2023 연구성과 (25 / 285)

※ 컨트롤 + 클릭으로 열별 다중 정렬 가능합니다.
Excel 다운로드
WoS SCOPUS Document Type Document Title Abstract Authors Affiliation ResearcherID (WoS) AuthorsID (SCOPUS) Author Email(s) Journal Name JCR Abbreviation ISSN eISSN Volume Issue WoS Edition WoS Category JCR Year IF JCR (%) FWCI ⓘ FWCI Update Date WoS Citation SCOPUS Citation Keywords (WoS) KeywordsPlus (WoS) Keywords (SCOPUS) KeywordsPlus (SCOPUS) Language Publication Stage Publication Year Publication Date DOI JCR Link DOI Link WOS Link SCOPUS Link
○ ○ Article ⓘ Hidden Markov Model Based on Logistic Regression ⓘ A hidden Markov model (HMM) is a useful tool for modeling dependent heterogeneous phenomena. It can be used to find factors that affect real-world events, even when those factors cannot be directly observed. HMMs differ from traditional methods by using state variables and mixture distributions to model the hidden states. This allows HMMs to find relationships between variables even when the variables cannot be directly observed. HMM can be extended, allowing the transition probabilities to depend on covariates. This makes HMMs more flexible and powerful, as they can be used to model a wider range of sequential data. Modeling covariates in a hidden Markov model is particularly difficult when the dimension of the state variable is large. To avoid these difficulties, Markovian properties are achieved by implanting the previous state variables to the logistic regression model. We apply the proposed method to find the factors that affect the hidden state of matsutake mushroom growth, in which it is hard to find covariates that directly affect matsutake mushroom growth in Korea. We believe that this method can be used to identify factors that are difficult to find using traditional methods. ⓘ Lee, Byeongheon; Park, Joowon; Kim, Yongku ⓘ Kyungpook Natl Univ, Dept Stat, Daegu 41566, South Korea; Kyungpook Natl Univ, Sch Forest Sci & Landscape Architecture, Daegu 41566, South Korea ⓘ Ferreira, Manuel Alberto M./N-3136-2013 ⓘ 58665453300; 55791550500; 47962102500 ⓘ gijo0104@naver.com;joowon72@knu.ac.kr;kim.1252@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 20 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.25 2025-06-25 ⓘ 1 1 Bayesian analysis; hidden Markov model; hierarcical modeling; logistic regression ⓘ LONGITUDINAL DATA ⓘ Bayesian analysis; hidden Markov model; hierarcical modeling; logistic regression ⓘ ⓘ English 2023 2023-10 10.3390/math11204396 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Homogeneous Li deposition guided by ultra-thin lithiophilic layer for highly stable anode-free batteries ⓘ Li-metal batteries (LMBs) are intensively studied to keep up with the growing demand for sustainable and high-capacity energy storage devices. However, the practical implementation of LMBs is still challenging owing to the catastrophic side effects associated with the growth of dendritic Li and inferior Coulombic efficiency. To enhance the long-term electrochemical stability and high-rate performance of LMBs, it is crucial to control the morphology of Li deposition over the current collectors. Herein, we propose surface-modified current collectors to investigate how a lithiatable layer affects the morphology of Li deposition and thus contributes to the stable electrochemical performance of LMBs at high current densities. The lithiatable layer improves the electrolyte wetting ability, which efficiently diminishes the interfacial resistance between electrode and electrolyte. Moreover, the lithiatable layer induces homogeneous Li nucleation, consequently leading to the uniform Li deposition over the surface of current collectors. Due to these synergistic effects, the anode-free cells with surface-modified current collectors have achieved excellent cycle stability in comparison to that with conven-tional Cu current collector. ⓘ Kim, Junghwan; Lee, Gyeong Ryul; Chung, Roy Byung Kyu; Kim, Patrick Joohyun; Choi, Junghyun ⓘ Korea Inst Ceram Engn & Technol, Energy Storage Mat Ctr, Jinju 52851, South Korea; Kyungpook Natl Univ, Dept Appl Chem, Daegu 41566, South Korea; Kyungpook Natl Univ, Sch Mat Sci & Engn, Daegu 41566, South Korea ⓘ ⓘ 55966669500; 57192425717; 16642183100; 57195611779; 59883103900 ⓘ roy.b.chung@knu.ac.kr;pjkim@knu.ac.kr;jchoi@kicet.re.kr; ⓘ ENERGY STORAGE MATERIALS ⓘ ENERGY STORAGE MATER ⓘ 2405-8297 2405-8289 61 SCIE CHEMISTRY, PHYSICAL;MATERIALS SCIENCE, MULTIDISCIPLINARY;NANOSCIENCE & NANOTECHNOLOGY ⓘ 2023 18.9 4.2 1.54 2025-06-25 ⓘ 16 16 Li metal batteries; Anode-free batteries; Surface-modified current collector; Uniform Li deposition; Atomic Layer Deposition ⓘ LITHIUM; WETTABILITY; CHALLENGES; SEI ⓘ Anode-free batteries; Atomic Layer Deposition; Li metal batteries; Surface-modified current collector; Uniform Li deposition ⓘ Atomic layer deposition; Electric current collectors; Electrolytes; Morphology; Wetting; Anode-free battery; Atomic-layer deposition; Current-collector; Li deposition; Li metal; Li metal