检索结果-内蒙古大学图书馆

MeDSLIP: Medical Dual-Stream Language-Image Pre-training for Fine-grained Alignment

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Fan, Wenrui Suvon, Mohammod Naimul Islam Zhou, Shuo Liu, Xianyuan Alabed, Samer Osmani, Venet Swift, Andrew Chen, Chen Lu, Haiping Department of Computer Science University of Sheffield Sheffield United Kingdom Centre of Machine Intelligence University of Sheffield Sheffield United Kingdom Department of Infection Immunity and Cardiovascular Disease University of Sheffield Sheffield United Kingdom Department of Clinical Radiology Sheffield Teaching Hospitals Sheffield United Kingdom INSIGNEO Institute for in Silico Medicine University of Sheffield Sheffield United Kingdom Information School University of Sheffield Sheffield United Kingdom Department of Engineering Science University of Oxford Oxford United Kingdom Department of Computing Imperial College London London United Kingdom

Vision-language pre-training (VLP) models have shown significant advancements in the medical domain. Yet, most VLP models align raw reports to images at a very coarse level, without modeling finegrained relationships between anatomical and pathological concepts outlined in reports and the corresponding semantic counterparts in images. To address this problem, we propose a Medical Dual-Stream Language- Image Pre-training (MeDSLIP) framework. Specifically, MeDSLIP establishes vision-language fine-grained alignments via disentangling visual and textual representations into anatomy-relevant and pathologyrelevant streams. Moreover, a novel vision-language Prototypical Contrastive Learning (ProtoCL) method is adopted in MeDSLIP to enhance the alignment within the anatomical and pathological streams. MeDSLIP further employs cross-stream Intra-image Contrastive Learning (ICL) to ensure the consistent coexistence of paired anatomical and pathological concepts within the same image. Such a cross-stream regularization encourages the model to exploit the synchrony between two streams for a more comprehensive representation learning. MeDSLIP is evaluated under zero-shot and supervised fine-tuning settings on three public datasets: NIH CXR14, RSNA Pneumonia, and SIIM-ACR Pneumothorax. Under these settings, MeDSLIP outperforms six leading CNN-based models on classification, grounding, and segmentation tasks. Copyright © 2024, The Authors. All rights reserved.

关键词： Alignment

Integrating computational thinking practices into early childhood education in culturally responsive ways: Insights from research–practice partnership

学校读者我要写书评

暂无评论

Future in Educational Research

Future in Educational Research 2024年第4期2卷 359-381页

作者： Jeanne Hubelbank Mia Dubosarsky Shakhnoza Kayumova Tiffany Davis Nea Sann Shawn Fortin Gillian Smith Independent Evaluator Sudbury Massachusetts USA The STEM Education Center Worcester Polytechnic Institute Worcester Massachusetts USA Department of STEM Education and Teacher Development University of Massachusetts Dartmouth Massachusetts USA Ashburnham Westminster Regional School District Ashburnham Massachusetts USA Worcester Child Development Head Start Worcester Massachusetts USA Teaching & Learning Belchertown Public Schools Belchertown Massachusetts USA Interactive Media & Game Development Department of Computer Science Worcester Polytechnic Institute Worcester Massachusetts USA

The study examines the research–practice partnership (RPP) model aimed at co-constructing and integrating computational thinking (CT) in culturally responsive (CR) ways within early childhood and elementary (PreK-5) classrooms, focusing on teachers' RPP experiences. Recent research underlines the importance of integrating CT in early childhood and elementary education, demonstrating that culturally, linguistically, and developmentally responsive computational activities in PreK-5 settings are key to such efforts. However, there is a gap in research regarding effective professional development models that would prepare early childhood teachers in terms of both disciplinary ideas and practices of CT, as well as attending to cultural, linguistic, and developmental differences in young children in asset-based ways. Our research contributes to this body of knowledge by examining how RPP can be a model for teacher learning in which teachers, administrators, and researchers can co-develop knowledge and confidence in integrating CT into the PreK-5 teaching curriculum and practices in CR ways. The study outlines the collaborative development of processes and frameworks co-constructed by the RPP educators, as well as teacher-developed curricular materials and lesson plans integrating CT in CR ways. Findings include teachers' experience of these lessons and teacher insights about the support needed to incorporate CT and CR practices into their existing curricula. The study concludes by highlighting the potential challenges and opportunities inherent in such endeavors, thereby contributing to the broader discourse on supporting CT integration into early childhood classrooms in CR ways.

关键词： computational thinking cultural responsive practices early childhood research practice partnership teacher professional development

A Novel Xi'An Drum Music Generation Method Based on Bi-Lstm Deep Reinforcement Learning

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Tian-Mian, Liang Peng, Li Yu-Mei, Cao Xiao-Ming, Wang Xiao-Jun, Wu Key Laboratory of Intelligent Computing Service Technology for Folk Song Ministry of Culture and Tourism Xi’an710119 China School of Computer Science Shaanxi Normal University Xi’an710119 China Engineering Laboratory of Teaching Information Technology of Shaanxi Province Xi’an710119 China Key Laboratory of Modern Teaching Technology Ministry of Education Xi’an710062 China Xi'an Key Laboratory of Cultural Tourism Resources Development and Utilization Xi’an710062 China

In order to solve the problem that most of the conventional deep learning LSTM music generation network models do not consider the rules of music theory and the scarcity of tracks faced by Xi'an drum music in music generation, as well as the difficulties in composition, this paper proposes a Bi-LSTM network based deep reinforcement learning model to generate Xi'an drum music. This paper proposes an improved method of constructing generation strategy through reward and punishment scores in the simulation generation stage. In order to allow the model to fully take into account the real folk drum music creation process, in view of the unique creation theoretical style of Xi'an drum music, such as pitch, chord, mode, etc., abstract concept constraints such as chord progression and music theory rules are converted into computer language. Through the construction of chord reward mechanism and music principle reward mechanism, the harmony constraint of drum music is realized by using the overall value return, and the systematic generation of drum music is realized. The experiment proves that based on Bi-LSTM deep reinforcement learning model, Xi'an drum music with high quality and artistic aesthetics can be basically generated. © 2023, The Authors. All rights reserved.

关键词： Music

Kernel Estimation of Truncated Volterra Filter Model Based on DFP Technique and Its Application to Chaotic Time Series Prediction

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2019年第1期28卷 127-135页

作者： ZHANG Yumei BAI Shulin LU Gang WU Xiaojun Key Laboratory of Modern Teaching Technology Ministry of EducationShaanxi Normal University School of Computer Science Shaanxi Normal University School of Electronics and Information Northwestern Polytechnical University

In order to overcome some problems caused by improper parameters selection when applying Least mean square(LMS), Normalized LMS(NLMS) or Recursive least square(RLS) algorithms to estimate coefficients of second-order Volterra filter, a novel DavidonFletcher-Powell-based Second-order Volterra filter(DFPSOVF) is proposed. Analysis of computational complexity and stability are presented. Simulation results of system parameter identification show that the DFP algorithm has fast convergence and excellent robustness than LMS and RLS algorithm. Prediction results of applying DFPSOVF model to single step predictions for Lorenz chaotic time series illustrate stability and convergence and there have not divergence problems. For the measured multiframe speech signals, prediction accuracy using DFPSOVF model is better than that of Linear prediction(LP).The DFP-SOVF model can better predict chaotic time series and the real measured speech signal series.

关键词： Chaos Davidon-Fletcher-Powell algorithm Prediction model Second-order Volterra filter Speech signal

Professional practice profile, treatment preferences, and the bases for clinical, educational, and research among Nigerian physiotherapists

学校读者我要写书评

暂无评论

Bulletin of Faculty of Physical Therapy 2023年第1期28卷 1-12页

作者： Mbada, Chidozie Emmanuel Onigbinde, Omotola Adam Oyewole, Olufemi Oyeleye Binuyo, Overcomer Temiloluwa Gebrye, Tadesse Egwu, Michael Ogbonnia Karstens, Sven Fatoye, Francis Department of Health Professions Faculty of Health and Education Manchester Metropolitan University Manchester UK Department of Medical Rehabilitation College of Health Sciences Obafemi Awolowo University Ile-Ife Nigeria Department of Physiotherapy Olabisi Onabanjo University Teaching Hospital Sagamu Nigeria College of Health Sciences University of KwaZulu-Natal Durban South Africa Therapeutic Sciences Department of Computer Science Trier University of Applied Sciences Trier Germany

Physiotherapy education and practice have country-specific peculiarities which may limit globalization in health care. This study aimed to characterize physiotherapy practice and treatment preferences, educational qualifications, and research in Nigeria, with a view of providing vital information for transnational integration and collaboration. A cross-sectional survey of 104 Nigerian physiotherapists was conducted. The Physical Therapy Practice Questionnaire and a self-developed proforma were used as survey tools. The mean age of respondents was 33.5 ± 9.4 years. About two-fifth of all respondents (39.4%) had an MSc and mostly practice as clinicians (51.0%) in teaching hospitals (34.6%). The respondents were mostly involved in general practice (50.0%), with a caseload of 1–10 patients per day (67.3%). Soft tissue mobilization (83%), proprioceptive neuromuscular facilitation (76%), breathing exercises (77%), and transcutaneous electrical neuromuscular stimulation (83%) were commonly used. Respondents were familiar databases and evidence-based resources (81.2%) and mostly utilize PubMed (73.3%). Regular case conferences with professional colleagues (47.6%) and treatment planning of between 11 and 30 min (40.6%) were common. Educators spend 1–3 h planning educational work (91.8%). Clinical decision-making is mostly based on professional experience, while journals are the primary resource for educational information. Physiotherapy practice in Nigeria is degree based and requires registration board’s licensure. Practitioners deal with a high caseload and utilize a wide range of techniques and modalities and have tendencies to utilize personal experience and research in making clinical decisions. The parity in education and practice with advanced climes inadvertently gives physiotherapy practice in Nigeria a global purview.

关键词：

PARITY OF 3-REGULAR PARTITION NUMBERS AND DIOPHANTINE EQUATIONS

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ballantine, Cristina Merca, Mircea Radu, Cristian-Silviu Department of Mathematics and Computer Science College of The Holy Cross WorcesterMA01610 United States Department of Mathematical Methods and Models Fundamental Sciences Applied in Engineering Research Center University Politehnica of Bucharest BucharestRO-060042 Romania Johannes Kepler University Linz4040 Austria

Let b3(n) be the number of 3-regular partitions of n. Recently, W. J. Keith and F. Zanello discovered infinite families of Ramanujan type congruences modulo 2 for b3(2n) involving every prime p with p ≡ 13, 17, 19, 23 (mod 24), and O. X. M. Yao provided new infinite families of Ramanujan type congruences modulo 2 for b3(2n) involving every prime p > 5. In this paper, we introduce new infinite Ramanujan type congruences modulo 2 for b3(2n). They complement naturally the results of Keith-Zanello and Yao and involve primes in P = {p prime: ∃ j ∈ {1, 4, 8}, x, y ∈ Z, gcd(x, y) = 1 with x2 + 216y2 = jp} whose Dirichlet density is 1/6. As a key ingredient in our proof we show that of the number of primitive solutions for x2 + 216y2 = pm, p ∈ P, p ∤ m and pm ≡ 1 (mod 24), is divisible by 8. Here, the difficulty arises from the fact that 216 is not idoneal. We also give a conjectural exact formula for the number of solutions for this Diophantine equation. In the second part of the article, we study reversals of Euler-type identities. These are motivated by recent work of the second author on a reversal of Schur’s identity which involves 3-regular partitions weighted by the parity of their *** Codes 11D09, 11D45, 11P83, 05A17, 11F33 © 2022, CC BY.

关键词： Algebra

ASYMPTOTIC EXPECTED T-FUNCTIONALS OF RANDOM POLYTOPES WITH APPLICATIONS TO Lp SURFACE AREAS

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Hoehner, Steven Li, Ben Roysdon, Michael Thäle, Christoph Department of Mathematics & Computer Science Longwood University United States Department of Mathematical Methods in Physics Warsaw University Poland Institute for Computational and Experimental Research in Mathematics United States Department of Pure Mathematics Tel Aviv University Israel Faculty of Mathematics Ruhr University Bochum Germany

An asymptotic formula is proved for the expected T -functional of the convex hull of independent and identically distributed random points sampled from the Euclidean unit sphere in n according to an arbitrary positive continuous density. As an application, the approximation of the sphere by random polytopes in terms of Lp surface area differences is *** Codes 52A22, 52A27, 60D05 (Primary) 52A20, 52B11 (Secondary) Copyright © 2022, The Authors. All rights reserved.

关键词： Topology

Bending of a circular flexible plate taking into account the initial curvature 49

学校读者我要写书评

暂无评论

Bending of a circular flexible plate taking into account the...

49th International Conference on Vibroengineering

作者： Mamatova, G.U. Sugirbekova, A.K. Bekaulov, N.M. Department of Chemical Physics and Materials Science Al-Farabi Kazakh National University Almaty050040 Kazakhstan Department of Chair of Mathematics Physics and Informatics Teaching Methods Abai Kazakh National Pedagogical University Almaty050010 Kazakhstan Department of Computer Engineering and Information Security International IT University Almaty050000 Kazakhstan

In this article, an analytical solution of the problem of bending a round plate with an initial curvature is obtained. A round plate with an initial curvature as a structural element is widely used, the calculation of which leads to many questions related to the design of round Foundation plates, turbine disks, flexible shaft connections and other. The nonlinear theory of round flexible plates provides the key to explaining the process of loss of stability, which often leads to complete destruction of the structure. Therefore, there is a high need for analytical methods for solving problems of calculating the stress-strain state of round flexible plates, taking into account the initial curvature. These problems are mathematically reduced to differential equations with variable coefficients, the exact solution of which, as a rule, does not exist. Therefore, the construction of analytical solutions to these problems is very relevant. Copyright © 2020 G. U. Mamatova, et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

关键词： Differential equations

Learning in Audio-visual Context: A Review, Analysis, and New Perspective

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wei, Yake Hu, Di Tian, Yapeng Li, Xuelong The Gaoling School of Artificial Intelligence Beijing Key Laboratory of Big Data Management and Analysis Methods Renmin University of China Beijing100872 China The Department of Computer Science University of Texas at Dallas RichardsonTX75080 United States Northwestern Polytechnical University Xi’an710072 China

Sight and hearing are two senses that play a vital role in human communication and scene understanding. To mimic human perception ability, audio-visual learning, aimed at developing computational approaches to learn from both audio and visual modalities, has been a flourishing field in recent years. A comprehensive survey that can systematically organize and analyze studies of the audio-visual field is expected. Starting from the analysis of audio-visual cognition foundations, we introduce several key findings that have inspired our computational studies. Then, we systematically review the recent audio-visual learning studies and divide them into three categories: audio-visual boosting, cross-modal perception and audio-visual collaboration. Through our analysis, we discover that, the consistency of audio-visual data across semantic, spatial and temporal support the above studies. To revisit the current development of the audio-visual learning field from a more macro view, we further propose a new perspective on audio-visual scene understanding, then discuss and analyze the feasible future direction of the audio-visual learning area. Overall, this survey reviews and outlooks the current audio-visual learning field from different aspects. We hope it can provide researchers with a better understanding of this area. A website including constantly-updated survey is released: https://***/audio-visual-learning/. Copyright © 2022, The Authors. All rights reserved.

关键词： Surveys