Article
Operations Research & Management Science
Wen Huang, Ke Wei
Summary: This paper discusses the problem of minimizing a differentiable function and a nonsmooth function on a Riemannian manifold. Various versions of Riemannian proximal gradient methods have been proposed to solve this problem. However, their convergence analyses require exact solutions to the Riemannian proximal mapping, which is costly or impractical. In this paper, an inexact Riemannian proximal gradient method is studied, and it is proven that accurate solutions to the proximal mapping guarantee global convergence and local convergence rate based on the Riemannian Kurdyka-Lojasiewicz property. Practical conditions for the accuracy of solving the Riemannian proximal mapping are also provided. Experimental results on sparse principal component analysis validate the proposed practical conditions.
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS
(2023)
Article
Geochemistry & Geophysics
Xiangfei Shen, Haijun Liu, Xinzheng Zhang, Kai Qin, Xichuan Zhou
Summary: The study introduces a local-global sparse regression unmixing method, which combines local sparsity regularization and global sparsity regularization to effectively estimate the abundance of a given image. Experimental results demonstrate the effectiveness of the proposed algorithm.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Mathematics, Applied
Guoqiang Wang, Wenjian Yu, Xiubo Liang, Yuanqing Wu, Bo Yu
Summary: This paper presents an efficient algorithm for large-scale sparse least squares problems, reducing the original problem into a sequence of subproblems by utilizing the sparsity of the gradient. The algorithm is shown to globally converge and locally converge at a linear rate, with a complexity of O(mp ln(1/epsilon). Numerical comparisons demonstrate the robustness and high performance of the algorithm in solving LASSO problems and image deblurring problems.
SIAM JOURNAL ON SCIENTIFIC COMPUTING
(2022)
Article
Computer Science, Software Engineering
Gilles Bareilles, Franck Iutzeler, Jerome Malick
Summary: Proximal methods are used to identify the underlying substructure of nonsmooth optimization problems. This paper introduces the integration of proximal gradient steps with Riemannian Newton-like methods to achieve superlinear convergence when solving nondegenerated nonsmooth nonconvex optimization problems.
MATHEMATICAL PROGRAMMING
(2023)
Article
Computer Science, Software Engineering
Wen Huang, Ke Wei
Summary: The paper presents a Riemannian proximal gradient method and its accelerated variant for optimization problems constrained on a manifold. Global convergence and a convergence rate of O(1/k) for the RPG algorithm are established, with the sequence generated converging to a single stationary point under the assumption of a Riemannian KL property. Additionally, the flexibility of RPG on the Stiefel manifold covers a variety of problems, including sparse PCA.
MATHEMATICAL PROGRAMMING
(2022)
Article
Automation & Control Systems
Bokun Wang, Shiqian Ma, Lingzhou Xue
Summary: This paper presents two Riemannian stochastic proximal gradient methods for minimizing nonsmooth functions over the Stiefel manifold and provides IFO complexity analysis. Experimental results demonstrate the superior performance of the proposed methods in online sparse PCA and robust low-rank matrix completion problems.
JOURNAL OF MACHINE LEARNING RESEARCH
(2022)
Article
Geochemistry & Geophysics
Jing-Jie Cao, Gang Yao, Nuno da Silva
Summary: Seismic data interpolation is an important tool for providing complete data in the presence of incompleteness. This study proposes a novel nonconvex regularization model that combines the L-1 norm and a hyperbolic tangent-based function to achieve seismic interpolation. Synthetic and field data examples demonstrate the effectiveness and robustness of the proposed method.
PURE AND APPLIED GEOPHYSICS
(2022)
Article
Operations Research & Management Science
Xian Zhang, Dingtao Peng
Summary: This paper studies the constrained group sparse regularization optimization problem and proposes a continuous relaxation model and corresponding algorithm. By analyzing the optimality conditions and relationships between the relaxation problem and the original problem, the equivalence of the two problems is revealed to some extent. The numerical results of the algorithm demonstrate its efficiency.
