187037 research outputs found
Sort by
Bandwidth Parameterized by Cluster Vertex Deletion Number
No full textGiven a graph G and an integer b, Bandwidth asks whether there exists a bijection π from V(G) to {1,…,|V(G)|} such that max{u,v}∈E(G)|π(u)-π(v)|≤b. This is a classical NP-complete problem, known to remain NP-complete even on very restricted classes of graphs, such as trees of maximum degree 3 and caterpillars of hair length 3. In the realm of parameterized complexity, these results imply that the problem remains NP-hard on graphs of bounded pathwidth, while it is additionally known to be W[1]-hard when parameterized by the tree-depth of the input graph. In contrast, the problem does become FPT when parameterized by the vertex cover number. In this paper we make progress in understanding the parameterized (in)tractability of Bandwidth. We first show that it is FPT when parameterized by the cluster vertex deletion number cvd plus the clique number ω, thus significantly strengthening the previously mentioned result for vertex cover number. On the other hand, we show that Bandwidth is W[1]-hard when parameterized only by cvd. Our results develop and generalize some of the methods of argumentation of the previous results and narrow some of the complexity gaps.
STAR: Sea Turtle Basic Activity Recognizer Network via Efficient Transformer
No full textSea turtles are endemic to several islands and are crucial in maintaining ecosystem balance. However, they face significant threats from poaching, extreme weather conditions, and abnormal activities. Monitoring and protecting their habitats has become a key focus in blue economy research, a priority program in several developed nations. Observing sea turtle behavior not only aids in conservation efforts but also supports the development of behavioral analysis that can serve as metrics for further research. In this work, we propose STAR, a basic turtle activity recognition system, employing an efficient deep learning approach. This system categorizes simple activities such as swimming, eating, hiding, resting, and other distinct behaviors. An effective network is offered by utilizing enhanced modules to discriminate activity features involving sequential frames. In addition, a lightweight transformer is presented as a novel attentive module that improves the feature extraction ability. Another contribution offers a new video dataset of sea turtle activity recognition captured underwater from various turtle poses. Several extensive experiments analyzed the performances of essential modules and achieved satisfactory results compared to other lightweight models. The efficiency of the STAR model shows that it obtained fast data processing speed without abundant computational resources in single-frame analysis. Applying this recognition system requires sea turtle detection to capture the area of interest in the beginning process.
Middle Interlayer Engineered Ferroelectric NAND Flash Overcoming Reliability and Stability Bottlenecks for Next-Generation High-Density Storage Systems
No full textMultilevel storage and low-voltage operation position ferroelectric transistors as promising candidates for next-generation nonvolatile memory. Among them, gate-injection-type ferroelectric transistors offer improved vertical scalability and power efficiency for three-dimensional (3D) NAND flash. However, their intricate interplay between polarization switching and charge trapping complicates systematic understanding of degradation mechanisms, limiting strategies to improve reliability and stability. Here, gate stack engineering incorporating middle interlayers within HfZrOx matrix is presented to modulate polarization dynamics, strengthening the coupling of dual mechanisms and overcoming long-standing reliability and stability bottlenecks in ferroelectric NAND operation. This approach achieves a memory window up to 11 V, an operating voltage below 18 V, triple-level-cell retention beyond 10 years, disturbance immunity, and 54% reduced threshold voltage variability. A 20% reduction in program voltage compared to conventional NAND enables aggressive vertical scaling, leading to 25% higher bit-density. Furthermore, analytical modeling provides insights into gate stack optimization. These findings establish ferroelectric NAND as a scalable, energy-efficient solution for next-generation storage.
Physics-based landslide susceptibility machine learning model for mountainous solar power plants
No full textBetween 2018 and 2020, several landslides occurred on solar power plants constructed on hillsides after extreme rainfall in South Korea. This study presents a physics-informed machine-learning model to conduct real-time physically-based landslide susceptibility assessment on solar panels installed on mountains. Through a statistical filtering process, fourteen landslide triggering factors related to the topography, soil geotechnical properties, soil hydrological properties, meteorological effects, and solar panels model were selected. While accounting for the presence of solar panels, 136,262 numerical simulations of rainfall infiltration transient seepage and slope stability analyses were performed. Among three machine learning models (random forest, support vector regression, and multi-layer perceptron) developed from numerical simulation data points, the multi-layer perceptron (MLP) model showed the highest prediction accuracy (R2 = 96% and mean square error = 0.001) without indicating overfitting. The sensitivity analysis on the developed MLP model indicated that the soil strength properties strongly influenced the factors of safety (FOS), but hydraulic properties showed a relatively small impact. The developed MLP model was applied to a solar power plant in Jangsu-gun, Jeonbuk-do, South Korea, to verify the model’s predictability and showed the applicability of the developed MLP model for physically-based landslide susceptibility assessment. Furthermore, the Jangsu-gun example demonstrated the potentiality of the MLP model for a landslide early warning system (LEWS) for mountainous solar power plants thanks to its fast computational speed at predicting real-time FOS.
Combined application of deep learning and conventional computer vision for kidney ultrasound image classification in chronic kidney disease: preliminary study
No full textPurpose: This study evaluates the feasibility of combining deep learning (DL) and conventional computer vision techniques to classify kidney ultrasound (US) images for the presence or absence of chronic kidney disease (CKD). Methods: A retrospective analysis was conducted on 258 kidneys (124 normal and 134 with CKD). A DL model was trained using midsagittal US images of the right kidney and corresponding contour maps to automate measurements of parenchymal thickness and parenchyma-to-sinus ratios. These features were integrated with a convolutional neural network for classification. The ground truth was determined based on clinical CKD diagnosis and laboratory data. Results: The combined DL and conventional feature extraction model achieved an accuracy of 82%, with a specificity of 93% and a negative predictive value of 97%. This approach outperformed models that relied solely on raw US images using DL, which achieved an accuracy of 64%. The inclusion of contour-based parenchymal measurements enhanced classification performance. Conclusion: The integration of DL with automated feature extraction enables accurate classification of CKD using minimal user input. This proof-of-concept study highlights the potential of combining artificial intelligence-driven analysis with traditional metrics to serve as a noninvasive adjunct for CKD diagnosis and monitoring.
Structure Is All You Need: Structural Representation Learning on Hyper-Relational Knowledge Graphs
No full textDeepECtransformer: AI-based deep learning model for enzyme function annotation and discovery
No full text
Closing the Modality Gap: Integrating LLMs with LiDAR for 3D Object Detection and Object-level Understanding
No full textWIDEBAND EXTREMALY LARGE ARRAY ANTENNA SYSTEM AND OFDM-BASED HYBRID BEAMFORMING METHOD FOR HYBRID-FIELD INTERFERENCE CONTROL OF THE SAME
No full text본 개시는 광대역 거대 배열 안테나 시스템 및 그의 필드 간 간섭 제어를 위한 OFDM 기반 하이브리드 빔포밍 방법에 관한 것이다. 본 개시의 광대역 거대 배열 안테나 시스템은 복수의 송신 안테나들을 갖는 송신기, 및 송신기와 통신 가능하며, 복수의 수신 안테나들을 갖는 수신기를 포함하고, 송신기와 수신기 사이의 전체 부반송파들에 대하여 공통의 아날로그 빔포밍 행렬을 설계하는 제 1 단계, 및 부반송파들의 각각에 대한 기저대역 빔포밍 행렬을 설계하는 제 2 단계를 포함하여 하이브리드 빔포밍 설계를 수행하도록 구성될 수 있다