Similar Features Research Articles

A consequential one-night stand: Episodic historical hybridization leads to mitochondrial takeover in sympatric desert ant-eating spiders

Disentangling the genomic intricacies underlying speciation and the causes of discordance between sources of evidence can offer remarkable insights into evolutionary dynamics. The ant-eating spider Zodarion nitidum, found across the Middle East and Egypt, displays yellowish and blackish morphs that co-occur sympatrically. These morphs additionally differ in behavioral and physiological features and show complete pre-mating reproductive isolation. In contrast, they possess similar sexual features and lack distinct differences in their mitochondrial DNA. We analyzed both Z. nitidum morphs and outgroups using genome-wide and additional mitochondrial DNA data. The genomic evidence indicated that Yellow and Black are reciprocally independent lineages without signs of recent admixture. Interestingly, the sister group of Yellow is not Black but Z. luctuosum, a morphologically distinct species. Genomic gene flow analyses pinpointed an asymmetric nuclear introgression event, with Yellow contributing nearly 5 % of its genome to Black roughly 320,000 years ago, intriguingly aligning with the independently estimated origin of the mitochondrial DNA of Black. We conclude that the blackish and yellowish morphs of Z. nitidum are long-diverged distinct species, and that the ancient and modest genomic introgression event registered resulted in a complete mitochondrial takeover of Black by Yellow. This investigation underscores the profound long-term effects that even modest hybridization events can have on the genome of organisms. It also exemplifies the utility of phylogenetic networks for estimating historical events and how integrating independent lines of evidence can increase the reliability of such estimations.

Hierarchical Degradation-aware Network for Full-Reference Image Quality Assessment

Full-Reference Image Quality Assessment (FR-IQA) algorithms excel in evaluating perceptual distortions by comparing reference and distorted images. However, as the severity and quantity of distortions in datasets increase, existing FR-IQA methods struggle to capture complex nonlinear perceptual features. This limitation results in reduced adaptability and inaccurate assessments for images with more severe or multiple distortions. Recognizing the importance of understanding image degradation mechanisms, we propose a novel hierarchical degradation-aware network (HDaN) method. First, by exploring the degradation mechanisms from the reference image to the distorted image, our degradation network matches distortions that align more closely with the human visual system (HVS). Next, we design a convertor to project the matched features into multiple spaces, creating multidimensional feature representations that more comprehensively capture the complexity of image distortions rather than being confined to a single feature space. Then, we calculate a similarity matrix between the distorted and mapped features, selecting the most similar (top-k) features for merging. Finally, a regression network maps the merged features to quality scores, providing the final quality prediction. The experimental results demonstrate that our proposed HDaN method outperforms traditional deep learning-based FR-IQA methods. Specifically, the HDaN shows higher PLCC and SROCC metrics on benchmark datasets, significantly improving over existing methods. Moreover, the method exhibits better adaptability to images with varying degrees and types of distortions, thereby greatly enhancing the overall performance of IQA.

A synthesis of the REE-Fe-polymetallic mineral system of the REE-line, Bergslagen, Sweden: New mineralogical and textural-paragenetic constraints

