Dense Clusters Research Articles

Abstract P4-09-02: A molecular classification system for basal-like breast cancer based on the tumor microenvironment is prognostic for survival

Abstract Recent advancements in molecular profiling have revealed distinct breast cancer subtypes, but many clinical NGS assays rely on gene panels, such as PAM50, limiting their clinical utility. Basal-like breast cancer (BLBC), one of the most aggressive subtypes, has highly variable molecular and clinical characteristics. Now, the tumor microenvironment (TME) is recognized as a vital participant in tumor progression and therapeutic response. The development of more refined classifications based on the TME, capable of accounting for tissue heterogeneity, may improve NGS clinical utility for BLBC. Here, we apply our transcriptomic-based approach, recently described by Bagaev et. al., to classify the BLBC TME into discrete immune portraits, to potentially improve clinical outcomes and facilitate therapeutic decisions. We collected a cohort of 1,708 BLBC samples based on the expression levels of 50 genes (PAM50) from 10 publicly available datasets, with clinical outcomes available (n = 819). Using methodology described by Bagaev et. al., 31 functional gene expression signatures (Fges) were selected, and unsupervised dense Louvain clustering was performed to identify TME subtypes. A novel RNA-seq deconvolution algorithm was used to determine the cell types within the TME. Validation of the histological features, including stroma, tumor infiltrating lymphocytes (TILs), and tertiary lymphoid structures (TLS), relative to gene expression patterns in the TME subtypes was performed by automated and manual annotation of BLBC H&E slides (n = 146) from an independent TCGA cohort. Overall survival (OS) analysis was performed using Kaplan-Meier and Cox regression methods. We revealed 5 BLBC subtypes with distinct expression patterns: immune-enriched, non-fibrotic (IE, 19%), B-cell–enriched, TLS-like (TLS, 25.5%), granulocyte-enriched (G, 12.8%), fibrotic (F, 28%), and immune desert (D, 17.7%) (Table). IE tumors featured an active immune TME, with high immune checkpoint expression and T cell activity. The TLS subtype also had an immune-rich TME, presenting high levels of B cells, T helper cells, and TLS (p < 0.001). The TLS subtype exhibited the highest number of stromal TILs on TCGA H&E slides. The G subtype was characterized by high expression of granulocytes and granulocyte traffic molecules. In accordance with our findings, deconvolution predicted the highest percent of neutrophils in the G subtype (p < 0.001). The F subtype demonstrated the highest levels of angiogenesis, stromal Fges, and VEGFR1-3, FGFR1, and EGFR expression. By histological evaluation, 84% of F subtype samples demonstrated a medium or high level of fibrosis. The D subtype showed a high proliferation rate and low stromal and immune Fges. Indicative of proliferation rate, CCNB1 and cyclin B1 were highest in G, D, and IE subtypes. OS analysis revealed a significant association between TME subtypes (TLS = baseline; log HR G = 0.87, p < 0.05; IE = 0.39, p = 0.18; F = 0.99, p < 0.05; D = 1.21, p < 0.05), and survival outcomes. The immune-enriched subtypes, IE and TLS, demonstrated good prognosis and higher expression of immune checkpoint genes, while immune desert D and granulocyte-enriched G subtypes exhibited the worst OS. Using our transcriptomic-based approach, BLBC was classified into 5 distinct subtypes, each with unique therapeutic vulnerabilities. Further investigation of these TME subtypes may lead to potential clinical utility as a prognostic tool to improve clinical decision making. Table. Characteristics of Basal-like breast cancer (BLBC) tumor microenvironment (TME) subtypes. TLS - Tertiary lymphoid structures; CAF - cancer-associated fibroblasts Citation Format: Svetlana Khorkova, Diana Shamsutdinova, Vladimir Kushnarev, Lev Popyvanov, Daniil Dymov, Anastasia Zotova, Ivan Valiev, Zoya Antysheva, Anna Love, Jessica H. Brown, Alexander Bagaev, Nikita Kotlov, Nathan Fowler. A molecular classification system for basal-like breast cancer based on the tumor microenvironment is prognostic for survival [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P4-09-02.

