Similar Models Research Articles

Comparative analysis of deep Siamese models for medical reports text similarity

Comparative analysis of deep Siamese models for medical reports text similarity

SUMEX: A hybrid framework for Semantic textUal siMilarity and EXplanation generation

Measuring semantic similarity between two pieces of text is a widely known problem in Natural language processing(NLP). It has many applications, such as finding similar medical notes of patients to accelerate the diagnosis process, plagiarism detection, and document clustering. Most state-of-the-art models are based on machine/deep learning and lack sufficient explanations for their results, limiting their adoption in critical domains like healthcare. This paper presents a hybrid framework SUMEX (Semantic textUal siMilarity and EXplanation generation) that uniquely combines ontology with a state-of-the-art embedding-based model for semantic textual similarity. The primary strength of the framework is that it explains its results in human-understandable natural language, which is vital in critical domains such as healthcare. Experiments have been conducted on two datasets of clinical notes using four embeddings: ScispaCy, BioWord2Vec, ClinicalBERT, and a customized Word2Vec trained on clinical notes. The SUMEX framework outperforms the embedding-based model on the benchmark datasets of ClinicalSTS by improving average precision scores by 7 % and reducing the false-positives-rate by 23 %. On the Patients Similarity Dataset, the average top-five and top-three precision scores were improved by 14% and 10%, respectively, using SUMEX. The SUMEX also generates explanations for its results in natural language. The domain experts evaluated the quality of the explanations. The results show that the generated explanations are of significantly good quality, with a score of 90 % and 93 % for measures of Completeness and Correctness, respectively. In addition, ChatGPT was also used for similarity score and generating explanations. The experiments show that the SUMEX framework performed better than the ChatGPT.

Non-Autoregressive Line-Level Code Completion

Software developers frequently use code completion tools to accelerate software development by suggesting the following code elements. Researchers usually employ AutoRegressive (AR) decoders to complete code sequences in a left-to-right, token-by-token fashion. To improve the accuracy and efficiency of code completion, we argue that tokens within a code statement have the potential to be predicted concurrently. In this article, we first conduct an empirical study to analyze the dependency among the target tokens in line-level code completion. The results suggest that it is potentially practical to generate all statement tokens in parallel. To this end, we introduce SANAR, a simple and effective syntax-aware non-autoregressive model for line-level code completion. To further improve the quality of the generated code, we propose an adaptive and syntax-aware sampling strategy to boost the model’s performance. The experimental results obtained from two widely used datasets indicate that our model outperforms state-of-the-art code completion approaches of similar model size by a considerable margin, and is faster than these models with up to 9× speed-up. Moreover, the extensive results additionally demonstrate that the enhancements achieved by SANAR become even more pronounced with larger model sizes, highlighting their significance.

The Use of Rabbits Used to Propagate Human Lice for Research.

The globally important human diseases of trench fever, epidemic typhus, and relapsing fever are vectored by the human louse Pediculus humanus humanus. Although these conditions are epidemically quiescent at present, they persist in socially dysfunctional situations of war, deprivation, and crowding. The taxonomically closely related head louse, Pediculus humanus capitis, does not respect economic or social status and is quite common in most countries. The 2 types of lice are now recognized as conspecific ecotypes of a single species. While the body louse has been adapted for propagation in the laboratory by feeding in vivo on live rabbits, a similar animal model has not been developed for the host-specific head louse. Accordingly, research for treatment and control of the head louse has largely been performed by using laboratory-reared body lice. This review describes methods for the propagation of body lice in the laboratory and outlines at least 4 areas of research that require sufficient numbers of aged body louse cohorts produced in rabbits for use in controlled studies: 1) pediculicide development and resistance, 2) immunity and vaccine potential, 3) endosymbiotic bacteria needed by lice for nutrition, and 4) lice as vectors of human disease. The review concludes with a discussion of several ethical issues involved with the standard method of using unsedated rabbits and recommends consideration of providing sedating anesthetics for rabbits used in louse feeding procedures.