Rare earth elements (REE) have gained increasing significance for numerous technologies, particularly in today’s rapidly expanding “green transition” applications such as wind generators and electric vehicle traction engines. Among the more well-known REE mineralisation types in Sweden, together with alkaline intrusions and apatite-iron oxide ores, is the classic yet enigmatic REE-Fe-polymetallic mineral system of Bastnäs-type. The mineralisation type is regional in context and occurs in a discontinuous SW-NE-striking belt (the REE-line) in the west-central part of the Palaeoproterozoic Bergslagen ore province, Sweden. This contribution is aimed at integrating and synthesising existing geological mineralogical, and textural features with new observations from both well-known and several lesser-known, underexplored or previously unrecognised REE-enriched occurrences within this belt, and to discuss key features within the context of mineral systems modelling. A considerable diversity in both the style and abundance of REE mineralisation as well as in discrete REE mineralogy is evident both regionally across the entire REE-line and locally within different ore districts or mine fields. These variations also extend laterally within or across different stratigraphic levels, and within different host rocks, primarily skarn-altered metacarbonates but also variably altered felsic metavolcanic rocks. Many of the mineralisations share similar textural features, which record a protracted evolution with multiple stages of formation or replacement of REE-minerals. The earliest recognised REE assemblages feature fine-grained cerite-(CeCa) with minor bastnäsite-(Ce) – bastnäsite-(La) or fluorbritholite-(Ce) – fluorbritholite-(Y) or locally britholite-(Ce) – britholite-(Y) minerals. Such assemblages typically display anhedral-granoblastic textures appearing in folded assemblages, all suggesting recrystallisation and ductile deformation during regional metamorphism of REE-minerals that had formed during an early stage of the Svecokarelian orogeny. Overprinting overgrowths and cross-cutting vein-like features of allanite-group minerals likely represent different stage(s) of REE mineralisation and (re)-mobilisation during this orogenic evolution. Several of the REE-enriched occurrences contain variably abundant and diverse polymetallic Cu-Mo-Bi-(Co) sulphide mineralisation that typically occur in late paragenetic positions and show a prevalence to REE-rich assemblages, often dominated by different allanite-group minerals. Sulphide and REE mineralisation are locally strongly associated with metamorphic minerals formed during metamorphism of variably Mg-(Fe)-altered metavolcanic rocks. The diversity in style and intensity of the REE mineralisations, along with variations in textures and specific mineralogy, suggest slight differences in the ore-forming conditions or environment at the time of mineralisation. Additionally, these differences may also reflect variations in the preservation or modification processes that operated later in the evolution of the Svecokarelian orogeny, also featuring remobilisation of REEs and sulphidic minerals. The insights gained from all available evidence synthesised herein, from micro-scale to province-scale, help define key proxies for prospectivity mapping. The combined evidence supports that the primary ore-forming stages in the mineral system coincided with felsic volcanism and associated sub-volcanic to plutonic processes at around 1.9 Ga.

A novel cryptic splice donor due to synonymous variant in VPS13A as an underlying cause of a chorea-acanthocytosis in a large family

Chorea-acanthocytosis (ChAc) is a rare inherited disease of the nervous system. In this disease the neurological manifestations are associated with acanthocytosis of the red blood cells. The clinical features appear in the third to fourth decades of life. Generalized weakness, choreiform movement disorder, decline in cognition, and psychiatric symptoms are the characteristic features of the disease. The differential diagnosis between Huntington's disease and ChAc is difficult because both the diseases share similar neurological features. Herein, we recruited a large family with multiple individuals initially diagnosed as having Huntington's disease. Analysis of the DNA samples of affected individuals by exome sequencing detected a synonymous variant (NM_001018037.2; c.5040C > T) in the VPS13A. Multiple splice site detection tools were used to predict the potential pathogenicity of the novel synonymous variant. The variant, identified in this study, was predicted to be a cryptic splice donor site that may lead to aberrant pre-mRNA splicing. Reverse transcriptase PCR analyses of patient blood-derived RNA showed activation of a cryptic mid‐exon splice donor, leading to frameshift. The variant was confirmed in all other affected and unaffected individuals using Sanger sequencing.This is the first report of synonymous variants of VPS13A as an underlying cause of ChAc. Our results provide the first direct evidence of the involvement of a synonymous variant of VPS13A in ChAc. Additionally, this study emphasized the importance of considering VPS13A gene mutations in the screening of Huntington's patients.