COMPUTATIONAL OPTIMIZATION AND APPLICATIONS
(2022)
Article
Management
Rafael Blanquero, Emilio Carrizosa, Cristina Molero-Rio, Dolores Romero Morales
Summary: This paper proposes an optimal regression tree model based on a continuous optimization problem, which aims to strike a balance between prediction accuracy and sparsity. The model can fulfill important properties for regression tasks and provide local explanations due to the smoothness of predictions. The computational experience demonstrates the superiority of this approach in terms of prediction accuracy compared to standard benchmark regression methods, and the scalability with respect to sample size is also illustrated.
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH
(2022)
Article
Mathematics, Applied
Yuncheng Liu, Fuquan Xia
Summary: In this paper, we propose a proximal variable smoothing gradient method for solving a class of three-composite nonconvex nonsmooth optimization problems. We also introduce a new proximal variable smoothing incremental aggregated gradient method based on this approach. Preliminary numerical experiments demonstrate the effectiveness of the proposed methods.
NUMERICAL ALGORITHMS
(2023)
Article
Engineering, Electrical & Electronic
Zhuan Zhang, Shuisheng Zhou, Dong Li, Ting Yang
Summary: Sparse two-dimensional principal component analysis (2DPCA) is an effective dimensionality reduction technique widely used in face recognition. The Manifold Proximal Gradient Algorithm (ManPG) is efficient and robust for solving the non-convex and non-smooth optimization problem of sparse 2DPCA, but its computational complexity is high. In order to reduce this complexity, the Riemannian Proximal Stochastic Gradient Descent algorithm (RPSGD) is proposed, which maintains sublinear convergence and offers advantages in large-scale sparse 2DPCA applications.
DIGITAL SIGNAL PROCESSING
(2022)
Article
Statistics & Probability
Leo Miolane, Andrea Montanari
Summary: The Lasso is a popular regression method for high-dimensional problems, with statistical properties related to soft-thresholding denoisers. The method can be used to evaluate the performance of various data-driven procedures and has been shown to be effective in dealing with Gaussian noise.
ANNALS OF STATISTICS
(2021)
Article
Computer Science, Artificial Intelligence
Yu Guo, Yuan Sun, Zheng Wang, Feiping Nie, Fei Wang
Summary: In this article, a novel unsupervised feature selection model, DSFEL, is proposed. DSFEL includes a module for learning a block-diagonal structural sparse graph to represent the clustering structure and another module for learning a completely row-sparse projection matrix using the l(2,0)-norm constraint to select distinctive features. Experimental results on nine real-world datasets demonstrate that the proposed method outperforms existing state-of-the-art unsupervised feature selection methods.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Mathematics, Applied
Guiyun Xiao, Zheng-Jian Bai
Summary: In this paper, the authors consider the problem of sparse least squares regression with a probabilistic simplex constraint. They reformulate the problem as a nonconvex and nonsmooth l(1)-regularized minimization problem over the unit sphere, and propose a geometric proximal gradient method to solve it.
JOURNAL OF SCIENTIFIC COMPUTING
(2022)
Article
Automation & Control Systems
Michael R. Metel, Akiko Takeda
Summary: This paper focuses on stochastic proximal gradient methods for optimizing a smooth non-convex loss function with a non-smooth non-convex regularizer and convex constraints. The authors present the first non-asymptotic convergence bounds for this class of problem and compare their algorithms with the current state-of-the-art deterministic algorithm, finding superior convergence in a numerical experiment.
JOURNAL OF MACHINE LEARNING RESEARCH
(2021)
Article
Engineering, Electrical & Electronic
Myonghoon Kwak, Wooseok Choi, Seongjae Heo, Chuljun Lee, Revannath Nikam, Seyoung Kim, Hyunsang Hwang
Summary: A novel neuromorphic architecture was proposed using different types of RRAM for neural network training, achieving a pattern recognition accuracy of 98%.