battery; Surface-modified; Surface-modified current collector; Ultra-thin; Uniform li deposition; Anodes ⓘ English 2023 2023-08 10.1016/j.ensm.2023.102899 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Improving the Performance of Object Detection by Preserving Balanced Class Distribution ⓘ Object detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and validation typically originate from public datasets that are well-balanced in terms of the number of objects ascribed to each class in an image. However, in real-world scenarios, handling datasets with much greater class imbalance, i.e., very different numbers of objects for each class, is much more common, and this imbalance may reduce the performance of object detection when predicting unseen test images. In our study, thus, we propose a method that evenly distributes the classes in an image for training and validation, solving the class imbalance problem in object detection. Our proposed method aims to maintain a uniform class distribution through multi-label stratification. We tested our proposed method not only on public datasets that typically exhibit balanced class distribution but also on private datasets that may have imbalanced class distribution. We found that our proposed method was more effective on datasets containing severe imbalance and less data. Our findings indicate that the proposed method can be effectively used on datasets with substantially imbalanced class distribution. ⓘ Lee, Heewon; Ahn, Sangtae ⓘ Kyungpook Natl Univ, Sch Elect Engn, Daegu 41566, South Korea; Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 41566, South Korea ⓘ Ahn, Sangtae/AFQ-7342-2022 ⓘ 58621495000; 55468016100 ⓘ saebuk2000@knu.ac.kr;stahn@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 21 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.74 2025-06-25 ⓘ 2 3 computer vision; object detection; imbalanced class distribution; multi-label stratification ⓘ ⓘ computer vision; imbalanced class distribution; multi-label stratification; object detection ⓘ ⓘ English 2023 2023-11 10.3390/math11214460 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Multi-Layer Decomposition and Synthesis of HDR Images to Improve High-Saturation Boundaries ⓘ Recently, high dynamic range (HDR) imaging has been used in many fields such as display, computer graphics, and digital cameras. Various tone mapping operators (TMOs) are used for effective HDR imaging. TMOs aim to express HDR images without loss of information and natural images on existing display equipment. In this paper, to improve the color distortion that occurs during tone mapping, multi-layer decomposition-based color compensation and global color enhancement of the boundary region are proposed. Multi-layer decomposition is used to preserve the color information of the input image and to find the area where color distortion occurs. Color compensation and enhancement are especially used to improve the color saturation of the border area, which is degraded due to color distortion and tone processing. Color compensation and enhancement are processed in IPT color space with excellent hue linearity to improve effective performance by reducing interference between luminance and chrominance components. The performance of the proposed method was compared to the existing methods using naturalness, distortion, and tone-mapped image quality metrics. The results show that the proposed method is superior to the existing methods. ⓘ Kwon, Hyuk-Ju; Lee, Sung-Hak ⓘ Kyungpook Natl Univ, Sch Elect & Elect Engn, Daegu 702701, South Korea ⓘ ⓘ 55169908300; 7601395661 ⓘ shak2@ee.knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 3 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.74 2025-06-25 ⓘ 3 3 HDR; tone compression; multi-layer decomposition; color compensation; color enhancement; IPT color space ⓘ QUALITY ASSESSMENT; TONE; MODEL; APPEARANCE ⓘ color compensation; color enhancement; HDR; IPT color space; multi-layer decomposition; tone compression ⓘ ⓘ English 2023 2023-02 10.3390/math11030785 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Multi-Task Learning Approach Using Dynamic Hyperparameter for Multi-Exposure Fusion ⓘ High-dynamic-range (HDR) image synthesis is a technology developed to accurately reproduce the actual scene of an image on a display by extending the dynamic range of an image. Multi-exposure fusion (MEF) technology, which synthesizes multiple low-dynamic-range (LDR) images to create an HDR image, has been developed in various ways including pixel-based, patch-based, and deep learning-based methods. Recently, methods to improve the synthesis quality of images using deep-learning-based algorithms have mainly been studied in the field of MEF. Despite the various advantages of deep learning, deep-learning-based methods have a problem in that numerous multi-exposed and ground-truth images are required for training. In this study, we propose a self-supervised learning method that generates and