Symptomatic Low- And High-degree Carotid Stenoses Exhibit Similar Features of Plaque Instability

Symptomatic Low- And High-degree Carotid Stenoses Exhibit Similar Features of Plaque Instability

Adaptive positioning of large-scale aircraft components using a simplified image-based visual servoing via online measurement of ball-head center

In digital alignment systems, ensuring fast and accurate alignment between the ball head on large aircraft components and the ball socket of the positioner is the key to aircraft assembly efficiency and quality. Image-based visual servoing is one of the most advanced techniques with high efficiency and positioning accuracy, but it does not account for measuring the ball-head position. Moreover, variable target positions, similar feature interference, and poor lighting conditions pose great challenges to the fast and robust visual measurement of ball-head position. Hence, an online visual measurement method is introduced to measure the absolute position of the ball head in real time. An adaptive region of interest extraction algorithm and an improved edge following method are introduced to extract the elliptical contour from the ball-head image, and ellipse detection is achieved through a simple triplet selection algorithm. A geometric model based on the perspective projection of the sphere is established for estimating the ball-head center. Then, the simplified image-based visual servoing method is developed by constantizing the image Jacobian matrix. Next, an adaptive positioning method of large-scale aircraft components with multiple ball heads is proposed from a holistic perspective. Finally, experiments show that the final positioning accuracy of each ball head in the X/Y plane is less than ± 0.05 mm, with an efficiency almost ten times that of the conventional method. After adaptive adjustment, the position and angle errors of the simulated component are reduced within 0.6 mm and 0.3°, respectively.

Innovative multi-stage matching for counting anything

Few-shot counting (FSC) is the task of counting the number of objects in an image that belong to the same category, by using a provided exemplar pattern. By replacing the exemplar, we can effectively count anything, even in cases where we have no prior knowledge of that category’s exemplar. However, due to the variations within the same category and the impact of inter-class similarity, it is challenging to achieve accurate intra-class similarity matching using conventional similarity comparison methods. To tackle these issues, we propose a novel few-shot counting method called Multi-stage Exemplar Attention Match Network (MEAMNet), which increases the accuracy of matching, reduces the impact of noise, and enhances similarity feature matching. Specifically, we propose a multi-stage matching strategy to obtain more stable and effective matching results by acquiring similar feature in different feature spaces. In addition, we propose a novel feature matching module called Exemplar Attention Match (EAM). With this module, the intra-class similarity representation in each stage will be enhanced to achieve a better matching of the key feature. Experimental results indicate that our method not only significantly surpasses the state-of-the-art (SOTA) methods in most evaluation metrics on the FSC-147 dataset but also achieves comprehensive superiority on the CARPK dataset. This highlights the outstanding accuracy and stability of our matching performance, as well as its exceptional transferability. We will release the code at https://github.com/hzg0505/MEAMNet.

Scientific and practical basis of cephalopods dehydration processes

Nowadays, it is relevant to solve the problems of adapting optimal resource- and energy-efficient modes to technologies for processing food raw materials. The work is focused on the analysis of the prerequisites to find similar features in the dehydration of food capillary-porous colloidal materials by means of various methods. The authors have studied the patterns of cephalopods dehydration using the example of octopus and squid under different heat treatment regimes. It has been proved that there is only one critical point characterizing the transition from moisture with a lower binding energy to moisture with a higher one, and an equation has been proposed for finding the critical humidity during dehydration of the objects been studied. A generalized mathematical model has been developed to determine the duration of the process at the current moisture content of the food material based on dimensionless similarity criteria characterizing the phenomena been investigated. The dependences of the moisture removal rate for different heat treatment methods of cephalopods are given according to their initial moisture content, the specific surface area of the material and the operating factors that determine the process. The obtained dependencies and patterns open up unique opportunities for labor-saving design in the food industry using rational operating modes of industrial enterprises, optimization of heat treatment processes in general and the creation of energy-efficient equipment new designs.

Combined LC-MS/MS feature grouping, statistical prioritization, and interactive networking in msFeaST.