Low-frequency Waves Due to Newborn Interstellar Pickup Ions Observed from 43 to 47 au by the Voyager 1 Spacecraft

Interstellar neutral atoms enter the heliosphere at a relatively slow speed corresponding to the motion of the Sun through the local interstellar medium, which is approximately 25 km s−1. Neutral hydrogen atoms enter from the approximate location of the Voyager spacecraft and are eventually ionized primarily by collision with thermal solar wind ions. An earlier analysis by Hollick et al. examined low-frequency magnetic waves observed by the Voyager spacecraft from launch through 1990 that are thought to arise from the scattering of newborn interstellar pickup H+ and He+. We report an analysis of Voyager 1 observations in 1991, which is the last year of high-resolution magnetic field data that are publicly available, and find 70 examples of low-frequency waves with the characteristics that suggest excitation by pickup H+ and 10 examples of waves consistent with excitation by pickup He+. We find a particularly dense cluster of observations at the tail end of what is thought to be a Merged Interaction Region (MIR) that was previously studied by Burlaga & Ness using Voyager 2 observations. This is not unexpected if the MIR is followed by a large rarefaction region, as they tend to be regions of reduced turbulence levels that permit the growth of the waves over the long time periods that are generally required of this instability.

Great balls of FIRE II: The evolution and destruction of star clusters across cosmic time in a Milky Way-mass galaxy

ABSTRACT The current generation of galaxy simulations can resolve individual giant molecular clouds, the progenitors of dense star clusters. But the evolutionary fate of these young massive clusters, and whether they can become the old globular clusters (GCs) observed in many galaxies, is determined by a complex interplay of internal dynamical processes and external galactic effects. We present the first star-by-star N-body models of massive (N ∼ 105–107) star clusters formed in a FIRE-2 MHD simulation of a Milky Way-mass galaxy, with the relevant initial conditions and tidal forces extracted from the cosmological simulation. We select 895 (∼30 per cent) of the YMCs with >6 × 104 M⊙ from Grudić et al. 2022 and integrate them to z = 0 using the cluster Monte Carlo code, CMC. This procedure predicts a MW-like system with 148 GCs, predominantly formed during the early, bursty mode of star formation. Our GCs are younger, less massive, and more core-collapsed than clusters in the Milky Way or M31. This results from the assembly history and age-metallicity relationship of the host galaxy: Younger clusters are preferentially born in stronger tidal fields and initially retain fewer stellar-mass black holes, causing them to lose mass faster and reach core collapse sooner than older GCs. Our results suggest that the masses and core/half-light radii of GCs are shaped not only by internal dynamical processes, but also by the specific evolutionary history of their host galaxies. These results emphasize that N-body studies with realistic stellar physics are crucial to understanding the evolution and present-day properties of GC systems.

Differential roles of CaMKII isoforms in phase separation with NMDA receptors and in synaptic plasticity

Calcium calmodulin-dependent kinase II (CaMKII) is critical for synaptic transmission and plasticity. Two major isoforms of CaMKII, CaMKIIα and CaMKIIβ, play distinct roles in synaptic transmission and long-term potentiation (LTP) with unknown mechanisms. Here, we show that the length of the unstructured linker between the kinase domain and the oligomerizing hub determines the ability of CaMKII to rescue the basal synaptic transmission and LTP defects caused by removal of both CaMKIIα and CaMKIIβ (double knockout [DKO]). Remarkably, although CaMKIIβ binds to GluN2B with a comparable affinity as CaMKIIα does, only CaMKIIα with the short linker forms robust dense clusters with GluN2B via phase separation. Lengthening the linker of CaMKIIα with unstructured "Gly-Gly-Ser" repeats impairs its phase separation with GluN2B, and the mutant enzyme cannot rescue the basal synaptic transmission and LTP defects of DKO mice. Our results suggest that the phase separation capacity of CaMKII with GluN2B is critical for its cellular functions in the brain.