Risk assessment of high-speed railway CTC system based on improved game theory and cloud model

PurposeIn order to solve the problem of inaccurate calculation of index weights, subjectivity and uncertainty of index assessment in the risk assessment process, this study aims to propose a scientific and reasonable centralized traffic control (CTC) system risk assessment method.Design/methodology/approachFirst, system-theoretic process analysis (STPA) is used to conduct risk analysis on the CTC system and constructs risk assessment indexes based on this analysis. Then, to enhance the accuracy of weight calculation, the fuzzy analytical hierarchy process (FAHP), fuzzy decision-making trial and evaluation laboratory (FDEMATEL) and entropy weight method are employed to calculate the subjective weight, relative weight and objective weight of each index. These three types of weights are combined using game theory to obtain the combined weight for each index. To reduce subjectivity and uncertainty in the assessment process, the backward cloud generator method is utilized to obtain the numerical character (NC) of the cloud model for each index. The NCs of the indexes are then weighted to derive the comprehensive cloud for risk assessment of the CTC system. This cloud model is used to obtain the CTC system's comprehensive risk assessment. The model's similarity measurement method gauges the likeness between the comprehensive risk assessment cloud and the risk standard cloud. Finally, this process yields the risk assessment results for the CTC system.FindingsThe cloud model can handle the subjectivity and fuzziness in the risk assessment process well. The cloud model-based risk assessment method was applied to the CTC system risk assessment of a railway group and achieved good results.Originality/valueThis study provides a cloud model-based method for risk assessment of CTC systems, which accurately calculates the weight of risk indexes and uses cloud models to reduce uncertainty and subjectivity in the assessment, achieving effective risk assessment of CTC systems. It can provide a reference and theoretical basis for risk management of the CTC system.

Pericyte Control of Gene Expression in the Blood-Brain Barrier Endothelium: Implications for Alzheimer's Disease.

A strong body of evidence suggests that cerebrovascular pathologies augment the onset and progression of Alzheimer's disease (AD). One distinctive aspect of this cerebrovascular dysfunction is the degeneration of brain pericytes-often overlooked supporting cells of blood-brain barrier endothelium. The current study investigates the influence of pericytes on gene and protein expressions in the blood-brain barrier endothelium, which is expected to facilitate the identification of pathophysiological pathways that are triggered by pericyte loss and lead to blood-brain barrier dysfunction in AD. Bioinformatics analysis was conducted on the RNA-Seq expression counts matrix (GSE144474), which compared solo-cultured human blood-brain barrier endothelial cells against endothelial cells co-cultured with human brain pericytes in a non-contact model. We constructed a similar cell culture model to verify protein expression using western blots. The insulin resistance and ferroptosis pathways were found to be enriched. Western blots of the insulin receptor and heme oxygenase expressions were consistent with those observed in RNA-Seq data. Additionally, we observed more than 5-fold upregulation of several genes associated with neuroprotection, including insulin-like growth factor 2 and brain-derived neurotrophic factor. Results suggest that pericyte influence on blood-brain barrier endothelial gene expression confers protection from insulin resistance, iron accumulation, oxidative stress, and amyloid deposition. Since these are conditions associated with AD pathophysiology, they imply mechanisms by which pericyte degeneration could contribute to disease progression.

Potential cognitive risks of generative transformer-based AI chatbots on higher order executive functions.

Chat generative retrained transformer (ChatGPT) represents a groundbreaking advancement in Artificial Intelligence (AI-chatbot) technology, utilizing transformer algorithms to enhance natural language processing and facilitating their use for addressing specific tasks. These AI chatbots can respond to questions by generating verbal instructions similar to those a person would provide during the problem-solving process. ChatGPT has become the fastest growing software in terms of user adoption in history, leading to an anticipated widespread use of this technology in the general population. Current literature is predominantly focused on the functional aspects of these technologies, but the field has not yet explored hypotheses on how these AI chatbots could impact the evolutionary aspects of human cognitive development. Thesis: The "neuronal recycling hypothesis" posits that the brain undergoes structural transformation by incorporating new cultural tools into "neural niches," consequently altering individual cognition. In the case of technological tools, it has been established that they reduce the cognitive demand needed to solve tasks through a process called "cognitive offloading." In this theoretical article, three hypotheses were proposed via forward inference about how algorithms such as ChatGPT and similar models may influence the cognitive processes and structures of upcoming generations. By forecasting the neurocognitive effects of these technologies, educational and political communities can anticipate future scenarios and formulate strategic plans to either mitigate or enhance the cognitive influence that these factors may have on the general population. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Complete characterization of NADH dehydrogenase subunit 1 gene in human hydatid cysts