IEEE ELECTRON DEVICE LETTERS
(2021)
Article
Materials Science, Ceramics
Gyeongran Shim, Soo-Hyun Kim, Seyoung Kim, In-Sub Han, Hyung-Joon Bang, Young-Hoon Seong, Seulhee Lee, Wan Sik Kim, Kwangbok Shin
Summary: SiCf/SiC composites fabricated through a two-step pyrolysis process and LSI process exhibit increased density, reduced porosity, and higher flexural strength compared to those fabricated through a one-step pyrolysis process. Among specimens processed with the same two-step pyrolysis, those subjected to LSI at 1500 degrees Celsius show significantly higher flexural strength than those at 1450 or 1550 degrees. Additionally, mechanical properties of SiCf/SiC specimens processed at 1500 degrees Celsius are superior to those at 1550 degrees Celsius despite similar porosity and density due to the rapid degradation of Tyranno-S grade SiC fibers at higher LSI process temperatures.
CERAMICS INTERNATIONAL
(2022)
Article
Polymer Science
Jewon Choi, Seyoung Kim, Huy Ju Mun, Jin Yoo, Soo-Hyung Choi, Kookheon Char
Summary: ABC triblock copolymer hydrogels with hydrophobic cores and supramolecular junctions exhibit multifunctional antifreeze properties, making them potentially valuable for cryo-preservable and bio-injectable drug storage and delivery.
MACROMOLECULAR RAPID COMMUNICATIONS
(2022)
Article
Polymer Science
Sanghoon Song, Yohan Chang, Seung-Hwan Oh, Soyoon Kim, Seungsoo Choi, Seyoung Kim, Jin-Kyun Lee, Soo-Hyung Choi, Jeewoo Lim
Summary: We developed a method for the synthesis of heavily fluorinated polynorbornenes through fluoropolymer dispersion ring-opening metathesis polymerization. The use of a fluorinated block copolymer as a stabilizer enabled complete monomer conversion within minutes, resulting in fluorous polynorbornenes with predetermined molecular weights and low molecular weight distributions.
Article
Energy & Fuels
Laidong Zhou, Tong-Tong Zuo, Chun Yuen Kwok, Se Young Kim, Abdeljalil Assoud, Qiang Zhang, Juergen Janek, Linda F. Nazar
Summary: A series of new lithium mixed-metal chlorospinels are reported as solid electrolytes for all-solid-state lithium batteries, which exhibit high ionic conductivity and low electronic conductivity, showing potential for high-performance all-solid-state lithium batteries.
Article
Biochemistry & Molecular Biology
Seyoung Kim, Daniel J. Fesenmeier, Sungwan Park, Sandra E. Torregrosa-Allen, Bennett D. Elzey, You-Yeon Won
Summary: This study investigates the clearance pathways and kinetics of amphiphilic poly(styrene-block-ethylene glycol) (PS-PEG) nanoparticles in the lungs. The results demonstrate that PS-PEG nanoparticles have a prolonged retention time in the lungs and are cleared through both mucociliary escalator mechanism and alveolar clearance by macrophages. Additionally, a portion of the nanoparticles translocates directly into the circulation.
Article
Medicine, Research & Experimental
Jin-Ah Beak, Min-Jung Park, Se-Young Kim, JooYeon Jhun, Jin Seok Woo, Jeong Won Choi, Hyun Sik Na, Soon Kyu Lee, Jong Young Choi, Mi-La Cho
Summary: The combination of Lactobacillus acidophilus and FK506 has been found to effectively control graft-versus-host disease (GvHD) after allogeneic hematopoietic stem cell transplantation. It inhibits the differentiation of pro-inflammatory T cells and promotes the differentiation of regulatory T cells, as well as modulating the immune function and alleviating GvHD symptoms. This combination treatment also shows potential in regulating immune responses in healthy individuals and liver transplant patients.
JOURNAL OF TRANSLATIONAL MEDICINE
(2022)
Article
Immunology
Soon Kyu Lee, Min-Jung Park, Jeong Won Choi, Jin-Ah Baek, Se-Young Kim, Ho Joong Choi, Young Kyoung You, Jeong Won Jang, Pil Soo Sung, Si Hyun Bae, Seung Kew Yoon, Jong Young Choi, Mi-La Cho
Summary: In this study, a patient-derived avatar model was developed to predict immune homeostasis in liver transplantation patients. The model showed promise in evaluating treatment responses and reducing risks to patients.