Machine Learning Approach for Detection of Water Overgrowth in Azov Sea with Sentinel-2 Data

The Azov Sea estuaries play an important role in the reproduction of semi-anadromous fish species. Spawning efficiency is closely connected with overgrowing of those species spawning grounds; thus, the objective of the water vegetation research has vital fisheries importance. Thus, the main goal of the research was to develop a machine learning algorithm for the detection of water overgrowth with Phragmites australis based on Sentinel-2 data. The research was conducted based on field botanical and vegetation investigations in 2020–2021 in Soleniy and Chumyanniy firths. Collected field and remote sensing data were processed with the semi-automatic classification plugin for QGIS. For the classification of Azov Sea estuaries, a random forest algorithm was used. The obtained results showed that in 2020 the areas occupied by reeds reached 0.37 km2, while in 2021, they increased to 0.51 km2. There was a high level of Phragmites australis growth in the Soleniy and Chumyanniy firths. The rapid growth of Phragmites australis in the period of 2020–2021, where the area covered by the reed doubled, is primarily attributed to eutrophication. This is due to the nutrient enrichment from agricultural lands located in the northern part of the research area near Novonekrasovskiy village. Additionally, changes in water flows and hydrological conditions can also contribute to the favorable growth of the reed. This can result in a high growth rate of Phragmites australis, which can reach up to 2 m per year and can propagate both through vegetative and sexual means, leading to the formation of large and dense clusters.

Density peaks clustering algorithm based on fuzzy and weighted shared neighbor for uneven density datasets

Uneven density data refers to data with a certain difference in sample density between clusters. The local density of density peaks clustering algorithm (DPC) does not consider the effect of sample density difference between clusters of uneven density data, which may lead to wrong selection of cluster centers; the algorithm allocation strategy makes it easy to incorrectly allocate samples originally belonging to sparse clusters to dense clusters, which reduces clustering efficiency. In this study, we proposed the density peaks clustering algorithm based on fuzzy and weighted shared neighbor for uneven density datasets (DPC-FWSN). First, a nearest neighbor fuzzy kernel function is obtained by combining K-nearest neighbor and fuzzy neighborhood. Then, local density is redefined by the nearest neighbor fuzzy kernel function. The local density can better characterize the distribution characteristics of the sample by balancing the contribution of sample density in dense and sparse areas, in order to avoid the situation that the sparse cluster does not have a cluster center. Finally, the allocation strategy for weighted shared neighbor similarity is proposed to optimize the sample allocation at the boundary of the sparse cluster. Experiments are performed on IDPC-FA, FKNN-DPC, FNDPC, DPCSA and DPC for uneven density datasets, complex morphologies datasets and real datasets. The clustering results demonstrate that DPC-FWSN effectively handles datasets with uneven density distribution.

Observations for Sjögren’s Pigment Epithelial Reticular Dystrophy in a 16-Year-Old Boy—An Extremely Rare Retinal Case Report

The purpose of this publication is to present an extremely rare case of Sjögren's pigment epithelial reticular dystrophy. So far, 10 such publications have been found in world literature. A 16-year-old boy was diagnosed due to a slight loss of visual acuity, confirmed in static perimetry/24-2/. Abnormal dense clusters of retinal pigment epithelium (RPE) cells forming a reticular network pattern (resembling a fishing net) with marked knots were detected by fundoscopy in the macular area and the mid-periphery of the retina. No abnormalities were found in the anterior segment, intraocular pressure, kinetic perimetry, Ishihara or Farnsworth D-15 tests or OCT. Fluorescein angiography confirmed blocked fluorescence from the choroidal vessels caused by the pigment in RPE. An autofluorescence test showed hypofluorescent foci corresponding to symmetrical and bilateral retinal hyperpigmentation with an RPE reticular pattern. Multifocal ERG (mfERG) revealed slight cone photoreceptor and bipolar bioelectrical dysfunction. Electrooculography (EOG) showed significant asymmetry (Arden Ratio 1.8), suggesting bioelectrical dysfunction of RPE/photoreceptors. Flash ERG (ERG) revealed only slight increase in implicit time of the a and b waves of the rod and cone responses and exclude cone-rod dystrophies. This article highlights the importance of the results of ophthalmoscopy, fluorescein angiography, autofluorescence, mfERG, fERG, EOG and genetic tests for Sjögren's reticular dystrophy with a pathogenic variant in the region of the C2 gene-c.841_849+19del (dbSNP rs9332736).