Computational metabolomics workflows have revolutionized the untargeted metabolomics field. However, the organization and prioritization of metabolite features remains a laborious process. Organizing metabolomics data is often done through mass fragmentation-based spectral similarity grouping, resulting in feature sets that also represent an intuitive and scientifically meaningful first stage of analysis in untargeted metabolomics. Exploiting such feature sets, feature-set testing has emerged as an approach that is widely used in genomics and targeted metabolomics pathway enrichment analyses. It allows for formally combining groupings with statistical testing into more meaningful pathway enrichment conclusions. Here, we present msFeaST (mass spectral Feature Set Testing), a feature-set testing and visualization workflow for LC-MS/MS untargeted metabolomics data. Feature-set testing involves statistically assessing differential abundance patterns for groups of features across experimental conditions. We developed msFeaST to make use of spectral similarity-based feature groupings generated using k-medoids clustering, where the resulting clusters serve as a proxy for grouping structurally similar features with potential biosynthesis pathway relationships. Spectral clustering done in this way allows for feature group-wise statistical testing using the globaltest package, which provides high power to detect small concordant effects via joint modeling and reduced multiplicity adjustment penalties. Hence, msFeaST provides interactive integration of the semi-quantitative experimental information with mass-spectral structural similarity information, enhancing the prioritization of features and feature sets during exploratory data analysis. The msFeaST workflow is freely available through https://github.com/kevinmildau/msFeaST and built to work on MacOS and Linux systems.

Moho Depth and Crustal Velocity Structure Beneath Stations RTC and TAM from Teleseismic Receiver-Function Analysis

We apply the P-wave receiver-function technique to invert teleseismic data to investigate the S- wave velocity structure beneath stations RTC and TAM in West Africa. Teleseismic waveform receiver function analysis is a widely used technique for studying crustal structure beneath broadband seismic stations. Our results show that the crust beneath RTC is relatively complex with large velocity fluctuations. In the upper crust, a low velocity layer extends from 8 to 12 km deep. At Tamanrasset crustal structure east and west of the station differ. We find a high-velocity zone between 2 and 8 km to the east that we attribute to a high conductivity unit corresponding to intrusions described in the literature. We find no similar feature west of the station. Our velocity models for RTC and TAM are consistent with their differing tectonic environments. TAM, on a stable craton, displays a fairly homogeneous velocity structure while RTC, on thick sediments in an ocean-continent transition zone, has more heterogeneous structure. This structural difference is reflected as well by the depth to Moho, ~20 km at RTC and nearly 38 km at TAM, although both have gradational velocity increases with depth.

Comparative Analysis of Classification and Segmentation Performance of Different Dog Breeds Based on Mask R-CNN

The classification of different image data is a common task in real life, which has been achieved as early as convolutional neural networks. With the emergence of R-CNN (Region CNN), simple classification tasks began to develop towards object detection tasks, followed by various segmentation tasks. After Faster R-CNN [4], Mask R-CNN [3] achieved instance segmentation while detecting objects. This algorithm model needs to distinguish different target objects in the image panel and learn a large number of characteristics to represent the details of each target object. In dog data images, there are differences and similarities in the characteristics of dog species, such as fur color, texture, and body shape. So in this article, the author selected and created two different dog datasets, representing targets with significant differences and similarities in features. Compare and analyze the performance of Mask R-CNN object detection and instance segmentation using these two dog datasets.

Immune signature of gene expression pattern shared by autism spectrum disorder and Huntington's disease

Autism spectrum disorder (ASD) and Huntington's disease (HD) are complex neurological conditions with unclear causes and limited treatments, affecting individuals, families, and society. Despite ASD and HD representing two opposing stages of neuronal development and degeneration, they share similar clinical-pathological features in motor function. In this study, we leveraged transcriptomic data from the prefrontal cortex available in public databases to identify shared transcriptional characteristics of ASD and HD. Differential expression analysis revealed that the majority of differentially expressed genes (DEGs) were up-regulated in ASD carriers, whereas most DEGs were down-regulated in HD carriers. Among the DEGs shared between both diseases, three out of seven protein-coding genes were related to the immune system. Furthermore, we identified two enriched pathways shared between ASD and HD DEGs. The gene interaction network analysis unveiled four hub genes shared by both diseases, all of which are associated with immune functions. The findings suggest a shared gene expression pattern in the prefrontal cortex of people with ASD and HD, closely linked to the immune system. These findings will contribute to exploring the biological mechanisms underlying the shared phenotypes of these two diseases from an immunological perspective.