FRONTIERS IN IMMUNOLOGY
(2022)
Article
Chemistry, Physical
Daniel J. Fesenmeier, Sungwan Park, Seyoung Kim, You-Yeon Won
Summary: This study investigates the hypothesis that the surface mechanical properties of poly(styrene)-poly(ethylene glycol) (PS-PEG) micelles are influenced by the structure of the PEG corona. Changes in micelle aggregation number and the chemistry of the PS-PEG block copolymer are expected to alter the characteristics of the PEG corona and affect the surface mechanical properties of the resulting micelle film.
JOURNAL OF COLLOID AND INTERFACE SCIENCE
(2022)
Article
Engineering, Electrical & Electronic
Wooseok Choi, Wonjae Ji, Seongjae Heo, Donguk Lee, Kyungmi Noh, Chuljun Lee, Jiyong Woo, Seyoung Kim, Hyunsang Hwang
Summary: In this study, the read noise of RRAM is utilized for implementing probabilistic neural network, and the electrical characteristics of TiOx-based RRAM under different forming conditions are analyzed. An array mapping scheme is demonstrated to transfer weights to the 1T1R array, and through NN simulations, the promising results of the probabilistic NN over deterministic NN on nonlinear classification problem are validated.
IEEE ELECTRON DEVICE LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Donguk Lee, Jongwon Lee, Chuljun Lee, Seyoung Kim, Hyunsang Hwang
Summary: Neuromorphic computing has attracted considerable attention for its potential in efficient parallel computations. The proposed Li-ECRAM neuron offers exceptional area efficiency and low-power operation compared to other types of novel nonvolatile memory-based neurons, with linear integration characteristics eliminating the need for a nonlinear compensating circuit.
IEEE TRANSACTIONS ON ELECTRON DEVICES
(2022)
Article
Chemistry, Physical
SeYoung Kim, Hyungyeon Cha, Robert Kostecki, Guoying Chen
Summary: All-solid-state batteries, composed of a 4 V class layered oxide cathode, an inorganic solid-state electrolyte, and a lithium metal anode, are seen as the future of energy storage technologies. However, there are challenges related to dendrite formation and cathode instabilities that hinder their development. In this study, composite cathode structures were developed to address these challenges, resulting in exceptional performance and stability in ASSB cells. The study emphasizes the importance of proper cathode composite design for better-performing ASSB cells.
ACS ENERGY LETTERS
(2023)
Article
Andrology
Jihye Choi, Seyoung Kim, Seung-Ju Lee, Sangrak Bae, Sooyoun Kim
Summary: This study found that Korean mothers have limited knowledge about male HPV vaccination, have low perceived risk of HPV, and remain highly hesitant about vaccinating their sons. Therefore, it is important to raise public awareness about male HPV vaccination and alleviate vaccine hesitancy.
WORLD JOURNAL OF MENS HEALTH
(2023)
Article
Engineering, Electrical & Electronic
Wooseok Choi, Myonghoon Kwak, Donguk Lee, Sangmin Lee, Chuljun Lee, Seyoung Kim, Hyunsang Hwang
Summary: In this study, the inherent stochastic characteristics of the OTS threshold voltage are utilized to enhance the inference performance of neural networks through stochastic resonance. The threshold switching of the OTS device is characterized, and a signal detection method using the OTS device is proposed. It is confirmed that the inherent stochasticity in the OTS device can effectively restore degraded images and improve recognition accuracy in poor visibility conditions.
IEEE JOURNAL OF THE ELECTRON DEVICES SOCIETY
(2022)
Article
Chemistry, Physical
Seyoung Kim, Jung-Min Kim, Kathleen Wood, Soo-Hyung Choi
Summary: This study investigated the nanostructure of complex coacervate core hydrogels with varying compositions of cationic charged groups using small-angle X-ray/neutron scattering. The results showed that as the fraction of cationic end-block groups increased, the polymer volume fraction in the cores, interfacial tension, and salt resistance also increased.