Weakly supervised semantic segmentation of histological tissue via attention accumulation and pixel-level contrast learning

Objective. Histopathology image segmentation can assist medical professionals in identifying and diagnosing diseased tissue more efficiently. Although fully supervised segmentation models have excellent performance, the annotation cost is extremely expensive. Weakly supervised models are widely used in medical image segmentation due to their low annotation cost. Nevertheless, these weakly supervised models have difficulty in accurately locating the boundaries between different classes of regions in pathological images, resulting in a high rate of false alarms Our objective is to design a weakly supervised segmentation model to resolve the above problems. Approach. The segmentation model is divided into two main stages, the generation of pseudo labels based on class residual attention accumulation network (CRAANet) and the semantic segmentation based on pixel feature space construction network (PFSCNet). CRAANet provides attention scores for each class through the class residual attention module, while the Attention Accumulation (AA) module overlays the attention feature maps generated in each training epoch. PFSCNet employs a network model containing an inflated convolutional residual neural network and a multi-scale feature-aware module as the segmentation backbone, and proposes dense energy loss and pixel clustering modules are based on contrast learning to solve the pseudo-labeling-inaccuracy problem. Main results. We validate our method using the lung adenocarcinoma (LUAD-HistoSeg) dataset and the breast cancer (BCSS) dataset. The results of the experiments show that our proposed method outperforms other state-of-the-art methods on both datasets in several metrics. This suggests that it is capable of performing well in a wide variety of histopathological image segmentation tasks. Significance. We propose a weakly supervised semantic segmentation network that achieves approximate fully supervised segmentation performance even in the case of incomplete labels. The proposed AA and pixel-level contrast learning also make the edges more accurate and can well assist pathologists in their research.

Specific absorption rate in quasispherical and elongated aggregates of magnetite nanoparticles: Experimental characterization and numerical simulation

Assemblies of magnetite nanoparticles created by the ultrasonic mechanocavitation in a viscous liquid have been obtained for use in magnetic hyperthermia. The magnetite nanoparticles inherit the perfect crystal structure of the original coarse-grained magnetite powder, with characteristic sizes of 0.2–8 μm, and have a high saturation magnetization, Ms = 92 emu/g, close to the value of pure magnetite. Measurements of the specific absorption rate (SAR) of assemblies in an alternating (ac) magnetic field showed a significant dependence of SAR on the nanoparticle concentration, as well as on the shape of particle macroaggregates. A 4-fold increase in the SAR of elongated clusters of particles oriented parallel to the action of ac magnetic field was found in comparison with an assembly of quasi-spherical clusters. In addition a sharp dependence of the SAR of an assembly of elongated clusters on the ac magnetic field direction with respect to the cluster orientation axis was revealed. The results obtained are in qualitative agreement with the results of numerical simulation of dense clusters of magnetic nanoparticles based on the solution of the stochastic Landau-Lifshitz equation.

Mechanistic insights into CDCP1 clustering on non-small-cell lung cancer membranes revealed by super-resolution fluorescent imaging

CDCP1 is a transmembrane protein that is involved in a variety of important biologicalprocesses and upregulated in a variety of human solid malignancies; however, its spatial distribution and variation at the molecular level remain unclear. To solve this problem, we first analyzed its expression level and prognostic implications in lung cancer. Then, we used super-resolution microscopy to reveal the spatial organization of CDCP1 at different levels, and found that cancer cells generated more and larger CDCP1 clusters than normal cells. Furthermore, we found that CDCP1 can be integrated into larger and denser clusters as functional domains upon activation. Our findings elucidated the significant differences of CDCP1 clustering characteristics between cancer and normal cells, and revealed the relationship between its distribution and function, which will contribute to a comprehensive understanding of its oncogenic mechanism, and will be of great help for the development of CDCP1-targeted drugs for lung cancer.