learns reference images based on input images during the training process. In addition, we propose a method to train a deep learning model for an MEF with multiple tasks using dynamic hyperparameters on the loss functions. It enables effective network optimization across multiple tasks and high-quality image synthesis while preserving a simple network architecture. Our learning method applied to the deep learning model shows superior synthesis results compared to other existing deep-learning-based image synthesis algorithms. ⓘ Im, Chan-Gi; Son, Dong-Min; Kwon, Hyuk-Ju; Lee, Sung-Hak ⓘ Kyungpook Natl Univ, Sch Elect & Elect Engn, 80 Daehak Ro, Daegu 702701, South Korea ⓘ ; Son, Dong-Min/LZH-4025-2025 ⓘ 57942669400; 57216612214; 55169908300; 7601395661 ⓘ imchmcgi2@knu.ac.kr;forhollow@knu.ac.kr;olin1223@knu.ac.kr;shak2@ee.knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 7 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.25 2025-06-25 ⓘ 0 1 high dynamic range; multi exposure fusion; image fusion; deep learning ⓘ IMAGES ⓘ deep learning; high dynamic range; image fusion; multi exposure fusion ⓘ ⓘ English 2023 2023-04 10.3390/math11071620 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Optimization-Based Energy Disaggregation: A Constrained Multi-Objective Approach ⓘ Recently, optimization-based energy disaggregation (ED) algorithms have been gaining significance due to their capability to perform disaggregation with minimal information compared to the pattern-based ED algorithms, which demand large amounts of data for training. However, the performances of optimization-based ED algorithms depend on the problem formulation that includes an objective function(s) and/or constraints. In the literature, ED has been formulated as a constrained single-objective problem or an unconstrained multi-objective problem considering disaggregation error, sparsity of state switching, on/off switching, etc. In this work, the ED problem is formulated as a constrained multi-objective problem (CMOP), where the constraints related to the operational characteristics of the devices are included. In addition, the formulated CMOP is solved using a constrained multi-objective evolutionary algorithm (CMOEA). The performance of the proposed formulation is compared with those of three high-performing ED formulations in the literature based on the appliance-level and overall indicators. The results show that the proposed formulation improves both appliance-level and overall ED results. ⓘ Park, Jeewon; Ajani, Oladayo S.; Mallipeddi, Rammohan ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 37224, South Korea ⓘ ; AJANI, Oladayo/HIR-9607-2022; Mallipeddi, Rammohan/AAL-5306-2020 ⓘ 58098750300; 57465126000; 25639919900 ⓘ mallipeddi.ram@gmail.com; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 3 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 1.96 2025-06-25 ⓘ 8 8 energy disaggregation; non-intrusive load monitoring; optimization-based energy disaggregation; constrained multi-objective optimization; evolutionary algorithms ⓘ ALGORITHM ⓘ constrained multi-objective optimization; energy disaggregation; evolutionary algorithms; non-intrusive load monitoring; optimization-based energy disaggregation ⓘ ⓘ English 2023 2023-02 10.3390/math11030563 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Probabilistic Classification Method of Spiking Neural Network Based on Multi-Labeling of Neurons ⓘ Recently, deep learning has exhibited outstanding performance in various fields. Even though artificial intelligence achieves excellent performance, the amount of energy required for computations has increased with its development. Hence, the need for a new energy-efficient computer architecture has emerged, which further leads us to the neuromorphic computer. Although neuromorphic computing exhibits several advantages, such as low-power parallelism, it exhibits lower accuracy than deep learning. Therefore, the major challenge is to improve the accuracy while maintaining the neuromorphic computing-specific energy efficiency. In this paper, we propose a novel method of the inference process that considers the probability that after completing the learning process, a neuron can react to multiple target labels. Our proposed method can achieve improved accuracy while maintaining the hardware-friendly, low-power-parallel processing characteristics of a neuromorphic processor. Furthermore, this method converts the spike counts occurring in the learning process into probabilities. The inference process is conducted to implement the interaction between neurons by considering all the spikes that occur. The inferring circuit is expected to show a significant reduction in hardware cost and can afford an algorithm exhibiting a competitive computing performance. ⓘ Sung, Mingyu; Kim, Jaesoo; Kang, Jae-Mo ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea; Kyungpook Natl Univ, Dept Comp Sci, Daegu 41566, South Korea ⓘ ⓘ 57221328242; 