يمتلك طفيلي الاكياس العدرية سلالاتين متشابهتين (G1 و G3) ونظرا للدور البارز الذي تلعبه عائلة جينات نازعات الهيدروجين في التميز بين هاتين السلالاتين، لذا هدفت الدراسة الحالية الى اجراء التوصيف الجزيئي الكامل لاحد ابرز انواع هذه العائلة وهو الجين NAD1 (من شفرة البدء الى الي شفرة التوقف) للاكياس العدرية المعزولة جراحيا من اكباد المرضى (ولمحافظتي البصرة وميسان) في مستشفى الصدر التعليمي في محافظة البصرة، العراق. عزلت الرؤيسات الاولية من الاكياس تحت ظروف معقمة واستخدمت في استخلاص الحامض النووي منقوص الاوكسجين. اذ ضخم التتابع الكامل للجين في عينات الدراسة وانجز التتابع (السكونس) لهما باستخدام بادئات متخصصه. اظهرت نتائج تحليل السكونس المتكرر Consensus sequence امتلاك كل من عينات البصرة و ميسان 1038 زوجا من القواعد وتقع شفرة البدء GTG في الموقع 87-89 وشفرة التوقف TAA في الموقع 978-980. كما اظهرت استدلالات الاحماض الامينية الى وجود 297 حامض اميني لبروتينات عينات البصرة وميسان. فضلا عن ذلك بينت تحاليل المعلومات الحياتية احتواء هذه البروتينات على الحامض الاميني Isoleucine في الموقع 188 (I188) وهو موجود فقط في سلالة الاغنام G1. وان النموذج التركيبي لبروتينات عينات البصرة وميسان اظهر تشابها كبيرا و يقع I188 في الحلزون 10 (α10). بينت الدراسة ولاول مرة على مستوى العراق التوصيف الكامل للجين NAD1 والعائد الى سلالة الاغنام G1 المسببه لمرض الاكياس العدرية البشري.

A family reunion of "clinical cousins": Attachment and emotion in four family-oriented therapy models.

Attachment theory and the science of emotion provide a strong foundation for intervention at the family system level. Four therapeutic models in particular, Attachment-Based Family Therapy, Emotion-Focused Family Therapy, Dyadic Developmental Psychotherapy, and Emotionally Focused Family Therapy, demonstrate how a broad and accurate view of attachment relationships and emotion can be utilized to effectively intervene for a variety of presenting problems in a relational and empathic way for all involved. This paper continues a conversation that began at the Summit for Attachment and Emotion in Family Therapy in 2021 and aims to foster openness, collaboration, and affirmation between four different models of family therapy with shared theoretical roots. The presenters at the Summit and the authors of this paper view similarities across these models as validating and differences as opportunities to serve more families in unique ways, learning from one another's creativity to promote healing within families in the most effective and efficient ways possible. The paper frames the value of attachment theory and emotion science for family therapy, discusses the importance of learning from a variety of models with shared theoretical roots, presents brief summaries of the four models presented at the Summit, compares the models for similarities and complementarities, and shares highlights from each of the presenters from the Summit.

Multidimensional clustering analysis of mathematical knowledge difficulty based on Gspan

In addition to providing learners with a large amount of teaching resources, online teaching platforms can also provide learning resources and channels such as video courseware, Q&A tutoring groups, and forums. However, currently, there are still shortcomings in depth and dimensionality in mining student learning behavior data on the platform. In view of this situation, based on the learning interaction behavior, this study established the difficulty similarity model of knowledge points, and used spectral clustering to classify their difficulty. In addition, the study intended to use the maximum frequent subgraph under the Gspan framework to characterize learners’ implicit learning patterns. The outcomes expressed that the algorithm put forward in the study achieved the highest accuracy index of 98.8%, which was 1.4%, 4.0%, and 8.6% higher than Apriori-based graph mining algorithms, K-means, and frequent subgraph discovery algorithms. In terms of F1 index, the convergence value of the algorithm proposed in the study was 95.5%, which was about 2.5% higher than the last three algorithms. In addition, learners of all three cognitive levels had the highest maximum number of frequent subgraphs with sizes above 100 when the minSup value was 60%. And when the number of clusters was 3, the clustering accuracy of the three learners was the highest. In similarity calculation, the calculation method used in the study was at the minimum in terms of root mean square error and absolute error average index, which were 0.048% and 0.01% respectively. This indicated that the model proposed by the research had better classification effect on the difficulty of knowledge points for learners of different cognitive levels, and had certain application potential.