Temporopolar blurring signifies abnormalities of white matter in mesial temporal lobe epilepsy.

The single-center retrospective cohort study investigated underlying pathogenic mechanisms and clinical significance of patients with temporal lobe epilepsy and hippocampal sclerosis (TLE-HS), in the presence/absence of gray-white matter abnormalities (usually called "blurring"; GMB) in ipsilateral temporopolar region (TPR) on MRI. The study involved 105 patients with unilateral TLE-HS (60 GMB+ and 45 GMB-) who underwent standard anterior temporal lobectomy, along with 61 healthy controls. Resected specimens were examined under light microscope. With combined T1-weighted and DTI data, we quantitatively compared large-scale morphometric features and exacted diffusion parameters of ipsilateral TPR-related superficial and deep white matter (WM) by atlas-based segmentation. Along-tract analysis was added to detect heterogeneous microstructural alterations at various points along deep WM tracts, which were categorized into inferior longitudinal fasciculus (ILF), uncinate fasciculus (UF), and temporal cingulum. Comparable seizure semiology and postoperative seizure outcome were found, while the GMB+ group had significantly higher rate of HS Type 1 and history of febrile seizures, contrasting with significantly lower proportion of interictal contralateral epileptiform discharges, HS Type 2, and increased wasteosomes in hippocampal specimens. Similar morphometric features but greater WM atrophy with more diffusion abnormalities of superficial WM was observed adjacent to ipsilateral TPR in the GMB+ group. Moreover, microstructural alterations resulting from temporopolar GMB were more localized in temporal cingulum while evenly and widely distributed along ILF and UF. Temporopolar GMB could signify more severe and widespread microstructural damage of white matter rather than a focal cortical lesion in TLE-HS, affecting selection of surgical procedures.

Characterization of Endothelial Cell Subclusters in Localized Scleroderma Skin with Single-Cell RNA Sequencing Identifies NOTCH Signaling Pathway

Localized scleroderma (LS) is an autoimmune disease characterized by inflammation and fibrosis, leading to severe cutaneous manifestations such as skin hardening, tightness, discoloration, and other textural changes that may result in disability. While LS shares similar histopathologic features and immune-fibroblast interactions with systemic sclerosis (SSc), its molecular mechanisms remain understudied. Endothelial cells (EC) are known to play a crucial role in SSc but have not been investigated in LS. Single-cell RNA sequencing (scRNA-seq) now allows for detailed examination of this cell type in the primary organ of interest for scleroderma, the skin. In this study, we analyzed skin-isolated cells from 27 LS patients (pediatric and adult) and 17 healthy controls using scRNA-seq. Given the known role of EC damage as an initial event in SSc and the histologic and clinical skin similarities to LS, we focused primarily on endothelial cells. Our analysis identified eight endothelial subclusters within the dataset, encompassing both disease and healthy samples. Interaction analysis revealed that signaling from diseased endothelial cells was predicted to promote fibrosis through SELE interaction with FGFBP1 and other target genes. We also observed high levels of JAG in arterial endothelial cells and NOTCH in capillary endothelial cells, indicating the activation of a signaling pathway potentially responsible for epidermal abnormalities and contributing to LS pathogenesis. In summary, our scRNA-seq analysis identified potential disease-propagating endothelial cell clusters with upregulated pathways in LS skin, highlighting their importance in disease progression.

Utilizing Inherent Bias for Memory Efficient Continual Learning: A Simple and Robust Baseline