The Role of Bone Marrow Biopsy Morphology and Clinical Characteristics in Facing the Challenges of Diagnosing Primary Myelofibrosis, Polycythemia Vera, and Essential Thrombocythemia at Cipto Mangunkusumo Hospital

BackgroundMyeloproliferative neoplasm (MPN) is a neoplasm characterized by the proliferation of one or more myeloid cells and their derivatives. The limitations of molecular examination in Indonesia make the diagnosis of MPN based on clinical and histopathological examination very crucial. The aim evaluate the clinicopathological profile of primary myelofibrosis (PMF), polycythemia vera (PV), and essential thrombocythemia (ET) and identify the typical morphological characteristics of bone marrow biopsy.MethodsA retrospective study of cases diagnosed with MPN was conducted at the Department of Anatomic Pathology FMUI/RSCM in 2015-2019. Clinical data and evaluation of bone marrow morphology consisting of cellularity, erythroid myeloid ratio, cluster and megakaryocyte morphology, blast cells, fibrosis, osteosclerosis, and sinus dilatation were analyzed.ResultsA total of 172 cases were diagnosed as MPN BCR-ABL1-negative (PMF: 74; ET: 56; PV: 42). On routine blood examination, there was an increase in hemoglobin (Hb) and hematocrit (Ht) in PV, a decrease in Hb in PMF and an increase in platelets in ET (p<0.001). Splenomegaly is mostly found in PMF. Myeloid erythroid ratio was decreased in PV, normal in ET and increased in PMF (p<0.001). Megakaryocytes were arranged in loose clusters in 88.1% of PV cases and 96.4% of ET, dense clusters were found in 91.9% of PMF cases. Staghorn-like megakaryocytes were found in all ET cases and bulbous/cloud-like megakaryocytes were found in 97.3% of PMF cases. Most cases of PV (90.5%) and all cases of ET showed pre-fibrotic bone marrow (grade 0-1), while 77% of PMF cases showed fibrosis grade 2-3.ConclusionIn diagnosing PV, ET, and PMF, correlation between clinical data, laboratory, and bone marrow histomorphological evaluations, especially cellularity, myeloid:erythroid ratio, cluster and megakaryocyte morphology, degree of fibrosis, osteosclerosis, and sinus dilatation is required.

Stress Features Inferred From Induced Earthquakes in the Weiyuan Shale Gas Block in Southwestern China

AbstractStress features, particularly local stress field and earthquake stress drops, are important to understand mechanism of induced earthquakes. Since shale gas exploitations in 2015, the Weiyuan shale gas block has experienced frequent earthquakes. In this paper, we determine focal mechanisms of 257 events with ML > 1.5 by fitting three‐component waveforms, invert for direction of maximum horizontal stress for two dense earthquake clusters, and then calculate stress drops for 17 earthquakes with moment magnitudes between 2.2 and 2.75 through the spectral ratio method. The focal mechanisms of all earthquakes are reverse faulting. The orientation of local maximum horizontal compressive stress is in the ESE direction, consistent with crustal movement indicated by GPS measurements. Both the focal mechanisms and microseismicity locations suggest the existence of steeply dipping faults with dip angles of ∼70°, which are relatively rare and difficult to slip under usual conditions, unlike major induced earthquakes with strike‐slip or low‐angle thrusting faults in the midwestern United States and western Canada. However, the steeply dipping reverse events can be induced by large pore pressure from hydraulic fracturing. The stress drops range from 2.5 to 54.7 MPa, comparable to those of potentially induced earthquakes in the midwestern United States. Our results imply earthquakes in the Weiyuan area are controlled by local tectonic stress and induced by large pore pressure from hydraulic fracturing, which advances our knowledge of reactivation of steeply dipping reverse faults.

Molecular mechanism of pH-dependent modulation of Sup35 phase separation by intrinsically disordered M domain.

In budding yeast, starvation stress can trigger a protective response where most cellular processes are downregulated. The transition to this fully reversible state is accompanied by cytoplasmic acidification and subsequent reorganization. Sup35, a translation termination factor with two intrinsically disordered regions (N and M domains), is known to phase separate upon cytoplasmic acidification. Previous work showed the pH dependence of Sup35's phase separation could be attributed to a particularly dense ionizable cluster of negatively-charged residues in the intrinsically disordered M domain, which is proposed to regulate the phase separation properties of the N domain. In this work, we use NMR spectroscopy to probe the pH sensitivity of the ionizable residues in the IDR regions and their interactions. Our aim is to determine the molecular mechanism by which the M domain regulates Sup35's phase separation.

