67639 research outputs found
Sort by
Principled priors for Bayesian inference of circular models
No full textAdvancements in computational power and methodologies have enabled research on massive datasets. However, tools for analyzing data with directional or periodic characteristics, such as wind directions and customers' arrival time in 24-hour clock, remain underdeveloped. While statisticians have proposed circular distributions for such analyses, significant challenges persist in constructing circular statistical models, particularly in the context of Bayesian methods. These challenges stem from limited theoretical development and a lack of historical studies on prior selection for circular distribution parameters. In this article, we propose a principled, practical and systematic framework for selecting priors that effectively prevents overfitting in circular scenarios, especially when there is insufficient information to guide prior selection. We introduce well-examined Penalized Complexity (PC) priors for the most widely used circular distributions. Comprehensive comparisons with existing priors in the literature are conducted through simulation studies and a practical case study. Finally, we discuss the contributions and implications of our work, providing a foundation for further advancements in constructing Bayesian circular statistical models
Optimization Methods and Software for Federated Learning
No full textFederated Learning (FL) is a novel, multidisciplinary Machine Learning paradigm where multiple clients, such as mobile devices, collaborate to solve machine learning problems. Initially introduced in Konečný et al. (2016a,b); McMahan et al. (2017), FL has gained further attention through its inclusion in the National AI Research and Development Strategic Plan (2023 Update) of the United States (Science and on Artificial Intelligence, 2023). The FL training process is inherently decentralized and often takes place in less controlled settings compared to data centers, posing unique challenges distinct from those in fully controlled environments.
In this thesis, we identify five key challenges in Federated Learning and propose novel approaches to address them. These challenges arise from the heterogeneity of data and devices, communication issues, and privacy concerns for clients in FL training. Moreover, even well-established theoretical advances in FL require diverse forms of practical implementation to enhance their real-world applicability.
Our contributions advance FL algorithms and systems, bridging theoretical advancements and practical implementations. More broadly, our work serves as a guide for researchers navigating the complexities of translating theoretical methods into efficient real-world implementations and software. Additionally, it offers insights into the reverse process of adapting practical implementation aspects back into theoretical algorithm design. This reverse process is particularly intriguing, as the practical perspective compels us to examine the underlying mechanics and flexibilities of algorithms more deeply, often uncovering new dimensions of the algorithms under study
Design and synthesis of multitarget triazinoindole derivatives with potent antiproliferative activity: Targeting EGFR, SIRT1, MDM2, Hsp90, PI3K, p53, and caspase-9.
No full textDriven by the urgent need for novel anticancer agents capable of overcoming limitations associated with conventional therapies, a new series of Benzyl-5H-[1,2,4]triazino[5,6-b]indoles derivatives bearing flexible pyrazole or pyrazoline moieties were designed, synthesized, and evaluated for antiproliferative efficacy. Most of the synthesized compounds demonstrated notable cytotoxicity compared to the reference compound inauhzin (INZ). Specifically, compounds 4, 10, 11, and 12 exhibited potent activity against A549 lung cancer cells, with IC50 values of 7.39 μM, 3.17 μM, 0.82 μM, and 5.38 μM, respectively, outperforming INZ (IC50 = 8.97 μM). Against Caco-2 colorectal cancer cells, compounds 4, 5, and 10 displayed IC50 values of 2.35 μM, 3.28 μM, and 2.63 μM, respectively, compared to INZ (IC50 = 3.64 μM). To elucidate the potential mechanisms underlying the anticancer activity of the most active compounds, a comprehensive set of assays was conducted, including cell cycle analysis, apoptosis evaluation, and quantification of key molecular targets such as EGFR, SIRT1, MDM2, Hsp90, PI3K, p53, and caspase-9. These biomarkers are intricately interconnected within cellular signaling pathways, whereby inhibition of EGFR and PI3K suppresses proliferative signaling, downregulation of SIRT1 and MDM2 leads to p53 activation, and subsequent induction of caspase-9-mediated apoptosis and cell cycle arrest. Notably, most of the tested compounds demonstrated promising results in modulating these targets, which supports their potential as effective anticancer agents. Additionally, selected compounds showed excellent binding to the target sites when docked into the active domains of EGFR, Hsp90, PI3K, SIRT1, and MDM2
HelioFill: Diffusion-Based Model for EUV Reconstruction of the Solar Farside