Learning from continuously evolving data is critical in real-world applications. This type of learning, known as Continual Learning (CL), aims to assimilate new information without compromising performance on prior knowledge. However, learning new information leads to a bias in the network towards recent observations, resulting in a phenomenon known as catastrophic forgetting. The complexity increases in Online Continual Learning (OCL) scenarios where models are allowed only a single pass over data. Existing OCL approaches that rely on replaying exemplar sets are not only memory-intensive when it comes to large-scale datasets but also raise security concerns. While recent dynamic network models address memory concerns, they often present computationally demanding, over-parameterized solutions with limited generalizability. To address this longstanding problem, we propose a novel OCL approach termed “Bias Robust online Continual Learning (BRCL).” BRCL retains all intermediate models generated. These models inherently exhibit a preference for recently learned classes. To leverage this property for enhanced performance, we devise a strategy we describe as ‘utilizing bias to counteract bias.’ This method involves the development of an Inference function that capitalizes on the inherent biases of each model towards the recent tasks. Furthermore, we integrate a model consolidation technique that aligns the first layers of these models, particularly focusing on similar feature representations. This process effectively reduces the memory requirement, ensuring a low memory footprint. Despite the simplicity of the methodology to guarantee expandability to various frameworks, extensive experiments reveal a notable performance edge over leading methods on key benchmarks, getting continual learning closer to matching offline training. (Source code will be made publicly available upon the publication of this paper.)

Role of inhibition in overcoming interferences of misconception under similar feature saliency: An eye-tracking study of the projectile motion problem

Misconceptions coexisting with scientific understanding pose significant challenges in physics education. Inhibitory control may enable individuals to overcome interference from misconceptions. However, discerning the role of inhibitory control becomes intricate when the saliency of scientific- and misconception-related features varies in a problem. This study investigates the role of inhibitory control in overcoming such misconceptions when scientific and misconception-related features are similarly salient, particularly focusing on the range-time (range determines flight time) misconception. We adopted a negative priming (NP) paradigm with the eye-tracking technique in the study. Thirty-six college physics majors participated in the experiment. In the NP paradigm, the participants first compared the flight times in a neutral (same range) or incongruent (longer range, less flight time) primes and then completed the same task in a congruent (longer range, more flight time) probe. Results revealed longer reaction times for incongruent primes compared to neutral primes and longer reaction times to respond to congruent probes following incongruent primes than when neutral primes were present beforehand, suggesting that the inhibitory control may be involved in overcoming the interference from the range-time misconception. Moreover, the eye-tracking analysis highlighted an early overt attention to the salient misconception-related feature (longer range) during primes, along with additional attentional investment necessary to inhibit the range-time misconception. During probes, an early shift of attention away from the salient misconception-related feature and a greater investment of attention on processing the congruent probes after the incongruent primes than that after the neutral primes were observed, indicating a possible effect of the previous inhibitory control in the primes on the subsequent cognitive process on the probes. These findings contribute to a deeper understanding of the cognitive process underlying physics conceptual change and provide insights into physics education. Published by the American Physical Society 2024

A LiDAR-depth camera information fusion method for human robot collaboration environment

With the evolution of human–robot collaboration in advanced manufacturing, multisensor integration has increasingly become a critical component for ensuring safety during human–robot interactions. Given the disparities in range scales, densities, and arrangement patterns among multisensor data, such as that from depth cameras and LiDAR, accurately fusing information from multiple sources has emerged as a pressing need to safeguard human–robot safety. This paper focuses on LiDAR and depth cameras, addressing the challenges posed by the differences in data collection range, point density, and distribution patterns which complicate information fusion. We propose a heterogeneous sensor information fusion method for human–robot collaborative environments. To solve the problem of substantial differences in point cloud range scales, a moving sphere space coarse localization algorithm is introduced, narrowing down the scale of interest based on similar features. Furthermore, to address the challenge of significant density differences and low overlap rates between point clouds, we present an improved FPFH coarse registration algorithm based on overlap ratio and an enhanced ICP fine registration algorithm based on the generation of corresponding points. The method proposed herein is applied to the fusion of information from a 64-line LiDAR and a depth camera within a human–robot collaboration scene. Experimental results demonstrate an absolute translational accuracy of 4.29 cm and an absolute rotational accuracy of 0.006 rad, meeting the requirements for heterogeneous sensor information fusion in the context of human–robot collaboration.

