• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.695-697
• /
• 2023

최근 몇 년 간 딥러닝 기반 모델의 규모와 복잡성이 증가하면서 강력하고, 높은 정확도가 확보되지만 많은 양의 계산 자원과 메모리가 필요하기 때문에 모바일 장치나 임베디드 시스템과 같은 리소스가 제한된 환경에서의 배포에 제약사항이 생긴다. 복잡한 딥러닝 모델의 배포 및 운영 시 요구되는 고성능 컴퓨터 자원의 문제점을 해결하고자 사전 학습된 대규모 모델로부터 가벼운 모델을 학습시키는 지식증류 기법이 제안되었다. 하지만 현대 딥러닝 기반 모델은 높은 정확도 대비 훈련 데이터에 과적합 되는 과잉 확신(overconfidence) 문제에 대한 대책이 필요하다. 본 논문은 효율적인 경량화를 위한 미리 학습된 모델의 과잉 확신을 방지하고자 초점 손실(focal loss)을 이용한 모델 보정 기법을 언급하며, 다양한 손실 함수 변형에 따라서 지식증류의 성능이 어떻게 변화하는지에 대해 탐구하고자 한다.

• IEIE Transactions on Smart Processing and Computing
• /
• v.2 no.6
• /
• pp.339-344
• /
• 2013

This paper presents an automatic focal length estimation method to correct the fisheye lens distortion in a spatially adaptive manner. The proposed method estimates the focal length of the fisheye lens by generating two reference focal lengths. The distorted fisheye lens image is finally corrected using the orthographic projection model. The experimental results showed that the proposed focal length estimation method is more accurate than existing methods in terms of the loss rate.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.2
• /
• pp.15-25
• /
• 2024

For effective analysis of animal ecosystems, technology that can automatically identify the current status of animal habitats is crucial. Specifically, animal sound classification, which identifies species based on their sounds, is gaining great attention where video-based discrimination is impractical. Traditional studies have relied on a single deep learning model to classify animal sounds. However, sounds collected in outdoor settings often include substantial background noise, complicating the task for a single model. In addition, data imbalance among species may lead to biased model training. To address these challenges, in this paper, we propose an animal sound classification scheme that combines predictions from multiple models using Focal Loss, which adjusts penalties based on class data volume. Experiments on public datasets have demonstrated that our scheme can improve recall by up to 22.6% compared to an average of single models.

• Annals of Clinical Neurophysiology
• /
• v.3 no.1
• /
• pp.40-42
• /
• 2001

Anorexia nervosa(AN) is a disorder characterized by disturbance of body image, fear of gaining weight, severe weight loss and, in female, amenorrhea. Compared with normal persons, patients with AN have neuropathic symptoms more frequently. But electrophysiologic abnormalities have rarely been reported. We experienced a case with recurrent neuropathic symptoms after severe weight loss. Further evaluation revealed AN. Electrophysiologic study showed sensorimotor polyneuropathy and focal neuropathy with conduction block. As far as we know, this feature of neuropathy in AN has not been described. We describe unusual feature of neuropathy in our patient with literature review.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.49-54
• /
• 2024

Data imbalance is a common issue encountered in classification problems, stemming from a significant disparity in the number of samples between classes within the dataset. Such data imbalance typically leads to problems in classification models, including overfitting, underfitting, and misinterpretation of performance metrics. Methods to address this issue include resampling, augmentation, regularization techniques, and adjustment of loss functions. In this paper, we focus on loss function adjustment, particularly comparing the performance of various configurations of loss functions (Cross Entropy, Balanced Cross Entropy, two settings of Focal Loss: 𝛼 = 1 and 𝛼 = Balanced, Asymmetric Loss) on Multi-Class black-box video data with imbalance issues. The comparison is conducted using the I3D, and R3D_18 models.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.2
• /
• pp.263-283
• /
• 2024

Users' comments after online shopping are critical to product reputation and business improvement. These comments, sometimes known as e-commerce reviews, influence other customers' purchasing decisions. To confront large amounts of e-commerce reviews, automatic analysis based on machine learning and deep learning draws more and more attention. A core task therein is sentiment analysis. However, the e-commerce reviews exhibit the following characteristics: (1) inconsistency between comment content and the star rating; (2) a large number of unlabeled data, i.e., comments without a star rating, and (3) the data imbalance caused by the sparse negative comments. This paper employs Bidirectional Encoder Representation from Transformers (BERT), one of the best natural language processing models, as the base model. According to the above data characteristics, we propose the F_MixBERT framework, to more effectively use inconsistently low-quality and unlabeled data and resolve the problem of data imbalance. In the framework, the proposed MixBERT incorporates the MixMatch approach into BERT's high-dimensional vectors to train the unlabeled and low-quality data with generated pseudo labels. Meanwhile, data imbalance is resolved by Focal loss, which penalizes the contribution of large-scale data and easily-identifiable data to total loss. Comparative experiments demonstrate that the proposed framework outperforms BERT and MixBERT for sentiment analysis of e-commerce comments.