Group-aware graph neural networks for sequential recommendation

Sequential recommendations play a crucial role in recommender systems. Existing methods commonly focus on extracting sequential patterns within individual item sequences to make personalized recommendations. However, this paper argues that such a closure strategy has limitations. Firstly, the inherent uncertainty and sparsity within these sequences pose considerable challenges in accurately capturing sequential patterns. Secondly, ignoring group behaviors beyond the sequence hinders the modeling of user similarities, which are essential for recommendations. To address these limitations, this paper proposes a novel collaborative sequential recommendation approach, called Group-aware Graph Neural Networks (GaGNN). GaGNN seeks to incorporate group behaviors within sequential recommendations, thereby enhancing the modeling of user preferences through the assimilation of collective wisdom. GaGNN constructs two transition graphs, each offering a distinct behavioral perspective. The group transition graph is designed to capture the collaborative effects and learn collaborative information, whereas the individual transition graph aims to articulate individual behaviors, thereby extracting sequential information. Additionally, a multi-source information fusion module is designed to bridge semantic gaps between the two types of information and yield a holistic understanding of user preferences. Through rigorous experimental comparisons and ablation studies, the superiority of GaGNN and the beneficial integration of collaborative information are emphatically validated.

Introspective Deep Metric Learning.

This paper proposes an introspective deep metric learning (IDML) framework for uncertainty-aware comparisons of images. Conventional deep metric learning methods focus on learning a discriminative embedding to describe the semantic features of images, which ignore the existence of uncertainty in each image resulting from noise or semantic ambiguity. Training without awareness of these uncertainties causes the model to overfit the annotated labels during training and produce overconfident judgments during inference. Motivated by this, we argue that a good similarity model should consider the semantic discrepancies with awareness of the uncertainty to better deal with ambiguous images for more robust training. To achieve this, we propose to represent an image using not only a semantic embedding but also an accompanying uncertainty embedding, which describes the semantic characteristics and ambiguity of an image, respectively. We further propose an introspective similarity metric to make similarity judgments between images considering both their semantic differences and ambiguities. The gradient analysis of the proposed metric shows that it enables the model to learn at an adaptive and slower pace to deal with the uncertainty during training. Our framework attains state-of-the-art performance on the widely used CUB-200-2011, Cars196, and Stanford Online Products datasets for image retrieval. We further evaluate our framework for image classification on the ImageNet-1 K, CIFAR-10, and CIFAR-100 datasets, which shows that equipping existing data mixing methods with the proposed introspective metric consistently achieves better results (e.g., +0.44% for CutMix on ImageNet-1 K).

An Experimental Platform of Heating Network Similarity Model for Test of Integrated Energy Systems

An Experimental Platform of Heating Network Similarity Model for Test of Integrated Energy Systems

Elegans-AI: How the connectome of a living organism could model artificial neural networks

This paper introduces Elegans-AI models, a class of neural networks that leverage the connectome topology of the Caenorhabditis elegans to design deep and reservoir architectures. Utilizing deep learning models inspired by the connectome, this paper leverages the evolutionary selection process to consolidate the functional arrangement of biological neurons within their networks. The initial goal involves the conversion of natural connectomes into artificial representations. The second objective centers on embedding the complex circuitry topology of artificial connectomes into both deep learning and deep reservoir networks, highlighting their neural-dynamic short-term and long-term memory and learning capabilities. Lastly, our third objective aims to establish structural explainability by examining the heterophilic/homophilic properties within the connectome and their impact on learning capabilities. In our study, the Elegans-AI models demonstrate superior performance compared to similar models that utilize either randomly rewired artificial connectomes or simulated bio-plausible ones. Notably, these Elegans-AI models achieve a top-1 accuracy of 99.99% on both Cifar10 and Cifar100, and 99.84% on MNIST Unsup. They do this with significantly fewer learning parameters, particularly when reservoir configurations of the connectome are used. Our findings indicate a clear connection between bio-plausible network patterns, the small-world characteristic, and learning outcomes, emphasizing the significant role of evolutionary optimization in shaping the topology of artificial neural networks for improved learning performance.