57191684854; 56024930400 ⓘ jmkang@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 5 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0 2025-06-25 ⓘ 0 0 leaky integrate fire neuron; spiking neural network; spike-timing-dependent plasticity ⓘ ⓘ leaky integrate fire neuron; spike-timing-dependent plasticity; spiking neural network ⓘ ⓘ English 2023 2023-03 10.3390/math11051224 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Raindrop-Removal Image Translation Using Target-Mask Network with Attention Module ⓘ Image processing plays a crucial role in improving the performance of models in various fields such as autonomous driving, surveillance cameras, and multimedia. However, capturing ideal images under favorable lighting conditions is not always feasible, particularly in challenging weather conditions such as rain, fog, or snow, which can impede object recognition. This study aims to address this issue by focusing on generating clean images by restoring raindrop-deteriorated images. Our proposed model comprises a raindrop-mask network and a raindrop-removal network. The raindrop-mask network is based on U-Net architecture, which learns the location, shape, and brightness of raindrops. The rain-removal network is a generative adversarial network based on U-Net and comprises two attention modules: the raindrop-mask module and the residual convolution block module. These modules are employed to locate raindrop areas and restore the affected regions. Multiple loss functions are utilized to enhance model performance. The image-quality assessment metrics of proposed method, such as SSIM, PSNR, CEIQ, NIQE, FID, and LPIPS scores, are 0.832, 26.165, 3.351, 2.224, 20.837, and 0.059, respectively. Comparative evaluations against state-of-the-art models demonstrate the superiority of our proposed model based on qualitative and quantitative results. ⓘ Kwon, Hyuk-Ju; Lee, Sung-Hak ⓘ Kyungpook Natl Univ, Sch Elect & Elect Engn, 80 Deahakro, Daegu 41566, South Korea ⓘ ⓘ 55169908300; 7601395661 ⓘ olin1223@knu.ac.kr;shak2@ee.knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 15 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 2.21 2025-06-25 ⓘ 8 9 raindrop removal; U-Net; attention mechanism; generative adversarial network ⓘ GENERATIVE ADVERSARIAL NETWORK ⓘ attention mechanism; generative adversarial network; raindrop removal; U-Net ⓘ ⓘ English 2023 2023-08 10.3390/math11153318 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Redesigning Embedding Layers for Queries, Keys, and Values in Cross-Covariance Image Transformers ⓘ There are several attempts in vision transformers to reduce quadratic time complexity to linear time complexity according to increases in the number of tokens. Cross-covariance image transformers (XCiT) are also one of the techniques utilized to address the issue. However, despite these efforts, the increase in token dimensions still results in quadratic growth in time complexity, and the dimension is a key parameter for achieving superior generalization performance. In this paper, a novel method is proposed to improve the generalization performances of XCiT models without increasing token dimensions. We redesigned the embedding layers of queries, keys, and values, such as separate non-linear embedding (SNE), partially-shared non-linear embedding (P-SNE), and fully-shared non-linear embedding (F-SNE). Finally, a proposed structure with different model size settings achieved 71.4%,77.8%, and 82.1% on ImageNet-1k compared with 69.9%,77.1%, and 82.0% acquired by the original XCiT models, namely XCiT-N12, XCiT-T12, and XCiT-S12, respectively. Additionally, the proposed model achieved 94.8% in transfer learning experiments, on average, for CIFAR-10, CIFAR-100, Stanford Cars, and STL-10, which is superior to the baseline model of XCiT-S12 (94.5%). In particular, the proposed models demonstrated considerable improvements on the out-of-distribution detection task compared to the original XCiT models. ⓘ Ahn, Jaesin; Hong, Jiuk; Ju, Jeongwoo; Jung, Heechul ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea; Captos Co Ltd, Yangsan 50652, South Korea ⓘ ; Jung, Heechul/HTL-7199-2023 ⓘ 57212010622; 57353688600; 55651859700; 55652175200 ⓘ ajs0420@knu.ac.kr;hong4497@knu.ac.kr;veryju@captos.co.kr;heechul@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 8 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.25 2025-06-25 ⓘ 0 1 vision transformer; Q/K/V embedding; shared embedding; non-linear embedding; image classification ⓘ ⓘ image classification; non-linear embedding; Q/K/V embedding; shared embedding; vision transformer ⓘ ⓘ English 2023 2023-04 10.3390/math11081933 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Resilient Corrective Control of Asynchronous Sequential Machines Against Intermittent Loss of Actuator Outputs ⓘ This article proposes a resilient corrective control scheme for input/state asynchronous sequential machines (ASMs) against a class of actuator faults in which