67639 research outputs found
Sort by
I2M2Net: Inter/Intra-modal Feature Masking Self-distillation for Incomplete Multimodal Skin Lesion Diagnosis
No full textMultimodal learning has demonstrated promising advantages over single-modal approaches in the diagnosis of skin lesions. However, these methods often suffer from significant accuracy degradation when encountering missing modalities, hindering their clinical deployment. In this paper, we introduce a novel and effective framework, I2M2Net, for incomplete multimodal learning, focusing on adaptively and progressively mining knowledge about modal feature-aware combinations. Specifically, one branch conducts normal classification using the original complete multimodal features extracted by heterogeneous modal encoders, while another branch shares the same structures and weights, designed to perform self-distillation with masked modality combinations. These combinations are imposed on the complete features using two masking strategies simultaneously: 1) random dropout of modality (i.e. inter-modal feature masking) to simulate different missing modality combinations and foster combination-invariant dependencies, and 2) randomly mask patches of the remaining modal features (i.e. intra-modal feature masking) to promote combination-specific representations. Additionally, we design a combination-based curriculum learning (CCL) algorithm to identify weak combinations and progressively guide our network to facilitate incomplete modality learning on challenging combinations. This is achieved by adaptively adjusting the probabilities of masking based on the consistency between the complete combination and other combinations. Experimental results on the multimodal skin disease dataset Derm7pt demonstrate that our method outperforms other state-of-the-art approaches
Bending the curve of land degradation to achieve global environmental goals
No full textLand has a vital role in sustaining human communities, nurturing diverse ecosystems and regulating the climate of our planet. As such, current rates of land degradation pose a major environmental and socioeconomic threat, driving climate change, biodiversity loss and social crises. Preventing and reversing land degradation are key objectives of the United Nations Convention to Combat Desertification and are also fundamental for the other two Rio Conventions: the United Nations Framework Convention on Climate Change and the Convention on Biological Diversity. Here we argue that the targets of these conventions can only be met by ‘bending the curve’ of land degradation and that transforming food systems is fundamental for doing so. We showcase multiple actions for tackling land degradation that also yield climate and biodiversity benefits while fostering sustainable food systems that contribute to avoiding the risk of a global food crisis. We also propose ambitious 2050 targets for the three Rio Conventions related to land and food systems. Finally, we urge collective action to acknowledge the pivotal role of land in achieving the goals of the Rio Conventions and to embed food systems within intergovernmental agreements, enabling decisive progress on the complex and interconnected global crises that we face.We thank the Office of the Vice President for Research at the King Abdullah University of Science and Technology (KAUST) for funding the writing workshop leading to this Review; S. Díaz for comments and suggestions on a previous version of this manuscript; and I. Gromicho and H. Hwang for preparing Fig. 2. F.T.M., C.M.D., M.F.M., Y.W., T.T., M.M., E.G. and H.E.B. acknowledge support from KAUST. B.F. acknowledges support from NSFC (no. 42430505). The views expressed here are those of the authors and do not necessarily reflect the views of the United Nations
Rapid identification of induced seismicity using deep learning in West Texas
No full textTimely identification of the triggering mechanism behind the observed seismicity in areas with multiple overlapping human activities is an important research topic that can facilitate effective measures to mitigate the seismic hazard. This task is particularly challenging when dealing with delayed operational data, uncertain focal depths, or uneven seismic monitoring coverage. Here, we propose a deep learning (DL) framework to identify which human activity triggered a certain earthquake in near real-time using only seismic waveforms as input. We use an advanced architecture, the compact convolutional transformer (CCT), to extract high-level abstract features from the three-component seismograms and then use an advanced capsule neural network to link the induced seismicity in West Texas with three potential causal factors, i.e., hydraulic fracturing (HF), shallow saltwater disposal (SWDsh), or deep saltwater disposal (SWDdp). The training data was prepared based on an established probabilistic approach that combined physics-based principles with both real and reshuffled injection data to hindcast past seismicity rates. In the end, each activity was assigned a confidence level for association at the 5 km spatial scale. Even though the training data include only 981 events, we obtain over 90% accuracy for all three causal factors for both the single- and multi-station versions of the model.The authors acknowledge there are no conflicts of interest recorded
Nonlinear Metasurfaces for Completed Control of Amplitude, Phase, and Polarization in Broadband Terahertz Generation.