Structure and function of SARS-CoV-2 nucleocapsid protein measured using optical tweezers, confocal fluorescence, and AFM.

The nucleocapsid (N) protein, one of four structural proteins expressed by SARS-CoV-2, primarily functions by binding and compacting the ∼30 kilobase viral RNA genome (>10 μm linear length) to enable packaging into viral particles (<100 nm diameter). We use optical tweezers to isolate a single ssDNA substrate and observe the binding and compacting function of N protein in real time. The compaction is a multistep process, with kinetics consistent with initial diffusion-limited binding of free protein to the substrate followed by a sudden stochastically triggered reorganization of the protein-DNA complex into a compacted form. Experiments utilizing truncated protein variants lacking either the ordered N or C terminal domains (NTD and CTD) and the intrinsically disordered linker demonstrate that initial binding is driven by the NTD while the formation of the compacted DNA structure requires the CTD. We further localize the binding of fluorescently labeled protein along the substrate using confocal imaging, showing N protein able to both uniformly coat the entire length of the substrate non specifically and subsequently form dense protein clusters through interprotein oligomerization. Finally, various DNA and RNA substrates of differing lengths and sequences are incubated with either full length or truncated N protein and imaged using AFM. Measurement of these nucleic acid-protein complexes reveal both the binding activity of N protein and its ability to remodel complex, folded nucleic acid structures. Our results help elucidate the biophysical mechanism through which N protein enables efficient packaging of the long vRNA under physiological conditions.

Complex optical transport, dynamics, and rheology of intermediately attractive emulsions

Introducing short-range attractions in Brownian systems of monodisperse colloidal spheres can substantially impact their structures and consequently their optical transport and rheological properties. Here, for size-fractionated colloidal emulsions, we show that imposing an intermediate strength of attraction, well above but not much larger than thermal energy (approx 5.6k_{textrm{B}}T), through micellar depletion leads to a striking notch in the measured inverse mean free path of optical transport, 1/ell ^*, as a function of droplet volume fraction, phi. This notch, which appears between the hard-sphere glass transition, phi _{textrm{g}}, and maximal random jamming, phi _{textrm{MRJ}}, implies the existence of a greater population of compact dense clusters of droplets, as compared to tenuous networks of droplets in strongly attractive emulsion gels. We extend a prior decorated core-shell network model for strongly attractive colloidal systems to include dense non-percolating clusters that do not contribute to shear rigidity. By constraining this extended model using the measured 1/ell ^*(phi ), we improve and expand the microrheological interpretation of diffusing wave spectroscopy (DWS) experiments made on attractive colloidal systems. Our measurements and modeling demonstrate richness and complexity in optical transport and shear rheological properties of dense, disordered colloidal systems having short-range intermediate attractions between moderately attractive glasses and strongly attractive gels.

Growth of a Massive Black Hole in a Dense Star Cluster Via Tidal Disruption Accretion

Stars that are tidally disrupted by a massive black hole (MBH) may contribute significantly to the growth of the MBH, especially in dense nuclear star clusters. Yet, this tidal disruption accretion (TDA) of stars onto the MBH has largely been overlooked compared to the gas accretion (GA) channel in most numerical experiments until now. In this work, we implement a black hole growth channel via TDA in the high-resolution adaptive mesh refinement code Enzo to investigate its influence on an MBH seed’s early evolution. We find that an MBH seed grows rapidly from 103 to ≳106 M ⊙ in 200 Myr in some of the tested simulations. Compared to an MBH seed that grows only via GA, TDA can enhance the MBH’s growth rate by more than 1 order of magnitude. However, as predicted, TDA mainly helps the early growth of the MBH (from 103–4 to ≲105 M ⊙) while the later evolution is generally dominated by GA. We also observe that the star formation near the MBH is suppressed when TDA is the most active, sometimes with a visible cavity in gas (of size ∼ a few pc) created in the vicinity of the MBH. It is because the MBH may grow expeditiously with both GA and TDA, and the massive MBH could consume its neighboring gas faster than being replenished by gas inflows. Our study demonstrates the need to consider different channels of black hole accretion that may provide clues for the existence of supermassive black holes at high redshifts.