certain actuator outputs cannot be generated temporarily. We first present a mathematical formulation to describe the reachability of the controlled ASM damaged by the intermittent loss of actuator outputs. Based on the mathematical formulation, we address the existence condition and design procedure for a state-feedback corrective controller and a diagnoser that achieve resilience, that is, to make the closed-loop system exhibit normal input/state behaviors despite the intermittent loss of actuator outputs. To validate the applicability of the proposed concept and methodology, the closed-loop system of a practical asynchronous digital system is implemented on a field-programmable gate array (FPGA) and experimental verifications are provided. ⓘ Yang, Jung-Min; Kwak, Seong Woo ⓘ Kyungpook Natl Univ, Sch Elect Engn, Daegu 41566, South Korea; Pukyong Natl Univ, Dept Control & Instrumentat Engn, Busan 48513, South Korea ⓘ ⓘ 57208450551; 59816855300 ⓘ jmyang@ee.knu.ac.kr;ksw@pknu.ac.kr; ⓘ IEEE TRANSACTIONS ON CYBERNETICS ⓘ IEEE T CYBERNETICS ⓘ 2168-2267 2168-2275 53 10 SCIE AUTOMATION & CONTROL SYSTEMS;COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE;COMPUTER SCIENCE, CYBERNETICS ⓘ 2023 9.4 4.2 0.44 2025-06-25 ⓘ 4 4 Actuators; Transient analysis; Closed loop systems; Behavioral sciences; Field programmable gate arrays; Resilience; Fault tolerant systems; Asynchronous sequential machines (ASMs); actuator faults; corrective control; cyber security; field-programmable gate array (FPGA); resilient control ⓘ FAULT-TOLERANT CONTROL; NONLINEAR MULTIAGENT SYSTEMS; FEEDBACK CONTROL; INPUT/OUTPUT CONTROL; COMMUNICATION; MAXIMIZATION; FRAMEWORK; NETWORKS; ATTACKS; SCHEME ⓘ actuator faults; Asynchronous sequential machines (ASMs); corrective control; cyber security; field-programmable gate array (FPGA); resilient control ⓘ Behavioral research; Closed loop systems; Field programmable gate arrays (FPGA); Hysteresis; Logic gates; State feedback; Transient analysis; Actuator fault; Asynchronoi sequential machine; Asynchronous sequential machines; Behavioral science; Closed-loop system; Corrective control; Cyber security; Fault- tolerant systems; Field programmable gate array; Field programmables; Field-programmable gate array; Programmable gate array; Resilience; Resilient control; article; Actuators ⓘ English 2023 2023-10 10.1109/tcyb.2022.3167483 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Rewarded Meta-Pruning: Meta Learning with Rewards for Channel Pruning ⓘ Convolutional neural networks (CNNs) have gained recognition for their remarkable performance across various tasks. However, the sheer number of parameters and the computational demands pose challenges, particularly on edge devices with limited processing power. In response to these challenges, this paper presents a novel approach aimed at enhancing the efficiency of deep learning models. Our method introduces the concept of accuracy and efficiency coefficients, offering a fine-grained control mechanism to balance the trade-off between network accuracy and computational efficiency. At our core is the Rewarded Meta-Pruning algorithm, guiding neural network training to generate pruned model weight configurations. The selection of this pruned model is based on approximations of the final model's parameters, and it is precisely controlled through a reward function. This reward function empowers us to tailor the optimization process, leading to more effective fine-tuning and improved model performance. Extensive experiments and evaluations underscore the superiority of our proposed method when compared to state-of-the-art techniques. We conducted rigorous pruning experiments on well-established architectures such as ResNet-50, MobileNetV1, and MobileNetV2. The results not only validate the efficacy of our approach but also highlight its potential to significantly advance the field of model compression and deployment on resource-constrained edge devices. ⓘ Shibu, Athul; Kumar, Abhishek; Jung, Heechul; Lee, Dong-Gyu; Yang, Xinsong ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea ⓘ Yang, Xinsong/GLV-4131-2022; Kumar, Abhishek/ABA-5251-2021; , Xinsong/O-1510-2016; Jung, Heechul/HTL-7199-2023 ⓘ 58090284100; 57206266703; 55652175200; 57169003900 ⓘ athulshibu@knu.ac.kr;dglee@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 23 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0 2025-06-25 ⓘ 0 0 convolutional neural networks; meta-pruning; ResNet-50; reward function; channel pruning ⓘ ⓘ channel pruning; convolutional neural networks; meta-pruning; ResNet-50; reward function ⓘ ⓘ English 2023 2023-12 10.3390/math11234849 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Self-Adaptive Spherical Search With a Low-Precision Projection Matrix for Real-World Optimization ⓘ Since the last three decades, numerous search strategies have been introduced within the framework of different evolutionary algorithms (EAs). Most of the popular search strategies operate