https://scholar.dgist.ac.kr/handle/20.500.11750/115 2026-04-05T13:58:07Z https://scholar.dgist.ac.kr/handle/20.500.11750/17096 Title: TENAS: Using Taylor Expansion and Channel-level Skip Connection for Neural Architecture Search Author(s): Lim, Heechul; Kim, Min-Soo Abstract: There is growing interest in automating designing good neural network architectures. The NAS methods proposed recently have significantly reduced the architecture search cost by sharing parameters, but there is still a challenging problem in designing search space. The existing operation-level architecture search methods require a large amount of computing power or designing the search space of operations very carefully. In this paper, we investigate the possibility of achieving competitive performance with them only using a small amount of computing power and without designing search space carefully. We propose TENAS using Taylor expansion and only a fixed type of operation. The resulting architecture is sparse in terms of channel and has different topology at different cells. The experimental results for CIFAR-10 and ImageNet show that a fine-granular and sparse model searched by TENAS achieves very competitive performance with dense models searched by the existing methods. Author 2022-07-31T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/15461 Title: Tweaking Deep Neural Networks Author(s): Kim, Jinwook; Yoon, Heeyong; Kim, Min-Soo Abstract: Deep neural networks are trained so as to achieve a kind of the maximum overall accuracy through a learning process using given training data. Therefore, it is difficult to fix them to improve the accuracies of specific problematic classes or classes of interest that may be valuable to some users or applications. To address this issue, we propose the synaptic join method to tweak neural networks by adding certain additional synapses from the intermediate hidden layers to the output layer across layers and additionally training only these synapses, if necessary. To select the most effective synapses, the synaptic join method evaluates the goodness of all the possible candidate synapses between the hidden neurons and output neurons based on the distribution of all the possible proper weights. The experimental results show that the proposed method can effectively improve the accuracies of specific classes in a controllable way. CCBY 2022-08-31T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/10398 Title: MRPrimerW2: an enhanced tool for rapid design of valid high-quality primers with multiple search modes for qPCR experiments Author(s): Jeon, Hajin; Bae, Jeongmin; Hwang, Sang Hyun; Hwang, Kyu-Young; Lee, Hyun-Seob; Kim, Hyerin; Kim, Min-Soo Abstract: For the best results in quantitative polymerase chain reaction (qPCR) experiments, it is essential to design high-quality primers considering a multitude of constraints and the purpose of experiments. The constraints include many filtering constraints, homology test on a huge number of off-target sequences, the same constraints for batch design of primers, exon spanning, and avoiding single nucleotide polymorphism (SNP) sites. The target sequences are either in database or given as FASTA sequences, and the experiment is for amplifying either each target sequence with each corresponding primer pairs designed under the same constraints or all target sequences with a single pair of primers. Many websites have been proposed, but none of them including our previous MRPrimerW fulfilled all the above features. Here, we describe the MRPrimerW2, the update version of MRPrimerW, which fulfils all the features by maintaining the advantages of MRPrimerW in terms of the kinds and sizes of databases for valid primers and the number of search modes. To achieve it, we exploited GPU computation and a disk-based key-value store using PCIe SSD. The complete set of 3 509 244 680 valid primers of MRPrimerW2 covers 99% of nine important organisms in an exhaustive manner. Free access: http://MRPrimerW2.com. © The Author(s) 2019. Published by Oxford University Press on behalf of Nucleic Acids Research. 2019-06-30T15:00:00Z https://scholar.dgist.ac.kr/handle/20.500.11750/9779 Title: Weakly supervised lesion localization for age-related macular degeneration detection using optical coherence tomography images Author(s): Yang, Hyun Lim; Kim, Jong Jin; Kim, Jong Ho; Kang, Yong Koo; Park, Dong Ho; Park, Han Sang; Kim, Hong Kyun; Kim, Min-Soo Abstract: Age-related macular degeneration (AMD) is the main cause of irreversible blindness among the elderly and require early diagnosis to prevent vision loss, and careful treatment is essential. Optical coherence tomography (OCT), the most commonly used imaging method in the retinal area for the diagnosis of AMD, is usually interpreted by a clinician, and OCT can help diagnose disease on the basis of the relevant diagnostic criteria, but these judgments can be somewhat subjective. We propose an algorithm for the detection of AMD based on a weakly supervised convolutional neural network (CNN) model to support computer-aided diagnosis (CAD) system. Our main contributions are the following three things. (1) We propose a concise CNN model for OCT images, which outperforms the existing large CNN models using VGG16 and GoogLeNet architectures. (2) We propose an algorithm called Expressive Gradients (EG) that extends the existing Integrated Gradients (IG) algorithm so as to exploit not only the input-level attribution map, but also the high-level attribution maps. Due to enriched gradients, EG can highlight suspicious regions for diagnosis of AMD better than the guided-backpropagation method and IG. (3) Our method provides two visualization options: Overlay and top-k bounding boxes, which would be useful for CAD. Through experimental evaluation using 10,100 clinical OCT images from AMD patients, we demonstrate that our EG algorithm outperforms the IG algorithm in terms of localization accuracy and also outperforms the existing object detection methods in terms of class accuracy. © 2019 Yang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. 2019-03-31T15:00:00Z