Multimodal, microspectroscopic speciation of legacy phosphorus in two US mid‐Atlantic agricultural soils

AbstractTo understand phosphorus (P) mobility in agricultural soils and its potential environmental risk, it is essential to directly measure solid phase P speciation. Often, bulk P K‐edge X‐ray absorption near edge structure (XANES) spectroscopy followed by linear combination fitting (LCF) is utilized to determine the solid P phases in soil. However, this method may limit results to only a few major phases. Additionally, XANES spectra for different P species may have very similar features, leading to an over‐ or underestimate of their contribution to LCF. Here, an improved P speciation by pairing multimodal microbeam‐X‐ray fluorescence (µ‐XRF) mapping coupled with µ‐XANES (microbeam‐X‐ray absorption near edge structure) analysis to directly speciate major and minor P phases on the micron scale is provided. We combined maps of both tender (P, sulfur, aluminum, and silicon) and hard energy (calcium, iron [Fe], and manganese) elements to evaluate the elemental co‐locations with P. To better account for uncertainty assigning XANES peaks to individual compounds, a more quantitative fingerprinting by “spectral feature analysis” was completed. With this analysis, an R‐factor is reported for the fit. These results were compared to traditional LCF. Pre‐edge fitting results revealed the presence of a two‐component pre‐edge feature for phosphate adsorbed to ferrihydrite. Additionally, phytate co‐precipitated with ferrihydrite (Phytate‐Fe‐Cop) had a pre‐edge feature, indicating direct association with Fe. Lastly, a unique P species associated with manganese oxide was identified in the soil via multimodal mapping and µ‐XANES. These results allow for better prediction of P dissolution and mobility.

ANALISIS TINGKAT KEPENTINGAN DAN KINERJA (IMPORTANCE-PERFORMANCE ANALYSIS) NILAI KEGUNAAN APLIKASI M-COMMERCE

Good application quality can bring user satisfaction to form user’s loyalty and trust in the company. The level of user satisfaction is different between Gojek and Grab even though each them has similarities in application features. Gojek and Grab are trying to make position as the most popular by the community in getting people's needs quickly and efficiently with huge competitive. This study aims to analyze and compare the usability qualities of Gojek and Grab based on the User Experience Questionnaire (UEQ) approach on the dimensions of attractiveness, perception, efficiency, dependability, stimulation and novelty. Measurements are made using Importance-Performance Analysis to graphically measure customer satisfaction so that service quality improvement priorities can be established. The data was collected by questionnaire to 32 selected respondents who accessed the Gojek and Grab applications in the Greater Jakarta area in August-December 2022. Based on the results of the study, the position of the usability quality attributes in Gojek and Grab was visually almost the same for each IPA quadrant. The Gojek application is superior in 5 dimensions to Grab, in terms of attractiveness, perception, dependability, stimulation, and novelty. Grab has one dimension that is superior to Gojek, in terms efficiency.

Contrastive Speaker Representation Learning with Hard Negative Sampling for Speaker Recognition.

Speaker recognition is a technology that identifies the speaker in an input utterance by extracting speaker-distinguishable features from the speech signal. Speaker recognition is used for system security and authentication; therefore, it is crucial to extract unique features of the speaker to achieve high recognition rates. Representative methods for extracting these features include a classification approach, or utilizing contrastive learning to learn the speaker relationship between representations and then using embeddings extracted from a specific layer of the model. This paper introduces a framework for developing robust speaker recognition models through contrastive learning. This approach aims to minimize the similarity to hard negative samples-those that are genuine negatives, but have extremely similar features to the positives, leading to potential mistaken. Specifically, our proposed method trains the model by estimating hard negative samples within a mini-batch during contrastive learning, and then utilizes a cross-attention mechanism to determine speaker agreement for pairs of utterances. To demonstrate the effectiveness of our proposed method, we compared the performance of a deep learning model trained with a conventional loss function utilized in speaker recognition with that of a deep learning model trained using our proposed method, as measured by the equal error rate (EER), an objective performance metric. Our results indicate that when trained with the voxceleb2 dataset, the proposed method achieved an EER of 0.98% on the voxceleb1-E dataset and 1.84% on the voxceleb1-H dataset.