• Molecules and Cells
• /
• v.42 no.8
• /
• pp.589-596
• /
• 2019

${\beta}Pix$ is a guanine nucleotide exchange factor for the Rho family small GTPases, Rac1 and Cdc42. It is known to regulate focal adhesion dynamics and cell migration. However, the in vivo role of ${\beta}Pix$ is currently not well understood. Here, we report the production and characterization of ${\beta}Pix$-KO mice. Loss of ${\beta}Pix$ results in embryonic lethality accompanied by abnormal developmental features, such as incomplete neural tube closure, impaired axial rotation, and failure of allantois-chorion fusion. We also generated ${\beta}Pix$-KO mouse embryonic fibroblasts (MEFs) to examine ${\beta}Pix$ function in mouse fibroblasts. ${\beta}Pix$-KO MEFs exhibit decreased Rac1 activity, and defects in cell spreading and platelet-derived growth factor (PDGF)-induced ruffle formation and chemotaxis. The average size of focal adhesions is increased in ${\beta}Pix$-KO MEFs. Interestingly, ${\beta}Pix$-KO MEFs showed increased motility in random migration and rapid wound healing with elevated levels of MLC2 phosphorylation. Taken together, our data demonstrate that ${\beta}Pix$ plays essential roles in early embryonic development, cell spreading, and cell migration in fibroblasts.

• Applied Microscopy
• /
• v.28 no.4
• /
• pp.577-590
• /
• 1998

Puromycin aminonucleoside (PAN) nephropathy was induced in a group of Sprague-Dawley rat by a single dose of intraperitoneal Injection to study an ultrastructural change of glomerulus. The experimental rats developed proteinuria three days after PAN injection. Electron microscopic studies of glomeruli showed the loss of epithelial foot processes, formation of cytoplasmic vacuoles, microvillous formation and increased numbers of lysosomes in the cytoplasm of podocytes. It is strongly suggested that proteinuria in PAN nephrosis may be primarily due to a glomerular epithelial lesion, leading to focal disarray of anionic sites or focal defects in the epithelial covering of the basement membrane. The loss of anionic sites in the basement membrane nay be caused by the foot process fusion and the epithelial detachment from the basement membrane.

• Computers and Concrete
• /
• v.32 no.6
• /
• pp.615-623
• /
• 2023

Crack detection is an essential method to ensure the safety of dam concrete structures. Low-quality crack images of dam concrete structures limit the application of neural network methods in crack detection. This research proposes a modified attentional mechanism model to reduce the disturbance caused by uneven light, shadow, and water spots in crack images. Also, the focal loss function solves the small ratio of crack information. The dataset collects from the network, laboratory and actual inspection dataset of dam concrete structures. This research proposes a novel method for crack detection of dam concrete structures based on the U-Net neural network, namely AF-UNet. A mutual comparison of OTSU, Canny, region growing, DeepLab V3+, SegFormer, U-Net, and AF-UNet (proposed) verified the detection accuracy. A binocular camera detects cracks in the experimental scene. The smallest measurement width of the system is 0.27 mm. The potential goal is to achieve real-time detection and localization of cracks in dam concrete structures.

• Journal of Veterinary Science
• /
• v.24 no.2
• /
• pp.25.1-25.6
• /
• 2023

An eight-year-old Maltese dog presented with diarrhea and anorexia. Ultrasonography revealed marked focal wall thickening with loss of layering in the distal ileum. Contrast-enhanced computed tomography (CT) revealed a preserved wall layer with hypoattenuating middle wall thickening. In some segments of the lesion, small nodules protruding toward the mesentery from the outer layer were observed. Histopathology revealed focal lipogranulomatous lymphangitis (FLL) with lymphangiectasia. This is the first report to describe the CT features of FLL in a dog. CT features of preserved wall layers with hypoattenuating middle wall thickening and small nodules can assist in diagnosing FLL in dogs.