on the hypercube (HC) search model, and search models based on other hypershapes, such as hyper-spherical (HS), are not investigated well yet. The recently developed spherical search (SS) algorithm utilizing the HS search model has been shown to perform very well for the bound-constrained and constrained optimization problems compared to several state-of-the-art algorithms. Nevertheless, the computational burdens for generating an HS locus are higher than that for an HC locus. We propose an efficient technique to construct an HS locus by approximating the orthogonal projection matrix to resolve this issue. As per our empirical experiments, this technique significantly improves the performance of the original SS with less computational effort. Moreover, to enhance SS's search capability, we put forth a self-adaptation technique for choosing the effective values of the control parameters dynamically during the optimization process. We validate the proposed algorithm's performance on a plethora of real-world and benchmark optimization problems with and without constraints. Experimental results suggest that the proposed algorithm remains better than or at least comparable to the best-known state-of-the-art algorithms on a wide spectrum of problems. ⓘ Kumar, Abhishek; Das, Swagatam; Kong, Lingping; Snasel, Vaclav ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea; Indian Stat Inst, Elect & Commun Sci Unit, Kolkata 700108, India; VSB Tech Univ Ostrava, Fac Elect Engn & Comp Sci, Ostrava 70800, Czech Republic ⓘ ; Das, Swagatam/AAG-6753-2019; Snasel, Vaclav/B-8094-2009; Kong, Lingping/KPB-3641-2024; Kumar, Abhishek/ABA-5251-2021 ⓘ 57206266703; 24729258600; 56031337800; 57195632134 ⓘ abhishek.kumar.eee13@iitbhu.ac.in;swagatam.das@isical.ac.in;lingping_kong@yahoo.com;vaclav.snasel@vsb.cz; ⓘ IEEE TRANSACTIONS ON CYBERNETICS ⓘ IEEE T CYBERNETICS ⓘ 2168-2267 2168-2275 53 7 SCIE AUTOMATION & CONTROL SYSTEMS;COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE;COMPUTER SCIENCE, CYBERNETICS ⓘ 2023 9.4 4.2 0.66 2025-06-25 ⓘ 9 7 Search problems; Statistics; Sociology; Metaheuristics; Maintenance engineering; Heuristic algorithms; Computational modeling; Constrained optimization; evolutionary algorithms (EAs); parameter adaptation technique; real-world optimization problem (RWOPs); search style; spherical search (SS) ⓘ DIFFERENTIAL EVOLUTION; ALGORITHM; STRATEGY ⓘ Constrained optimization; evolutionary algorithms (EAs); parameter adaptation technique; real-world optimization problem (RWOPs); search style; spherical search (SS) ⓘ Algorithms; Biological Evolution; Computer Simulation; Benchmarking; Evolutionary algorithms; Heuristic algorithms; Precision engineering; Spheres; Adaptation techniques; Computational modelling; Evolutionary algorithm; Heuristics algorithm; Metaheuristic; Optimization problems; Parameter adaptation; Parameter adaptation technique; Real-world optimization; Real-world optimization problem; Search problem; Search style; Spherical search; algorithm; computer simulation; evolution; Constrained optimization ⓘ English 2023 2023-07 10.1109/tcyb.2021.3119386 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Spectral Salt-and-Pepper Patch Masking for Self-Supervised Speech Representation Learning ⓘ Recent advanced systems in the speech recognition domain use large Transformer neural networks that have been pretrained on massive speech data. General methods in the deep learning area have been frequently shared across various domains, and the Transformer model can also be used effectively across speech and image. In this paper, we introduce a novel masking method for self-supervised speech representation learning with salt-and-pepper (S & P) mask which is commonly used in computer vision. The proposed scheme includes consecutive quadrilateral-shaped S & P patches randomly contaminating the input speech spectrum. Furthermore, we modify the standard S & P mask to make it appropriate for the speech domain. In order to validate the effect of the proposed spectral S & P patch masking for the self-supervised representation learning approach, we conduct the pretraining and downstream experiments with two languages, English and Korean. To this end, we pretrain the speech representation model using each dataset and evaluate the pretrained models for feature extraction and fine-tuning performance on varying downstream tasks, respectively. The experimental outcomes clearly illustrate that the proposed spectral S & P patch masking is effective for various downstream tasks when combined with the conventional masking methods. ⓘ Kim, June-Woo; Chung, Hoon; Jung, Ho-Young ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea; Elect & Telecommun Res Inst, Daejeon 34129, South Korea ⓘ ; Chung, Hwei-Ming/X-4986-2019 ⓘ 57219550643; 35885362900; 57198760619 ⓘ kaen2891@knu.ac.kr;hchung@etri.re.kr;hoyjung@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 15 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0 