SUNY Ulster

This paper explores the application of deep learning techniques in the fusion of high dynamic range (HDR) images, emphasizing its transformative impact on traditional HDR imaging methods. HDR images are renowned for capturing a broader range of luminosity; however, traditional methods face challenges such as camera shake and ghosting in dynamic scenes. The introduction of deep learning has automated and enhanced the HDR image generation process, particularly in image fusion, deblurring, and artifact correction. This paper reviews relevant deep learning algorithms and architectures, analyzes the strengths and limitations of current HDR imaging approaches, and suggests future research directions aimed at improving efficiency, accuracy, and applicability across various domains.

Whereas the demands for data privacy and security in healthcare have been tending to an ever-higher level, integration between blockchain and Artificial Intelligence (AI) technology has thereafter become a new innovative method that could solve these problems. Since many studies related to the application of blockchain in the field of data security and artificial intelligence in medical diagnosis have been proposed, research on the integration of this new technology in the medical field is still required. The paper focused essentially on the amalgamation of these two technologies, elaborating on novel practices in clinical diagnosis, medical records, and pharmaceutical supply chain management. This paper conducted a literature review and case analysis to examine how blockchain ensures data immutability with decentralized storage mechanisms and how AI utilizes this data through deep learning and real-time analysis for its decision-making system. The examination results indicate that blockchain decentralizes data for storing patients' privacy and integrity, while AI can enhance the accuracy of diagnosis and decision-making for treatment. Moreover, information transparency and operational efficiency will be further developed by integrating the two technologies. Based on the conclusion of this study, blockchain and AI integrated technology holds great potential to enable the healthcare industry toward intelligent and personalized development. Despite the technical standardization barrier and protection of privacy, future research should focus on optimizing scalability to further advance the integration of these two technologies in healthcare services.

This article provides a comprehensive review of face recognition research, focusing on advancements made over the past century. It presents a detailed examination of the core concepts, principles, steps, and classifications of face recognition technology. The review highlights the practical applications of face recognition in contemporary contexts and summarizes key datasets and preprocessing methods used in the field. The paper categorizes face recognition methods into three main types and places particular emphasis on hybrid methods. It explores the principles and research processes associated with these methods, offering an in-depth analysis of their results. Among the various techniques reviewed, deep learning methods emerge as the most promising for face recognition due to their superior performance. This review serves as a valuable resource for students and novice researchers by providing a clear overview of current research methodologies and tools. Additionally, it outlines potential research directions and contributes to the advancement of the field of computer vision.

The rapid development of artificial intelligence and deep learning has significantly influenced the domain of image creation, finding extensive applications in applications in fields like medical imaging, computer vision, and entertainment. Despite these advancements, challenges remain, especially in enhancing the quality and variety of produced images. This paper concentrates on applying Variational Autoencoders (VAEs) to image generation, a topic of increasing importance due to the model’s theoretical interpretability and stability. Through a detailed analysis of VAE principles, architecture, and applications, this research underscores the model’s capabilities in producing high-quality, varied images and its effectiveness in tasks such as image denoising and enhancement. The study also analysis the limitations of VAEs, like the inclination to generate blurry images, and discusses potential improvements, including hybrid models and enhanced loss functions. The results of this research enhance the comprehension of VAE’s capabilities and provide a foundation for future research aimed at advancing image generation technologies.

Hypervirulence and multi-drug resistance are two separate qualities of Klebsiella pneumoniae that have posed a public health threat in separate strains for decades now. The convergence of these two phenotypes into singular strains of Klebsiella pneumoniae have created a super bug, causing both severe infections in the young and healthy and having limited antibiotic treatment available against them. These strains were first discovered in Asia in 2013 but are now emerging all around the globe. In this study, hypervirulence was assessed in 493 known carabapenemase-producing Klebsiella pneumoniae strains isolated and cryopreserved between 2019 and 2021 from various specimen types from health institutions around the Maltese islands. Three phenotypic tests were used, namely, the string test for hypermucoviscosity, lateral flow assays for the identification of K1 and K2 capsular serotypes, and quantification of siderophore production. Since the latter test was the one that was known to be the most significant for the detection of hypervirulence, and none of the strains surpassed the set limit of a concentration of 30µg/ml, none of the 493 assayed strains were deemed hypervirulent. Hypermucoviscous, K1 and K2 capsular serotype belonging strains were found, though, so much so that a statistically significant relationship was found between hypermucoviscosity and capsular serotype. Other statistical tests carried out exploring the relationship between hypermucoviscosity and siderophore concentration and hypermucoviscosity and type of carabapenemase gene were not found to be statistically significant. According to this research, hypervirulent carabapenemase-producing Klebsiella pneumoniae strains have not yet arrived in the Maltese archipelago.