A multi-task approach for contrastive learning of handwritten signature feature representations

In spite of recent advances in computer vision, the classic problem of offline handwritten signature verification still remains challenging. The signature verification task has a high intra-class variability because a given user often shows high variability between its samples. Besides, signature verification is harder in the presence of skilled forgeries. Recently, in order to tackle these challenges, the research community has investigated deep learning methods for learning feature representations of handwritten signatures. When mapping signatures to a feature space, it is desired to obtain dense clusters of signature’s representations, in order to deal with intra-class variability. Besides, not only dense clusters are required but also a larger separation between different user’s clusters in the feature space. Finally, it is also desired to move away feature representations of skilled forgeries in relation to the respective dense cluster of genuine representations. This last property is hard to achieve in the real-world scenario because skilled forgeries are not readily available during training. In this work, we hypothesize that such properties can be achieved by means of a multi-task framework for learning handwritten signature feature representations based on deep contrastive learning. The proposed framework is composed of two objective-specific tasks. The first task aims to map signature examples of the same user closer within the feature space, while separating the feature representations of signatures of different users. The second task aims to adjust the skilled forgeries representations by adopting contrastive losses with the ability to perform hard negative mining. Hard negatives are examples from different classes with some degree of similarity that can be applied for training. We evaluated models obtained with the proposed framework in terms of the equal error rate on GPDSsynthetic, CEDAR and MCYT-75 datasets in writer-dependent and writer-independent verification approaches. Using synthetic and real signature datasets, Friedman tests with Bonferroni–Dunn post hoc tests were performed to compare the proposed multi-task contrastive models against the popular SigNet model as a baseline. Experiments demonstrated an statistically significant improvement in signature verification with a multi-task contrastive model based on the Triplet loss. Implementation of the method is available for download at https://github.com/tallesbrito/contrastive_sigver.

Clustering in gas‐fluidized riser flows of flexible fibers

AbstractClustering of flexible fibers in riser flows is investigated using a hybrid approach of Discrete Element Method and Computational Fluid Dynamics. Unlike spherical particles, the flexible fibers possess elongated shape, undergo significant deformation, and dissipate kinetic energies through rapid fiber deformation. The present studies show that these distinct features have significant impacts on the cluster characteristics of the fibers. A larger fiber aspect ratio leads to larger number and sizes of agglomerates, while it causes a reduction in heterogeneity of solids distribution due to the more dilute clusters with reduced packing densities. As the fibers become more flexible, the heterogeneity increases, and denser clusters are obtained. More significant effects of the fiber flexibility on the clustering are observed for the fibers with larger aspect ratios. The increased energy dissipation through the rapid fiber deformation enhances the clustering by augmenting the number and size of the agglomerates.

Scale-free chaos in the confined Vicsek flocking model.

The Vicsek model encompasses the paradigm of active dry matter. Motivated by collective behavior of insects in swarms, we have studied finite-size effects and criticality in the three-dimensional, harmonically confined Vicsek model. We have discovered a phase transition that exists for appropriate noise and small confinement strength. On the critical line of confinement versus noise, swarms are in a state of scale-free chaos characterized by minimal correlation time, correlation length proportional to swarm size and topological data analysis. The critical line separates dispersed single clusters from confined multicluster swarms. Scale-free chaotic swarms occupy a compact region of space and comprise a recognizable "condensed" nucleus and particles leaving and entering it. Susceptibility, correlation length, dynamic correlation function, and largest Lyapunov exponent obey power laws. The critical line and a narrow criticality region close to it move simultaneously to zero confinement strength for infinitely many particles. At the end of the first chaotic window of confinement, there is another phase transition to infinitely dense clusters of finite size that may be termed flocking black holes.

Dense clusters improve efficiency and foster colony development in restored Acropora cervicornis

Dense clusters improve efficiency and foster colony development in restored Acropora cervicornis