2025-06-25 ⓘ 0 0 self-supervised learning; speech representation learning; salt-and-pepper masking; spectrum patch masking ⓘ NETWORK ⓘ salt-and-pepper masking; self-supervised learning; spectrum patch masking; speech representation learning ⓘ ⓘ English 2023 2023-08 10.3390/math11153418 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Review ⓘ Tumor microenvironment signaling and therapeutics in cancer progression ⓘ Tumor development and metastasis are facilitated by the complex interactions between cancer cells and their microenvironment, which comprises stromal cells and extracellular matrix (ECM) components, among other factors. Stromal cells can adopt new phenotypes to promote tumor cell invasion. A deep understanding of the signaling pathways involved in cell-to-cell and cell-to-ECM interactions is needed to design effective intervention strategies that might interrupt these interactions. In this review, we describe the tumor microenvironment (TME) components and associated therapeutics. We discuss the clinical advances in the prevalent and newly discovered signaling pathways in the TME, the immune checkpoints and immunosuppressive chemokines, and currently used inhibitors targeting these pathways. These include both intrinsic and non-autonomous tumor cell signaling pathways in the TME: protein kinase C (PKC) signaling, Notch, and transforming growth factor (TGF-beta) signaling, Endoplasmic Reticulum (ER) stress response, lactate signaling, Metabolic reprogramming, cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) and Siglec signaling pathways. We also discuss the recent advances in Programmed Cell Death Protein 1 (PD-1), Cytotoxic T-Lymphocyte Associated Protein 4 (CTLA4), T-cell immunoglobulin mucin-3 (TIM-3) and Lymphocyte Activating Gene 3 (LAG3) immune checkpoint inhibitors along with the C-C chemokine receptor 4 (CCR4)- C-C class chemokines 22 (CCL22)/ and 17 (CCL17), C-C chemokine receptor type 2 (CCR2)- chemokine (C-C motif) ligand 2 (CCL2), C-C chemokine receptor type 5 (CCR5)- chemokine (C-C motif) ligand 3 (CCL3) chemokine signaling axis in the TME. In addition, this review provides a holistic understanding of the TME as we discuss the three-dimensional and microfluidic models of the TME, which are believed to recapitulate the original characteristics of the patient tumor and hence may be used as a platform to study new mechanisms and screen for various anti-cancer therapies. We further discuss the systemic influences of gut microbiota in TME reprogramming and treatment response. Overall, this review provides a comprehensive analysis of the diverse and most critical signaling pathways in the TME, highlighting the associated newest and critical preclinical and clinical studies along with their underlying biology. We highlight the importance of the most recent technologies of microfluidics and lab-on-chip models for TME research and also present an overview of extrinsic factors, such as the inhabitant human microbiome, which have the potential to modulate TME biology and drug responses. ⓘ Goenka, Anshika; Khan, Fatima; Verma, Bhupender; Sinha, Priyanka; Dmello, Crismita C.; Jogalekar, Manasi P.; Gangadaran, Prakash; Ahn, Byeong-Cheol ⓘ Northwestern Univ, Robert H Lurie Comprehens Canc Ctr, Ken & Ruth Davee Dept Neurol, Feinberg Sch Med, Chicago, IL 60611 USA; Northwestern Univ, Feinberg Sch Med, Dept Neurol Surg, Chicago, IL 60611 USA; Harvard Med Sch, Massachusetts Eye & Ear Infirm, Schepens Eye Res Inst, Dept Ophthalmol, Boston, MA 02114 USA; Harvard Med Sch, Massachusetts Gen Hosp, MassGeneral Inst Neurodegenerat Dis, Dept Neurol, Charlestown, MA 02129 USA; Univ Calif San Francisco, Helen Diller Family Comprehens Canc Ctr, San Francisco, CA 94143 USA; Kyungpook Natl Univ, Sch Med, BK21 FOUR KNU Convergence Educ Program Biomed Sci, Dept Biomed Sci, Daegu 41944, South Korea; Kyungpook Natl Univ, Kyungpook Natl Univ Hosp, Sch Med, Dept Nucl Med, Daegu 41944, South Korea ⓘ Dmello, Crismita/ABE-3901-2022; Sinha, Priyanka/HIR-6508-2022; Khan, Fatima/GQP-2552-2022; Gangadaran, Prakash/AAV-3102-2021; Goenka, Anshika/AAR-4277-2021; Jogalekar, Manasi/AAG-6925-2020 ⓘ 57218414224; 57205752395; 56007106200; 57219551769; 55151400900; 57194462336; 54393130400; 7202791511 ⓘ anshika.goenka@northwestern.edu;prakashg@knu.ac.kr;abc2000@knu.ac.kr; ⓘ CANCER COMMUNICATIONS ⓘ CANCER COMMUN ⓘ 2523-3548 43 5 SCIE ONCOLOGY ⓘ 2023 20.1 4.2 8.49 2025-06-25 ⓘ 133 132 3D-model; cancer therapy; gut microbiota; immune signaling; metabolism; signaling; tumor microenvironment ⓘ PROTEIN-KINASE-C; REGULATORY T-CELLS; ENDOPLASMIC-RETICULUM STRESS; ENDOTHELIAL GROWTH-FACTOR; IMMUNE CHECKPOINT; DENDRITIC CELLS; BREAST-CANCER; PHASE-I; ANTITUMOR-ACTIVITY; COLORECTAL-CANCER ⓘ 3D-model; cancer therapy; gut microbiota; immune signaling; metabolism; signaling; tumor microenvironment ⓘ Chemokines; Humans; Neoplasms; Neoplastic Processes; Receptors, Chemokine; Signal Transduction; Tumor Microenvironment; antisense