This work presents thermoelectric (TE) devices design of delta-doped $\beta$ -(AlxGa $_{{1}-{x}}$ )2O3/Ga2O3 metal insulator semiconductor high electron mobility transistors (MIS-HEMTs) using TCAD simulations. The TE properties of devices were comprehensively investigated with various temperature, gate voltages, gate lengths, delta-doping concentrations, and positions. With high delta-doping concentrations, a parasitic current channel is induced and that reduces electron chemical potential, resulting in high conductivity, a low Seebeck coefficient, and a reduced turn on voltage. Moving delta-doping positions closer to the $\beta$ -(AlxGa $_{{1}-{x}}$ )2O3/Ga2O3 interface enhances the concentration of the 2-D electron gas (2DEG), which screens the strong polar optical-phonon scattering and improves 2DEG mobility. For delta-doping positions at 1 nm, the power factor is improved due to quantum effect and energy filter effect, allowing the trade-off relationship between $\sigma$ and S to be mitigated. Expanding gate lengths increases channel electron temperature at gate edge near drain side. These results provide valuable insights and crucial guidance for the design of high-performance $\beta$ -(AlxGa $_{{1}-{x}}$ )2O3/Ga2O3 MIS-HEMTs for TE and temperature sensing applications.

Devices located in remote regions often lack coverage from well-developed terrestrial communication infrastructure. This not only prevents them from experiencing high quality communication services but also hinders the delivery of machine learning services in remote regions. In this paper, we propose a new federated learning (FL) methodology tailored to space-air-ground integrated networks (SAGINs) to tackle this issue. Our approach strategically leverages the nodes within space and air layers as both (i) edge computing units and (ii) model aggregators during the FL process, addressing the challenges that arise from the limited computation powers of ground devices and the absence of terrestrial base stations in the target region. The key idea behind our methodology is the adaptive data offloading and handover procedures that incorporate various network dynamics in SAGINs, including the mobility, heterogeneous computation powers, and inconsistent coverage times of incoming satellites. We analyze the latency of our scheme and develop an adaptive data offloading optimizer, and also characterize the theoretical convergence bound of our proposed algorithm. Experimental results confirm the advantage of our SAGIN-assisted FL methodology in terms of training time and test accuracy compared with various baselines.

We present the design and implementation of WaveFlex, the first smart surface that enhances Private 5G networks operating under the shared-license framework in the Citizens Broadband Radio Service frequency band. WaveFlex works in the presence of frequency diversity: multiple nearby base stations operating on different frequencies, as dictated by a Spectrum Access System coordinator. It also handles time dynamism: due to the dynamic sharing rules of the CBRS band, base stations occasionally switch channels, especially when priority users enter the network. Finally, WaveFlex operates independently of the network itself, not requiring access to nor modification of the gNB or UEs, yet it remains compliant with and effective on prevailing cellular protocols. We have designed and fabricated WaveFlex on a custom multi-layer PCB, software defined radio based network monitor, and supporting control software and hardware. Our experimental evaluation benchmarks operational Private 5G and LTE networks running at full line rate. In a realistic indoor office scenario, 5G experimental results demonstrate an 8.58~dB average SNR gain, and an average throughput gain of 10.77 Mbps under a single gNB, and 12.84 Mbps under three gNBs, corresponding to throughput improvements of 18.4% and 19.5%, respectively.

A bstract In CFTs, the partition function of a line defect with a cusp depends logarithmically on the size of the line with an angle-dependent coefficient: the cusp anomalous dimension. In the first part of this work, we study the general properties of the cusp anomalous dimension. We relate the small cusp angle limit to the effective field theory of defect fusion, making predictions for the first couple of terms in the expansion. Using a concavity property of the cusp anomalous dimension we argue that the Casimir energy between a line defect and its orientation reversal is always negative (“opposites attract”). We use these results to determine the fusion algebra of Wilson lines in $\mathcal{N}$ N = 4 SYM as well as pinning field defects in the Wilson-Fisher fixed points. In the second part of the paper we obtain nonperturbative numerical results for the cusp anomalous dimension of pinning field defects in the Ising model in d = 3, using the recently developed fuzzy-sphere regularization. We also compute the pinning field cusp anomalous dimension in the O ( N ) model at one-loop in the ε -expansion. Our results are in agreement with the general theory developed in the first part of the work, and we make several predictions for impurities in magnets.

A bstract We consider 3+1 dimensional Quantum Field Theories (QFTs) coupled to the dilaton and the graviton. We show that the graviton-dilaton scattering amplitude receives a universal contribution which is helicity flipping and is proportional to ∆ c − ∆ a along any RG flow, where ∆ c and ∆ a are the differences of the UV and IR c - and a -trace anomalies respectively. This allows us to relate ∆ c − ∆ a to spinning massive states in the spectrum of the QFT. We test our predictions in two simple examples: in the theory of a massive free scalar and in the theory of a massive Dirac fermion (a more complicated example is provided in a companion paper [1]). We discuss possible applications.

Background Therapeutic management of hair loss is frequently complicated by a lack of high‐quality evidence and reliant on the use of unlicensed therapies. Treatment decision‐making is predominantly based on expert opinion, local availability, personal experience, and cost, which make informed choices challenging for clinicians and patients in this area. Objectives The aims were to determine prescribing patterns amongst UK Dermatologists with a special interest in hair disorders, when treating mild‐moderate alopecia areata (AA), severe AA (including alopecia totalis/alopecia universalis), female pattern hair loss (FPHL) and frontal fibrosing alopecia (FFA). Methods Consultant members of the British Hair and Nail society, a special interest group affiliated to the British Association of Dermatologists, were invited to participate from across the United Kingdom. Participants were questioned on their current prescribing patterns in both NHS and private practice, were asked to rank their first‐to‐fifth line treatments for each condition and highlight the treatment they perceive as most effective for each disorder. Results Twenty‐six Consultant Dermatologists completed the questionnaire, from twenty‐three institutions. For treatment of mild‐moderate AA, topical corticosteroids were used first line amongst 65% (n = 17) of respondents, and 82% (n = 23) reported that intralesional corticosteroids were the most effective treatment. For severe AA, oral corticosteroids were used first line amongst 38% (n = 10) of respondents, and 25% (n = 8) reported that oral corticosteroids were the most effective treatment. For FPHL, topical minoxidil was used first line amongst 84% (n = 25) of respondents, and 42% (n = 10) reported that oral minoxidil was the most effective treatment. For FFA, topical corticosteroids were used first line amongst 62% (n = 16) of respondents, and 37% (n = 14) reported that hydroxychloroquine was the most effective treatment. Conclusions This study reports real‐world prescribing practices amongst dermatologists treating common hair loss conditions. These results aim to support clinicians with decision making for managing hair loss conditions.

A bstract We consider type IIB string theory with N D3 branes and various configurations of sevenbranes, such that the string coupling g s is fixed to a constant finite value. These are the simplest realizations of F-theory, and are holographically dual to rank N Argyres-Douglas conformal field theories (CFTs) with SU(2) and SU(3) flavor groups, and Minahan-Nemeschansky CFTs with E 6 , E 7 and E 8 flavor groups. We use the Seiberg-Witten curves of these theories to compute the mass deformed sphere free energy F ( m ) at large N in terms of novel matrix models with non-polynomial potentials. We show how F ( m ) can be used along with the analytic bootstrap to fix the large N expansion of flavor multiplet correlators in these CFTs, which are dual to scattering of gluons on AdS 5 × S ³ , and in the flat space limit determine the effective theory of sevenbranes in F-theory. As a first step in this program, we use the matrix models to compute the log N term in F ( m ) and thereby fix the logarithmic threshold in the AdS 5 × S ³ holographic correlator, which matches the flat space prediction.

An in-depth history of the European native oyster in Northern Irish waters has been absent from international and regional peer-reviewed publications. The knowledge of historical losses and a need to recover ecosystems for habitat and biodiversity purposes are primary drivers in an urgency to restore Ostrea edulis. However, a comprehensive record of O. edulis in Northern Ireland is required to assist with this work. The authors compiled a list of relevant references from grey material, rare historical archives, library collections, government reports and peer-reviewed publications. Archival reviews have been tabulated into a timeline, which documents site location, exploitation, sites of significant interest and socio-economic histories of the coastal communities who relied on the oyster. The reference material identified four distinctive phases of exploitation whereby harvesting transits from personal use to commercialization, collapse and then restoration. The study revealed that O. edulis harvests in the early 1800s in Northern Ireland were predominantly destined for export to supply collapsing stocks throughout Britain. Fishing was intense with the fishery closed by 1903. However, the species has proved to be extremely resilient with small artisanal fisheries still in existence today. This research will offer habitat managers guidance in relation to site selection and anthropogenic pressures when restoring the European flat oyster to the iconic historical beds of the Northern Irish Sea loughs.

We study various aspects of global symmetries in five-dimensional superconformal field theories. Whenever a supersymmetry-preserving relevant deformation is available, the infrared gauge theory description might exhibit a finite order mixed ’t Hooft anomaly between a 1-form symmetry and the instantonic symmetry. This anomaly constrains the flavor symmetry group acting faithfully on the SCFT and the consistency of certain RG flows. As an additional example, we consider the instructive case of three-dimensional $\mathcal {N}=4$ SQED. Finally, we discuss the compatibility between conformal invariance and the presence of 1-form and 2-group global symmetries.

Downs’ (1957) showed that it was irrational to vote, and irrational to acquire political information if one’s sole motivation is economic, self-interest. In response, three solutions have been proposed. Voters are motivated by: expressive identification; civic duty; and weak altruism. There is no consensus as to why individuals vote. One reason is that none of the previous tests tested the competing theories against each other. To overcome the limitation of past tests, I use a MTURK survey, which has measures of all the competing theories. I then perform non-nested (Vuong and Clarke) and nested tests of the competing theories. Non-nested testing assumes that the competing theories are strictly exclusive, i.e., only one theory is correct. Nested tests assume that elements of the competing models are combined. My evidence tentatively supports a nested model of voting behavior.

Purpose of Review While some parenting interventions designed for early-life obesity prevention have demonstrated short-term success, there is limited evidence of longer-term impacts and feasibility with underrepresented populations. The goal of this review was to examine existing general parenting programs for parents of children 0–5 years that were not designed to target obesity but investigated long-term effects on parenting and/or were conducted with underrepresented populations to offer recommendations for the modification or development of parenting-focused obesity prevention programs. Recent Findings Most studies with sustained impacts on parenting in underrepresented populations were brief, group-based programs for parents of children 2–5 years. Many effective interventions included guided practice of skills and cultural adaptations. Unique intervention approaches included remote or school-based delivery models and motivational interviewing. Summary Brief, group-based programs for parents of young children may be a promising approach to achieving longer-term impacts of parenting interventions on obesity risk among underrepresented populations.

DeepFakes have raised serious societal concerns, leading to a great surge in detection-based forensics methods in recent years. Face forgery recognition is a standard detection method that usually follows a two-phase pipeline, i.e ., it extracts the face first and then determines its authenticity by classification. While those methods perform well in ideal experimental environment, they face challenges when dealing with DeepFakes in the wild involving complex background and multiple faces of varying sizes. Moreover, most face forgery recognition methods can only process one face at a time. One straightforward way to address this issue is to simultaneous process multi-face by integrating face extraction and forgery detection in an end-to-end fashion by adapting advanced object detection architectures. However, as these object detection architectures are designed to capture the discriminative features of different object categories rather than the subtle forgery traces among the faces, the direct adaptation suffers from limited representation ability. In this paper, we propose Contrastive Multi-FaceForensics (COMICS), an end-to-end framework for multi-face forgery detection. COMICS integrates face extraction and forgery detection in a seamless manner and adapts to the advanced object detection architectures. The core of the proposed framework is a bi-grained contrastive learning approach that explores face forgery traces at both the coarse- and fine-grained levels. Specifically, coarse-grained level contrastive learning captures the discriminative features among positive and negative proposal pairs at multiple layers produced by the proposal generator, and the fine-grained level contrastive learning captures the pixel-wise discrepancy between the forged and original areas of the same face and the pixel-wise content inconsistency among different faces. Extensive experiments on the OpenForensics and FFIW datasets demonstrate that our method outperforms other counterparts and shows great potential for being integrated into various architectures. Codes are available at https://github.com/zhangconghhh/COMICS.