No full textThe loss of STEREO-B in 2014 created a persistent blind spot in Extreme Ultraviolet (EUV) imaging of the solar farside. We present HelioFill, to the authors' knowledge, the first denoising-diffusion inpainting model that restores full-Sun EUV coverage by synthesizing the STEREO-B sector from Earth-side (SDO) and STEREO-A views. Trained on full-Sun maps from 2011-2014 (when SDO+STEREO-A+B provided 360 degrees coverage), HelioFill couples a latent diffusion backbone with domain-specific additions: spectral gating, confidence weighting, and auxiliary regularizers, to produce operationally suitable 304 Angstrom reconstructions. On held-out data, the model preserves the observed hemisphere with mean SSIM 0.871 and mean PSNR 25.56 dB, while reconstructing the masked hemisphere with mean SSIM 0.801 and mean PSNR 17.41 dB and reducing boundary error by approximately 21 percent (Seam L2) compared to a state-of-the-art diffusion inpainting model. The generated maps maintain cross-limb continuity and coronal morphology (loops, active regions, and coronal-hole boundaries), supporting synoptic products and cleaner inner-boundary conditions for coronal/heliospheric models. By filling observational gaps with observationally consistent EUV emission, HelioFill maintains continuity of full-Sun monitoring and complements helioseismic farside detections, illustrating how diffusion models can extend the effective utility of existing solar imaging assets for space-weather operations.Firas Ben Ameur acknowledges support from the KAUST Global Fellowship Program (Award No. RFS-KGFP20236044). Rayan Dhib acknowledges support from the FWO SB PhD fellowship (No. 1S66825N). For computational resources, this research used the Ibex cluster, managed by the Supercomputing Core Laboratory at KAUST. This work was funded by KAUST via the student visiting program (Grant No. BAS/1/1663-01-01). Stefaan Poedts is funded by the European Union (ERC, Open SESAME, No. 101141362)
Molecular Basis for Catalysis and Regulation of the Strigolactone Catabolic Enzyme CXE15
No full textStrigolactones (SLs) are pivotal plant hormones involved in developmental, physiological, and adaptive processes. SLs also facilitate symbiosis with arbuscular mycorrhizal fungi and trigger germination of root parasitic Striga plants. The carboxylesterase CXE15, recently identified as the SL catabolic enzyme in Arabidopsis thaliana, plays a crucial role in regulating SL levels. Our study elucidates the structural and regulatory mechanisms of CXE15. We present four crystal structures capturing the conformational dynamics of CXE15, revealing a unique N-terminal extension (Nt) that transitions from a β-sheet in monomers to an intertwined helical structure in dimers. Only the dimeric form is catalytically active, as it forms a hydrophobic cavity for SLs between its two active sites. The moderate dimerisation affinity allows for genetic regulation through protein expression levels. Additionally, we identify an environment-controlled regulation mechanism. Under oxidising conditions, a disulphide bond forms between Cys14 of the two monomers, blocking the active site and inhibiting SL cleavage. This redox-sensitive inhibition of SL catabolism, triggered by reactive oxygen species (ROS) in response to abiotic stress, suggests a mechanism for maintaining high SL levels under beneficial conditions. Our findings provide molecular insights into the regulation of SL homeostasis and catabolism under stress conditions.This research was funded by King Abdullah University of Science and Technology (KAUST) through the baseline fund to S.A.B., and S.T.A. We acknowledge SOLEIL for provision of synchrotron radiation facilities and would like to thank A. Thompson, P. Legrand, S. Sirigu, M. Savko, B. Shepard and A. Thureau for assistance in using the beamlines PROXIMA 1, PROXIMA 2 A, and SWING (Proposal IDs:20210195, 20210932, 20220373, 20231539). For computer time, this research used the resources of the KAUST Supercomputing Laboratory, and experimental research was supported by the Bioscience Core Lab and the Imaging and Characterization Core Lab at King Abdullah University of Science & Technology (KAUST) in Thuwal, Saudi Arabia. We would like to thank the KAUST ACL proteomics core lab for the mass spectrometry analysis
LegoACE: Autoregressive Construction Engine for Expressive LEGO® Assemblies
No full textAutomated LEGO® design is challenging due to the extensive variety of LEGO® brick types and the necessity of constructing semantically meaningful models from individually meaningless components. Current automatic LEGO® generation methods face two key challenges: i) They typically rely on explicit modeling of brick connectivity to ensure structural validity. However, this requires extensive manual annotation, which is labor-intensive as the variety of LEGO® primitives increases. This limits training data diversity, restricting the variety of LEGO® bricks that can be effectively utilized. ii) To facilitate learning within neural networks, current methods often employ either volume or text-based descriptions to represent LEGO® models. However, volumetric representations are computationally expensive and hamper large-scale generative training, while text-based approaches rely on large language models and dedicated text-to-brick mapping rules, introducing a semantic gap between language tokens and 3D brick structures
Kinematic decomposition of volumetric particle tracking velocimetry data