oligonucleotide; bevacizumab; canakinumab; CD135 antigen; chemokine receptor CCR2; chemokine receptor CCR4; chemokine receptor CCR5; chimeric antigen receptor; cilengitide; cisplatin; crenolanib; cyclic GMP; cytotoxic T lymphocyte antigen 4; diacylglycerol; epidermal growth factor receptor variant III; etaracizumab; everolimus; fibroblast growth factor; FMS related receptor tyrosine kinase 3 ligand; fresolimumab; gap junction protein; granulocyte macrophage colony stimulating factor; hypoxia inducible factor; immunoglobulin enhancer binding protein; incyclinide; indoleamine 2,3 dioxygenase; interferon regulatory factor 3; interleukin 12; interleukin 1beta; interleukin 2; interleukin 6; jagged canonical notch ligand 2; lactate dehydrogenase; losartan; lymphocyte activation gene 3 protein; macrophage derived chemokine; macrophage inflammatory protein 1alpha; matrix metalloproteinase; mesenchymal epithelial transition factor; monocyte chemotactic protein 1; p21 activated kinase 4; pazopanib; pegvorhyaluronidase alfa; pexidartinib; platelet derived growth factor; platelet derived growth factor receptor; programmed death 1 receptor; prostate specific membrane antigen; protein kinase C; protein tyrosine phosphatase SHP 1; protein tyrosine phosphatase SHP 2; protocadherin; reactive oxygen metabolite; scatter factor; secondary lymphoid tissue chemokine; sialic acid binding immunoglobulin like lectin; sipuleucel T; STAT3 protein; stromal cell derived factor 1; T cell immunoglobulin mucin 3; talabostat; thymus and activation regulated chemokine; transforming growth factor; transforming growth factor beta; tryptophan 2,3 dioxygenase; tumor necrosis factor; tumor necrosis factor receptor superfamily member 8; unclassified drug; vasculotropin; chemokine; chemokine receptor; antigen presenting cell; cancer associated fibroblast; cancer growth; cancer therapy; CD8+ T lymphocyte; dendritic cell; effector cell; endoplasmic reticulum stress; endothelium cell; epithelial mesenchymal transition; extracellular matrix; gene; glioblastoma; head and neck squamous cell carcinoma; high throughput screening; human; intestine flora; mesenchymal stem cell; microbiome; microfluidic analysis; natural killer cell; nonhuman; Notch signaling; organoid; phase 1 clinical trial (topic); phase 2 clinical trial (topic); phase 3 clinical trial (topic); Review; signal transduction; stimulator of interferon gene; TGF beta signaling; tumor associated leukocyte; tumor microenvironment; tumor spheroid; tumor xenograft; tumor-associated macrophage; metabolism; neoplasm; oncogenesis and malignant transformation; signal transduction ⓘ English 2023 2023-05 10.1002/cac2.12416 ⓘ 바로가기 바로가기 바로가기 바로가기
○ ○ Article ⓘ Unsupervised Representation Learning with Task-Agnostic Feature Masking for Robust End-to-End Speech Recognition ⓘ Unsupervised learning-based approaches for training speech vector representations (SVR) have recently been widely applied. While pretrained SVR models excel in relatively clean automatic speech recognition (ASR) tasks, such as those recorded in laboratory environments, they are still insufficient for practical applications with various types of noise, intonation, and dialects. To cope with this problem, we present a novel unsupervised SVR learning method for practical end-to-end ASR models. Our approach involves designing a speech feature masking method to stabilize SVR model learning and improve the performance of the ASR model in a downstream task. By introducing a noise masking strategy into diverse combinations of the time and frequency regions of the spectrogram, the SVR model becomes a robust representation extractor for the ASR model in practical scenarios. In pretraining experiments, we train the SVR model using approximately 18,000 h of Korean speech datasets that included diverse speakers and were recorded in environments with various amounts of noise. The weights of the pretrained SVR extractor are then frozen, and the extracted speech representations are used for ASR model training in a downstream task. The experimental results show that the ASR model using our proposed SVR extractor significantly outperforms conventional methods. ⓘ Kim, June-Woo; Chung, Hoon; Jung, Ho-Young ⓘ Kyungpook Natl Univ, Dept Artificial Intelligence, Daegu 41566, South Korea; Elect & Telecommun Res Inst, Daejeon 34129, South Korea ⓘ ; Chung, Hwei-Ming/X-4986-2019 ⓘ 57219550643; 35885362900; 57198760619 ⓘ hoyjung@knu.ac.kr; ⓘ MATHEMATICS ⓘ MATHEMATICS-BASEL ⓘ 2227-7390 11 3 SCIE MATHEMATICS ⓘ 2023 2.3 4.2 0.25 2025-06-25 ⓘ 2 1 speech vector representation; representation learning; unsupervised learning; feature representation extractor; speech recognition; deep learning; neural network; speech processing ⓘ ⓘ deep learning; feature representation extractor; neural network; representation learning; speech processing; speech recognition; speech vector representation; unsupervised learning ⓘ ⓘ English 2023 2023-02 10.3390/math11030622 ⓘ 바로가기 바로가기 바로가기 바로가기
페이지 이동: