth. This article presents educational practice implications for other providers exploring and implementing block delivery in intensive learning and teaching modes. The University’s HealthyDMU philosophy recognises mental wellbeing is relevant to all, informed by a social model of wellbeing, in which a student’s experience of mental wellbeing is directly related to their environment and experiences, based on the five ways to wellbeing (Aked, Marks, Cordon & Thompson, 2008). Embedding this throughout the curriculum reduces wellbeing barriers to facilitate student success, establishing pro-active approaches to mental wellbeing and a health promoting environment. This article provides a unique focus on student experience from the position of mental wellbeing embedded in curriculum design and delivery. This paper presents an exploration of student responses to the introduction of intensive delivery at undergraduate level within a UK university, positioned around support for mental wellbeing.

Proof-of-concept model for instantaneous heart rate-drift correction during low and high exercise exertion.

Introduction: This study aimed to model below and above anaerobic threshold exercise-induced heart rate (HR) drift, so that the corrected HR could better represent kinetics during and after the exercise itself. Methods: Fifteen healthy subjects (age: 28 ± 5 years; : 50 ± 8 mL/kg/min; 5 females) underwent a maximal and a 30-min submaximal (80% of the anaerobic threshold) running exercises. A five-stage computational (i.e., delay block, new training impulse-calculation block, Sigmoid correction block, increase block, and decrease block) model was built to account for instantaneous HR, fitness, and age and to onset, increase, and decrease according to the exercise intensity and duration. Results: The area under the curve (AUC) of the hysteresis function, which described the differences in the maximal and submaximal exercise-induced and HR kinetics, was significantly reduced for both maximal (26%) and submaximal (77%) exercises and consequent recoveries. Discussion: In conclusion, this model allowed HR drift instantaneous correction, which could be exploited in the future for more accurate estimations.

Analytic representative element rate decline models for naturally fractured reservoir depletion

Representative single anisotropic matrix block 2D Green’s function models for depletion through fully-penetrating, vertical fractures through different numbers of fracture faces are constructed that analytically capture both fracture and block depletion with fracture-matrix mass transfer. The 1D Green’s function for a fracture system is likewise solved in terms of the time evolution of average fracture pressure. While transient average pressure values are not inherently measurable, they are transformed into cumulative production or instantaneous flowrate values, thus producing new rate decline model functional forms. Primary variables in assembling the interacting systems model are the volume ratio, Vf /Vm, permeability ratio, kf /kx, and geometry, (a/b)(ky/kx), with the last term accounting for both block shape and permeability anisotropy. We construct interacting systems models in terms of various ratios of Vf /Vm, and kf /kx for three fracture architecture prototypes: representative matrix blocks depleted by 4, 2, or 1 contacting fractures. The single matrix block models can be migrated to ones for heterogeneous systems using superposition and matrix block distributions, as demonstrated with a binary distribution of block sizes with variable fractions. Analytic solutions for rate decline problems can be used to understand the production signatures of naturally fractured reservoirs and interpretation of fracture volume fraction, permeability ratio, average matrix block size, and measures of heterogeneity.

A damage constitutive model of layered slate under the action of triaxial compression and water environment erosion

Based on the macroscopic structure control theory, The slate with a significant bedding plane is a composite rock mass composed of rock blocks containing microscopic defects, joint surface closure elements, and shear deformation elements. Considering the coupling damage effect of water erosion and triaxial compressive load on bedding structure plane, the transversely isotropic damage constitutive model of slate under triaxial compressive load is derived with the dip angle of bedding and confining pressure as the variable. Firstly, based on the statistical theory of continuous damage mechanics and the maximum tensile strain criterion, the transversely isotropic deformation constitutive model of rock block with micro-defects is given; Secondly, based on the phenomenological theory of closed deformation and shear-slip deformation mechanism of layered structural plane under the coupling action of water erosion and triaxial compression load, the calculation formula of axial deformation of layered structural plane under the coupling action is given; Finally, to verify the accuracy of the established constitutive model, triaxial compression tests are carried out to study the influence of dip angle and confining pressure on the macroscopic mechanical properties and mechanism of slate. The results show that: the established triaxial compression damage constitutive model of bedding slate can accurately describe the stress–strain relationship of bedding slate after water environment erosion. With the increase of bedding dip angle, the strength and deformation capacity of the bedding slate first decreases and then increases, showing a U-shaped distribution as a whole. There are three main types of failure: tension shear composite failure, shear slip failure, and splitting tension failure.

Modelling surface subsidence of coal mines using a bonded block numerical method

Surface subsidence is common geologic hazard in coal mines when longwall mining method is applied. In this research, a Bonded Block Model (BBM) is selected to numerically simulate and analyze the surface subsidence of coal mines. Parametric study is conducted and key factors controlling the subsidence are characterized. BBM models with no bedding planes realistically simulate the maximum subsidence, but overestimate the subsidence levels above the panel edges. The bedding planes introduced in the BBM models potentially provide more realistic subsidence profile. The separation of the planes and downward bending of the layered strata, as commonly observed in coal mines, are simulated and captured. In addition, high horizontal stress plays a key role in simulating more realistic subsidence profiles. The methodologies and conclusions of this research are expected to improve our understandings of the surface subsidence of coal mines and provide reliable tools to analyze and characterize this challenging issue.

A value analysis of longitudinal integrated clerkships: consideration of costs and benefits

ABSTRACT Longitudinal Integrated Clerkships (LICs) prioritise longitudinal relationships with faculty, patients, and place. Research shows that LICs benefit students and faculty, but most medical schools have limited LIC programmes. This is likely due to perceptions that LICs are more costly and complex than traditional block rotations (TBRs). The perceived cost versus evidence-based value related to clerkship education has not been examined in detail. Until recently, no ‘All-LIC’ medical school exemplars existed in the US, limiting the value of this model as well as the ability to examine relative cost and complexity. In this paper, we draw on our experience launching three ‘All-LIC’ medical schools in the United States – schools in which the entire clerkship class participates in a comprehensive clerkship-year LIC. We propose that the known benefits of LICs coupled with cost-mitigation strategies related to running an ‘All LIC’ model for core clinical clerkships, rather than block and LIC models simultaneously, results in a higher value for medical schools.

Analysis of equivalent substitution of rectangular stress block for nonlinear stress diagram

This article deals with the analysis of equivalency of the substitution of rectangular stress block for nonlinear stress diagram with descending branch for the calculation of flexural, eccentrically compressed and eccentrically tensioned reinforced concrete members of rectangular cross-section. The method for equivalent substitution of these diagrams is proposed. Analytical relationships of area and its centre for the nonlinear diagram with descending branch are presented. Coefficients for equivalent substitution of the said diagrams for the standard concrete strength classes are determined and given. It gives the opportunity to design reinforced concrete members using rectangular stress block model. Substitution of the diagrams applied in EC2 and in STR 2.05.05:2005 is compared with the equivalent substitution of rectangular stress diagram for nonlinear stress diagram with descending branch. It is shown that in STR 2.05.05:2005 description of the ratio between the depth of the rectangular diagram and that of the equivalent nonlinear one with descending branch is very poor. An explicit analytical relationship for this ratio and its linear approximation in respect to the concrete characteristic and design strengths are presented in this article as well.

Advancing 3D Dental Implant Finite Element Analysis: Incorporating Biomimetic Trabecular Bone with Varied Pore Sizes in Voronoi Lattices.

The human mandible's cancellous bone, which is characterized by its unique porosity and directional sensitivity to external forces, is crucial for sustaining biting stress. Traditional computer- aided design (CAD) models fail to fully represent the bone's anisotropic structure and thus depend on simple isotropic assumptions. For our research, we use the latest versions of nTOP 4.17.3 and Creo Parametric 8.0 software to make biomimetic Voronoi lattice models that accurately reflect the complex geometry and mechanical properties of trabecular bone. The porosity of human cancellous bone is accurately modeled in this work using biomimetic Voronoi lattice models. The porosities range from 70% to 95%, which can be achieved by changing the pore sizes to 1.0 mm, 1.5 mm, 2.0 mm, and 2.5 mm. Finite element analysis (FEA) was used to examine the displacements, stresses, and strains acting on dental implants with a buttress thread, abutment, retaining screw, and biting load surface. The results show that the Voronoi model accurately depicts the complex anatomy of the trabecular bone in the human jaw, compared to standard solid block models. The ideal pore size for biomimetic Voronoi lattice trabecular bone models is 2 mm, taking in to account both the von Mises stress distribution over the dental implant, screw retention, cortical bone, cancellous bone, and micromotions. This pore size displayed balanced performance by successfully matching natural bone's mechanical characteristics. Advanced FEA improves the biomechanical understanding of how bones and implants interact by creating more accurate models of biological problems and dynamic loading situations. This makes biomechanical engineering better.

An Updated Fault Coupling Model Along Major Block‐Bounding Faults on the Eastern and Northeastern Tibetan Plateau From a Stress‐Constrained Inversion of GPS and InSAR Data

AbstractLarge block‐bounding faults on the Tibetan plateau are significant geological structures that accommodate tectonic movements and accumulate stress, leading to large earthquakes. Quantifying the interseismic slip deficit rate helps to better assess the earthquake potential. We combine available InSAR (2015–2020) and interseismic GPS data to determine fault coupling along 14 major block‐bounding faults. Spatially dense InSAR measurements remarkably improve the resolution of the coupling model. Combined with a GPS‐constrained block model, we examine the performance of the inversion approach with the stress constraint and the common Laplacian smoothing based on both synthetic tests and real data. We suggest that, for continental strike‐slip faults, adding the stress constraint can mitigate unphysical coupling distributions due to unreasonable assumptions or modeling artifacts, reducing the model uncertainty and improving the accuracy of the coupling model. This is particularly useful for segments featured by a highly heterogeneous distribution of coupling along the transition zone from locking to creeping region, partially‐coupling segment, and junction zone between main and subsidiary faults. We present a large‐scale fault coupling map along the major block‐bounding faults on the northeastern and eastern Tibetan plateau, highlighting the distinct degrees of fault coupling and lateral variations. The collage of coupling maps along different faults demonstrates the kinematic features over a broad time scale during earthquake cycles ranging from early to late interseismic phases, such as the segments ruptured during the 2001 Kokoxili earthquake and the 1920 Haiyuan earthquake.

EV Charging Scheduling Under Demand Charge: A Block Model Predictive Control Approach

This paper studies the online scheduling of electric vehicle charging by a service provider subject to a demand charge in a distribution system. Demand charge imposes a penalty on the peak power consumption over each billing period, representing a substantial cost for the service provider with a large number of clients. Because the demand charge is calculated at the end of the billing period, it poses challenges in real-time scheduling when energy demand forecasts are inaccurate, resulting in either overly conservative power consumption or substantial demand charge. We propose a block model predictive control approach that decomposes the demand charge into a sequence of stage costs. Optimality conditions on demand patterns are also presented and analyzed. Numerical simulations demonstrate the efficacy of the proposed approach. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —This paper addresses a significant practical problem of minimizing the demand charge on the real-time scheduling of deferrable demands. In particular, we consider a setting where a commercial electric vehicle (EV) charging service provider has to manage the online scheduling of a large number of arriving EVs at a charging facility subject to a maximum charging power constraint and a tariff with the demand charge. A major practical challenge is to balance the tradeoff between maximizing profit in scheduling as much EV charging as possible and the need to minimize penalty on the peak charging power. We propose a model predictive control strategy that decomposes the overall demand charge into a sequence of terminal costs. Also addressed is the practical constraint arising from the mismatched EV charging decision period and the power measurement period used to compute the demand charge. Using real data collected at the Adaptive Charging Network (ACN) testbed in simulations, the proposed approach yields 8-12% improvement in operational profit over existing benchmarks, while it has yet been tested in actual charging systems. In the future research, we will address the charging scheduling under demand charge over multiple charging stations.

Breakdown of a concavity property of mutual information for non-Gaussian channels

Abstract Let $S$ and $\tilde S$ be two independent and identically distributed random variables, which we interpret as the signal, and let $P_{1}$ and $P_{2}$ be two communication channels. We can choose between two measurement scenarios: either we observe $S$ through $P_{1}$ and $P_{2}$, and also $\tilde S$ through $P_{1}$ and $P_{2}$; or we observe $S$ twice through $P_{1}$, and $\tilde{S}$ twice through $P_{2}$. In which of these two scenarios do we obtain the most information on the signal $(S, \tilde S)$? While the first scenario always yields more information when $P_{1}$ and $P_{2}$ are additive Gaussian channels, we give examples showing that this property does not extend to arbitrary channels. As a consequence of this result, we show that the continuous-time mutual information arising in the setting of community detection on sparse stochastic block models is not concave, even in the limit of large system size. This stands in contrast to the case of models with diverging average degree, and brings additional challenges to the analysis of the asymptotic behavior of this quantity.

Ambulatory Long Block: A Model of Precision Education and Assessment for Internal Medicine Residents.

High-quality precision education (PE) aims to enhance outcomes for learners and society by incorporating longitudinal data and analytics to shape personalized learning strategies. However, existing educational data collection methods often suffer from fragmentation, leading to gaps in understanding learner and program performance. In this article, the authors present a novel approach to PE at the University of Cincinnati, focusing on the Ambulatory Long Block, a year-long continuous ambulatory group-practice experience. Over the last 17 years, the Ambulatory Long Block has evolved into a sophisticated data collection and analysis system that integrates feedback from various stakeholders, as well as learner self-assessment, electronic health record utilization information, and clinical throughput metrics. The authors detail their approach to data prioritization, collection, analysis, visualization, and feedback, providing a practical example of PE in action. This model has been associated with improvements in both learner performance and patient care outcomes. The authors also highlight the potential for real-time data review through automation and emphasize the importance of collaboration in advancing PE. Generalizable principles include designing learning environments with continuity as a central feature, gathering both quantitative and qualitative performance data from interprofessional assessors, using this information to supplement traditional workplace-based assessments, and pairing it with self-assessments. The authors advocate for criterion referencing over normative comparisons, using user-friendly data visualizations, and employing tailored coaching strategies for individual learners. The Ambulatory Long Block model underscores the potential of PE to drive improvements in medical education and health care outcomes.

Experimental Study of a Sequential Membrane Process of Ultrafiltration and Nanofiltration for Efficient Polyphenol Extraction from Wine Lees.

The wine industry is a sector of great importance in the Spanish economy, contributing substantial annual revenues. However, one challenge facing the industry is the amount of waste generated, reaching millions of tons annually. These residues consist of organic matter of industrial interest, such as polyphenols. These substances are characterised by their excellent antioxidant properties, making them ideal for use in the food, cosmetic, and pharmaceutical industries. Modern techniques, such as membrane technology, are explored for their extraction based on separating compounds according to size. This work studies a sequential filtration process using ultrafiltration (UF) and nanofiltration (NF) membranes at different operating conditions (2 bar and 9.5 bar for UF and NF, respectively, at 20 °C) to extract polyphenols from wine lees. The results show a total polyphenols rejection rate for each process of 54% for UF and 90% for NF. Pore blocking models have been studied for the UF process and an intermediate pore blocking of the membrane upon wine lees filtration has been identified. A mathematical model that justifies the behavior of a polymeric NF membrane with the filtration of pre-treated vinasse residues has been validated. This study shows a viable process for extracting polyphenols from wine lees with sequential membrane technology.