No full textTerahertz (THz) generation is a crucial initial step in THz applications. However, the current THz sources face challenges in fully controlling the propagation properties of generated THz waves without the use of external devices. This limitation leads to bulky systems with unavoidable insertion losses and bandwidth constraints. To overcome these challenges and facilitate compact and versatile THz applications, a novel approach using nonlinear metasurfaces is proposed to control the amplitude, phase, and polarization of broadband THz waves directly and simultaneously at the emission stage. The basic design features an elaborated coupling-controlled chiral meta-atom, providing adjustable chirality and allowing an independent amplitude and phase control strategy under a circularly polarized (CP) pump. Furthermore, the polarization state of emitted THz wave can be arbitrarily customized by designing the superposition of the generated left-handed circularly polarized (LCP) and right-handed circularly polarized (RCP) components. This control is linearly predictable, eliminating the need for complex nonlinear simulations and interleaved supercell arrangements. The effectiveness of this method is demonstrated by experimentally generating two types of unique vectorial THz fields: spatial-polarization separable and nonseparable states. The proposed approach significantly enhances the capabilities of nonlinear metasurfaces, paving the way for versatile THz generation devices
From Wild Halophyte to Future Crop: Genomic Insights into Salicornia
No full textAmid escalating water scarcity and groundwater depletion, halophytes such as Salicornia offer sustainable solution for saltwater-based agriculture. Salicornia, a genus in the Amaranthaceae family, is valued for its edible succulent stems and oil-rich seeds, yet its genetic improvement has been constrained by limited genomic resources. We present chromosome-scale, annotated genome assemblies for six Salicornia species and identified four distinct subgenomes within the genus. Comparative transcriptomic analyses reveals genes associated with high shoot Na+ accumulation, a key salt tolerance mechanism. Phylogenetic clustering and population structure analyses of a globally representative resequencing panel highlight interspecies relationships and reveal misclassification at both the accession and species levels. The curated panel represents a robust resource for genetic studies and breeding. These genomic resources lay the foundation for the domestication of Salicornia as a climate-resilient crop for saltwater agricultural systems.Research reported in this publication was supported by the King Abdullah University of Science and Technology (KAUST). For computer time, this research used Shaheen III and the Ibex cluster managed by the Supercomputing Core Laboratory at King Abdullah University of Science & Technology (KAUST). The authors would like to thank Dr Mohammed Shahid (International Centre for Biosaline Agriculture, UAE) for sharing Salicornia europaea (UAQ and RAK) seeds and meta-data and Dr Muppala P. Reddy (former KAUST staff) for sharing Salicornia brachiata. Salicornia bigelovii seeds were kindly provided by Dr. E. Glenn of the Environmental Research Laboratory, University of Arizona, Tucson, USA. Salicornia europaea seeds were collected in the Dead Sea area and were kindly provided by Dr. Moshe Sagi of the Blaustein Institute for Desert Research (BIDR), Israel. The authors thank Peri Coleman (Delta Environmental Consultant) with her support assessing the Australian Tecticornia spp. and Ehsan Tavakkoli (University of Adelaide) for his help with assessment of local soil and water samples. The authors thank Dinara Utarbayeva for her technical support with early propagation efforts of various Salicornia species and Juan Pablo Arciniegas Vega for technical support with early molecular biology experiments. The authors would like to thank DASARANG company for their assistance with the collection of Salicornia samples in South Korea. The authors would like to thank Dovetail for their reference genome sequencing service support and KAUST Bioscience Core Labs for their support with HMW DNA sample quality analysis
Operational Current Amplifier Based Quadrature Oscillators Family
No full textThis paper investigates the application of the operational current amplifier (OCA) in the design of quadrature sinusoidal oscillators. There are numerous quadrature sinusoidal oscillators based on the three fundamental amplifiers namely the operational amplifier (OpAmp), the operational transconductance amplifier (OTA), and the operational transresistance amplifier (OTRA); however, their counterparts based on the OCA have not been reported. In this paper, several possible quadrature oscillator structures based on the OCA are systematically developed from their voltage mode active-RC counterparts using adjoint network theorem. Consequently, it is shown that the resulted oscillators do not necessarily preserve the attractive features of the voltage mode quadrature oscillators; namely providing explicit quadrature current output signals with equal amplitudes. When attempting to retain these features, it is found that the minimum OCAs needed is three. Analyzing this issue has paved the way for developing a novel OCA based integrator that promotes the invention of a new quadrature oscillator using only two OCAs. Simulation results obtained from a standard 150 nm CMOS process are provided. It is demonstrated that the proposed oscillator consumes approximately 350 μW while working in the 3 MHz range. Measurement results obtained from a prototype implemented using CMOS OCA ICs also confirms the proposed theory
Comparative study of different engine knock metrics for bracketing the octane number of fuels
No full textThis study presents a comparative analysis of different engine knock metrics used to evaluate the octane number (ON) of fuels in a Cooperative Fuel Research (CFR) engine. The knock metrics examined include knock intensity 20 (KI20), the maximum amplitude of pressure oscillations (MAPO), the maximum pressure rise rate (MPRR), the cumulative knock intensity (CKI), and the wavelet decomposition energy (WDE). Modified versions of standard CFR engine tests were conducted using both liquid and gaseous fuels, covering a range of research octane number (RON) from 60 to 100. The knock data were collected using both a detonation meter and an in-cylinder pressure transducer to compare traditional and pressure-based knock measurement methods. Results indicate that of the metrics investigated, MPRR is the most effective for bracketing octane numbers, showing higher validity and a closer resemblance to knockmeter readings compared to the others analyzed. Furthermore, the study explores the knock resistance of hydrogen, revealing discrepancies with standard RON evaluations. The findings of this work indicates that hydrogen’s RON, evaluated based on MPRR, falls within the range of 98–100. The results provide valuable insights for improving knock measurement accuracy, especially when evaluating fuels with high knock resistance, and for optimizing modern engine designs to meet emerging fuel standards.Funding and support for this work was provided by the King Abdullah University of Science and Technology and its Clean Energy Research Platform
Broadband water wave reflector with customisable frequency range enabled by floating metaplates
No full textResearch on water wave metamaterials based on local resonance has advanced rapidly. However, their application to floating structures for controlling surface gravity waves remains underexplored. In this work, we introduce the floating metaplate, a periodic array of resonators on a floating plate that leverages locally resonant bandgaps to effectively manipulate surface gravity waves. We employ the eigenfunction matching method combined with Bloch’s theorem to solve the wave–structure interaction problem and obtain the band structure of the floating metaplate. An effective model based on averaging is developed, which agrees well with the results of numerical simulation, elucidating the mechanism of bandgap formation. Both frequency- and time-domain simulations demonstrate the floating metaplate’s strong wave attenuation capabilities. Furthermore, by incorporating a gradient in the resonant frequencies of the resonators, we achieve the rainbow trapping effect, where waves of different frequencies are reflected at distinct locations. This enables the design of a broadband wave reflector with a tuneable operation frequency range. Our findings may lead to promising applications in coastal protection, wave energy harvesting and the design of resilient offshore renewable energy systems.The authors gratefully acknowledge the support by King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research (OSR) under Grant No. ORFS- CRG11-2022-5055, as well as KAUST Baseline Research Fund BAS/1/1626-01-01
Theory of Intermediate Twinning and Spontaneous Polarization in Ferroelectric Potassium Sodium Niobate
No full textPotassium sodium niobate is considered a prominent material system as a substitute for lead-containing ferroelectric materials. It exhibits first-order phase transformations and ferroelectricity with potential applications ranging from energy conversion to innovative cooling technologies, thereby addressing important societal challenges. However, a major obstacle in the application of potassium sodium niobate is its multi-scale heterogeneity and the lack of understanding of its phase transition pathway and microstructure. This can be seen from the findings of Pop-Ghe et al. (Ceram Int 47(14):20579–20585, 2021, https://doi.org/10.1016/j.ceramint.2021.04.067) which also reveal the occurrence of a phenomenon they term intermediate twinning during the phase transition. Here, we show that intermediate twinning is a consequence of energy minimization. We develop a geometrically nonlinear electroelastic energy function for potassium sodium niobate, including the cubic-tetragonal-orthorhombic transformations and ferroelectricity. The construction of the minimizers is based on compatibility conditions which ensure continuous deformations and pole-free interfaces. These minimizers agree with the experimental observations, including laminates between tetragonal variants under the cubic to tetragonal transformation, crossing twins under the tetragonal to orthorhombic transformation, intermediate twinning and spontaneous polarization. This shows how the full nonlinear electroelastic model provides a powerful tool in understanding, exploring, and tailoring the electromechanical properties of complex ferroelectric ceramics.Open access publishing provided by King Abdullah University of Science and Technology (KAUST). The work of GG, P-L P-G, and RDJ was partly supported by a Vannevar Bush Faculty Fellowship (Grant No. N00014-19-1-2623). The initial work of P-L P-G and EQ was funded by the DFG via the Reinhart Koselleck Project “Crystallographically compatible ceramic shape memory materials” (Project Number 313454214)
Cooperative Grasping for Collective Object Transport in Constrained Environments
No full textWe propose a novel framework for decision-making in cooperative grasping for two-robot object transport in constrained environments. The core of the framework is a Conditional Embedding (CE) model consisting of two neural networks that map grasp configuration information into an embedding space. The resulting embedding vectors are then used to identify feasible grasp configurations that allow two robots to collaboratively transport an object. To ensure generalizability across diverse environments and object geometries, the neural networks are trained on a dataset comprising a range of environment maps and object shapes. We employ a supervised learning approach with negative sampling to ensure that the learned embeddings effectively distinguish between feasible and infeasible grasp configurations. Evaluation results across a wide range of environments and objects in simulations demonstrate the model's ability to reliably identify feasible grasp configurations. We further validate the framework through experiments on a physical robotic platform, confirming its practical applicability.The work was supported by funding from King Abdullah University of Science and Technology (KAUST), and the SDAIA-KAUST Center of Excellence in Data Science and Artificial Intelligence (SDAIA-KAUST AI)