Across time, space, and genres: measuring probabilistic grammar distances between varieties of Mandarin

Abstract This paper aims to quantify distances between varieties of Mandarin (diachronic, regional, and situational) as a function of the similarity in the choice between syntactic variants in the Mandarin theme-recipient alternation (yŭ/gěi dative alternation). We use a novel corpus-based method, Variation-Based Distance and Similarity Modeling, which draws inspiration from work in comparative sociolinguistics and quantitative dialectometry. Analysis reveals that, while there is a relatively stable probabilistic grammar across the investigated varieties, historical varieties do exhibit a relatively higher degree of heterogeneity than synchronic varieties. Despite the overall high similarity of the latter, we identify substantial probabilistic differences between fictional writings of Modern Mainland Mandarin and all other synchronic varieties. Our findings thus provide evidence in support of the hypothesis that the transition from Early Mandarin to Modern Mandarin over the past two centuries has witnessed salient grammatical shifts and also empirically demonstrate the interaction between genre variability and regional variability in Modern Mandarin.

Online addiction analysis and identification of students by applying gd-LSTM algorithm to educational behaviour data

Abstract Internet has become the primary source of extracurricular entertainment for college students in today’s information age of Internet entertainment. However, excessive Internet addiction (IA) can negatively impact a student’s daily life and academic performance. This study used Stochastic models to gather data on campus education behaviour, extract the temporal characteristics of university students’ behaviour, and build a Stochastic dropout long short-term memory (LSTM) network by fusing Dropout and LSTM algorithms in order to identify and analyse the degree of IA among university students. The model is then used to locate and forecast the multidimensional vectors gathered, and finally to locate and evaluate the extent of university students’ Internet addiction. According to the experiment’s findings, there were 4.23% Internet-dependent students among the overall (5,861 university students), and 95.66% of those students were male. The study examined the model using four dimensions, and the experimental findings revealed that the predictive model suggested in the study had much superior predictive performance than other models, scoring 0.73, 0.72, 0.74, and 0.74 on each dimension, respectively. The prediction model outperformed other algorithms overall and in the evaluation of the four dimensions, performing more evenly than other algorithms in the performance comparison test with other similar models. This demonstrated the superiority of the research model.

Digital Sustainability for Human Resource Management Canvas Meta-Synthesis Approach

In the era of Digital Transformation (DT) and Sustainable Development (SD), the pivotal role of Human Resources (HR) in addressing business challenges has become increasingly evident. These challenges directly impact different HR departments and processes, prompting a need for a strategic overhaul. HR management should continuously adapt its service delivery model to align with the business model and create a balance between internal and external organizational expectations and employee needs, considering the two fundamental challenges of digital transformation and sustainable development, in an agile manner. The critical role of sustainable human resource management in fostering overall business sustainability is evident. However, in many cases, HR management has failed to recognize and demonstrate its cost structure, revenue flow, and social and environmental benefits transparently for the business. This research focuses on presenting the sustainable digital human resource management canvas and its connection to the business canvas to achieve sustainability in the digital era. Employing a qualitative research methodology, this study conducts a meta-synthesis of two pivotal concepts-sustainable human resource management and digital human resource management- leveraging data from reputable databases including Science Direct, Scopus, and Web of Science. Following meticulous screening and examination, a total of 17 articles from Q1-ranked journals were selected as the final articles for a more detailed and in-depth review. The findings from the meta-synthesis results consist of 5 dimensions and 23 components, ultimately presenting the sustainable digital human resource management canvas by modeling it after the business canvas. The distinct feature of this canvas, compared to other similar and conventional models, is the inclusion of individual sustainability in the value proposition and the creation of a digital value component, including artificial intelligence, the Internet of Things, and big data.

Can the giant planets of the Solar System form via pebble accretion in a smooth protoplanetary disc?