No full textThis paper describes a comprehensive kinematic decomposition of unstructured Lagrangian data from volumetric particle tracking velocimetry measurements. The method uses particle location data at an arbitrary time t and calculates linear affine mappings at a later time t+dt. The transformation produces the full velocity gradient tensor, which can then be further analyzed to identify the four types of fluid motion (i.e., translation, rotation, dilatation, and shear) without using spatial derivatives. The methodology provides insights into the underlying kinematics and facilitates the identification of coherent structures using, for example, the Q-criterion, within the flow without resorting to numerical differentiation or data assimilation methods. The method is first validated using analytical solutions and direct numerical simulations and then applied to experimental subsonic jet measurements. The method’s accuracy is discussed, and leading-order error sources are presented.We thank Zhao Pan and Lanyu Li for their RBF-QR codes. We gratefully acknowledge support from ONR (Grant Numbers N00014-21-1-2454 & N00014-21-1-2293), monitored by Dr. Leighton Meyers
NiOx gate oxide for enhanced thermal stability of threshold voltage in GaN MIS-HEMTs up to 400 °C
No full textThis study demonstrates the high-temperature operation of AlGaN/GaN metal–insulator–semiconductor high electron mobility transistors (MIS-HEMTs) based on nickel oxide (NiOx) as an intermediate gate oxide, achieving stable performance up to 400 °C. Compared to the control sample with only an SiNx gate dielectric, the proposed device exhibited significant improvements: (1) enhanced thermally induced VTH stability, (2) a flat transconductance (gm) curve indicating improved linearity, and (3) lower and stable drain-to-source saturation voltage (VDS,sat). Notably, the ΔVTH shift for D-mode MIS-HEMT with NiOx was effectively reduced to ∼+0.4 V, compared to ∼−1.4 V in the control sample, over a temperature range from 25 to 400 °C. This improvement is attributed to hole carrier generation in the NiOx layer, which increases the depletion region and stabilizes the stored charge underneath the gate at high temperatures. This work demonstrates that the NiOx gate oxide layer significantly enhances VTH stability and linearity in GaN MIS-HEMT, ensuring reliable and stable device operation at high temperatures.The authors thank Venkatesh Singaravelu [Nanofabrication Corelab at King Abdullah University of Science and Technology
(KAUST)] for his thoughtful discussion. The authors are grateful for KAUST Nanofabrication Corelab and funding support of the
Baseline Fund BAS/1/1664-01-01, the Near-term Grand Challenge Fund REI/1/4999-01-01, and the Impact Acceleration Fund REI/1/
5124-01-01
Promoting oxygen reduction reaction kinetics through manipulating electron redistribution in CoP/Cu<sub>3</sub>P@NC for aqueous/flexible Zn-air batteries
No full textZinc–air batteries (ZABs) are considered a promising energy storage technology due to their high energy density and environmental friendliness. However, the development of efficient and durable oxygen reduction reaction (ORR) catalysts remains a challenge. Herein, we report the synthesis of a highly efficient CoP/Cu3P@NC catalyst using a Zn-MOF template, which was transformed into N- and C-doped bimetallic phosphides via high-temperature phosphating. The CoP/Cu3P@NC-based ZAB exhibits remarkable performance with an open-circuit voltage of 1.50 V, a peak power density of 215 mW cm−2, and a specific capacity of 691 mA h gzn−1, outperforming conventional Pt/C-based ZABs. The catalyst maintained 93.5% of its initial activity after 300 h of cycling, demonstrating its excellent long-term stability. Furthermore, CoP/Cu3P@NC was applied in flexible ZABs, achieving a power density of 74 mW cm−2 and showing stable performance under various bending conditions. The superior performance is attributed to the synergistic effects of Co and Cu, optimized structural properties, and high porosity, enhancing mass transfer and oxygen activation. These results suggest that CoP/Cu3P@NC is a highly promising ORR catalyst for next-generation ZABs, offering both high efficiency and durability in flexible and conventional energy storage applications.This work has been supported by the National Natural Science Foundation of China (no. 52363028, 21965005), the Natural
Science Foundation of Guangxi Province\ (2021GXNSFAA076001, 2018GXNSFAA294077), the Guangxi Technology Base and Talent Subject (GUIKE AD23023004, GUIKE AD20297039), Project 202410602052 supported by the National Training Program of Innovation and
Entrepreneurship for Undergraduates, and the Cultivation Program for Outstanding Graduate Theses (EDT2024007)
Time-resolved Speciation of Shock-heated H<sub>2</sub>/NO and H<sub>2</sub>/CO<sub>2</sub> Mixtures
No full textThis study investigates hydrogen oxidation chemistry using NO and CO2 as oxygen sources over a wide temperature range (1600-4600 K) relevant to detonation engines and pressures (0.5-1.5 bar) in a low-pressure shock tube. A laser absorption spectroscopy-based diagnostic platform was developed to measure time-resolved mole fractions of critical species, including NO, CO, CO2, H2O, and OH. The experimental data was used to validate and refine kinetic models, providing new insights into the reaction pathways and temperature dependencies of the H-2-NO and H-2-CO2 systems under extreme conditions. The findings contribute to bridging gaps in hydrogen oxidation chemistry, optimizing combustion processes, and advancing cleaner engine technologies. Future work will aim to further refine these models, particularly for the H-2-CO2 system, addressing gaps in existing literature and enhancing predictive capabilities.This work is financially supported by the Impact Acceleration Fund provided by King Abdullah University of Science and Technology (KAUST), Kingdom of Saudi Arabia