Community detection in the human connectome: Method types, differences and their impact on inference.

Community structure is a fundamental topological characteristic of optimally organized brain networks. Currently, there is no clear standard or systematic approach for selecting the most appropriate community detection method. Furthermore, the impact of method choice on the accuracy and robustness of estimated communities (and network modularity), as well as method-dependent relationships between network communities and cognitive and other individual measures, are not well understood. This study analyzed large datasets of real brain networks (estimated from resting-state fMRI from = 5251 pre/early adolescents in the adolescent brain cognitive development [ABCD] study), and = 5338 synthetic networks with heterogeneous, data-inspired topologies, with the goal to investigate and compare three classes of community detection methods: (i) modularity maximization-based (Newman and Louvain), (ii) probabilistic (Bayesian inference within the framework of stochastic block modeling (SBM)), and (iii) geometric (based on graph Ricci flow). Extensive comparisons between methods and their individual accuracy (relative to the ground truth in synthetic networks), and reliability (when applied to multiple fMRI runs from the same brains) suggest that the underlying brain network topology plays a critical role in the accuracy, reliability and agreement of community detection methods. Consistent method (dis)similarities, and their correlations with topological properties, were estimated across fMRI runs. Based on synthetic graphs, most methods performed similarly and had comparable high accuracy only in some topological regimes, specifically those corresponding to developed connectomes with at least quasi-optimal community organization. In contrast, in densely and/or weakly connected networks with difficult to detect communities, the methods yielded highly dissimilar results, with Bayesian inference within SBM having significantly higher accuracy compared to all others. Associations between method-specific modularity and demographic, anthropometric, physiological and cognitive parameters showed mostly method invariance but some method dependence as well. Although method sensitivity to different levels of community structure may in part explain method-dependent associations between modularity estimates and parameters of interest, method dependence also highlights potential issues of reliability and reproducibility. These findings suggest that a probabilistic approach, such as Bayesian inference in the framework of SBM, may provide consistently reliable estimates of community structure across network topologies. In addition, to maximize robustness of biological inferences, identified network communities and their cognitive, behavioral and other correlates should be confirmed with multiple reliable detection methods.

3D printable model of an entomological pinning block, designed for precise positioning of entomological glue boards and labels.

The entomological pinning block is a device widely used by entomologists to facilitate the mounting of glue boards and labels on entomological pins. The most commonly used entomological blocks are wooden blocks with steps of varying heights, allowing the placement of individual glue boards and labels at different levels. Models of entomological blocks ready for 3D printing are scarce on the Internet. The proposed model of an entomological block is ready for printing on a standard 3D printer. In addition to the usual positioning of glue boards and labels along the Z-axis, the model also offers targeting devices that enable precise positioning of the entomological pin along the X- and Y-axes. The proposed model can be easily modified for use with glue boards or labels of a different width than proposed or with label level steps of varying heights along the pin. The design of the new pinning block is simple, yet effective, making it an accessible and useful tool for entomologists, museum curators and amateur collectors.

A Deep Dynamic Latent Block Model for Co-clustering of Zero-Inflated Data Matrices

The simultaneous clustering of observations and features of datasets (known as co-clustering) has recently emerged as a central machine learning application to summarize massive datasets. However, most existing models focus on continuous and dense data in stationary scenarios, where cluster assignments do not evolve over time. This work introduces a novel latent block model for the dynamic co-clustering of data matrices with high sparsity. To properly model this type of data, we assume that the observations follow a time and block dependent mixture of zero-inflated distributions, thus, combining stochastic processes with the time-varying sparsity modeling. To detect abrupt changes in the dynamics of both cluster memberships and data sparsity, the mixing and sparsity proportions are modeled through systems of ordinary differential equations. The inference relies on an original variational procedure whose maximization step trains fully connected neural networks in order to solve the dynamical systems. Numerical experiments on simulated and real world datasets demonstrate the effectiveness of the proposed methodology in the context of count data.