Context. Prevailing N-body planet formation models typically start with lunar-mass embryos and show a general trend of rapid migration of massive planetary cores to the inner Solar System in the absence of a migration trap. This setup cannot capture the evolution from a planetesimal to embryo, which is crucial to the final architecture of the system. Aims. We aim to model planet formation with planet migration starting with planetesimals of ~10−6−10−4 M⊕ and reproduce the giant planets of the Solar System. Methods. We simulated a population of 1000-5000 planetesimals in a smooth protoplanetary disc, which was evolved under the effects of their mutual gravity, pebble accretion, gas accretion, and planet migration, employing the parallelized N-body code SyMBAp. Results. We find that the dynamical interactions among growing planetesimals are vigorous and can halt pebble accretion for excited bodies. While a set of results without planet migration produces one to two gas giants and one to two ice giants beyond 6 au, massive planetary cores readily move to the inner Solar System once planet migration is in effect. Conclusions. Dynamical heating is important in a planetesimal disc and the reduced pebble encounter time should be considered in similar models. Planet migration remains a challenge to form cold giant planets in a smooth protoplanetary disc, which suggests an alternative mechanism is required to stop them at wide orbits.

Globotriaosylsphingosine improves risk stratification of kidney progression in Fabry disease patients

BackgroundKidney damage is common in patients with Fabry disease (FD), but more accurate information about the risk of progression to kidney failure is needed for clinical decision-making. In particular, FD patients with mild renal involvement often lack timely intervention and treatment. We aimed to utilize a model to predict the risk of renal progression in FD patients. MethodsBetween November 2011 and November 2019, ERT-naive patients with FD were recruited from three medical centers in China. To assess the risk of a 50% decline in the estimated glomerular filtration rate (eGFR) or end-stage kidney disease (ESKD), Cox proportional hazards models were utilized. The performance of these models was assessed using discrimination, calibration, and reclassification. ResultsA total of 117 individuals were enrolled. The mean follow-up time was 4.8 years, during which 35 patients (29.9 %) progressed to the composite renal outcomes. Male sex, baseline proteinuria, eGFR and globotriaosylsphingosine (Lyso-Gb3) were found to be independent risk factors for kidney progression by the Cox model, based on which a combined model containing those clinical variables and Lyso-Gb3 and clinical models including only clinical indicators were constructed. The two prediction models had relatively good performance, with similar model fit measured by R2 (59.8 % vs. 61.1 %) and AIC (51.54 vs. 50.08) and a slight increase in the C statistic (0.949 vs. 0.951). Calibration curves indicated closer alignment between predicted and actual renal outcomes in the combined model. Furthermore, subgroup analysis revealed that Lyso-Gb3 significantly improved the predictive performance of the combined model for kidney prognosis in low-risk patients with a baseline eGFR over 60 ml/min/1.73 m2 or proteinuria levels less than 1 g/d when compared to the clinical model. ConclusionsLyso-Gb3 improves the prediction of kidney outcomes in FD patients with a low risk of progression, suggesting that these patients may benefit from early intervention to assist in clinical management. These findings need to be externally validated.

Testing validity of 1D models for impurity fraction scaling for divertor detachment with EDGE2D-EIRENE

Predictions of the Huber–Chankin (HC) scaling for the upstream impurity fraction were verified in a series of EDGE2D-EIRENE (‘code’) runs for highly radiating plasmas with nitrogen injection. The main quantity extracted from the code was poloidally averaged, from X-point to X-point, separatrix impurity fraction cZ in the main scrape-off layer (SOL). Variation of the main working gas (H, D and T) revealed a qualitative agreement between the model and code results owing to the very large isotope difference in the predicted cZ values caused primarily by the inverse isotope mass dependence of the H-mode power threshold assumed in the HC model and implemented in the code. At the same time, the variation of the toroidal field and safety factor in deuterium cases yielded no correlation between the model predictions and code results. The code showed much higher local impurity fractions (fZ ) in the divertor compared to the main SOL, as well as large case-to-case variations in the divertor to the main SOL ratio of impurity fractions. The analysis of code results has wide-ranging consequences not only for the HC model, but also for other similar 1D models which use simple geometry ignoring strong neutral recycling in the divertor/ Different topology makes plasma parameters in the divertor and main SOL very different, resulting in different impurity charge state composition. Missing mechanisms in 1D codes (e.g. friction and thermo-forces exerted on impurity ions by main working gas ions) lead to impurity density redistribution. Neglecting all above factors, 1D models assume a constant impurity fraction along field lines.