• 대한조현병학회지
• /
• 제24권1호
• /
• pp.36-43
• /
• 2021

Objectives: Despite the high discontinuation rate of clozapine in refractory schizophrenia, there is limited evidence regarding the suggested treatment after clozapine discontinuation. Methods: The medical records of 37 patients who discontinued clozapine were retrospectively reviewed. The prescription patterns of antipsychotics, mood stabilizers, and antidepressants were compared at three points before and after clozapine treatment and at the most recent visit. Results: After clozapine discontinuation, 75.6% of the subjects were receiving antipsychotic polypharmacy, and 32.4% were taking more than 3 antipsychotics. The frequently used antipsychotics were olanzapine (21.5%), quetiapine (21.5%), and paliperidone (12.7%). The rates of augmentation with mood stabilizers and antidepressants were 43.2% and 29.7%, respectively. Furthermore, valproate was the most commonly used mood stabilizer (87.5%). Conclusion: Antipsychotic polypharmacy and augmentation are inevitable in schizophrenia patients for whom clozapine has been discontinued. Further research is required to improve the outcomes of polypharmacy and augmentation in schizophrenia patients.

• 한국군사과학기술학회지
• /
• 제25권3호
• /
• pp.300-310
• /
• 2022

A vast amount of labeled data is required for deep neural network training. A typical strategy to improve the performance of a neural network given a training data set is to use data augmentation technique. The goal of this work is to offer a novel image augmentation method for improving object detection accuracy. An object in an image is removed, and a similar object from the training data set is placed in its area. An in-painting algorithm fills the space that is eliminated but not filled by a similar object. Our technique shows at most 2.32 percent improvements on mAP in our testing on a military vehicle dataset using the YOLOv4 object detector.

• 융복합기술연구소 논문집
• /
• 제11권1호
• /
• pp.1-5
• /
• 2021

Deep learning based methods achieve state-of-the-art accuracy, however, they typically rely on supervised training with large labeled datasets. It is known in many medical applications that labeling medical images requires significant expertise and much time, and typical hand-tuned approaches for data augmentation fail to capture the complex variations in such images. This paper proposes a 3D image augmentation method to overcome these difficulties. It allows us to enrich diversity of training data samples that is essential in medical image segmentation tasks, thus reducing the data overfitting problem caused by the fact the scale of medical image dataset is typically smaller. Our numerical experiments demonstrate that the proposed approach provides significant improvements over state-of-the-art methods for 3D medical image segmentation.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2021년도 제64차 하계학술대회논문집 29권2호
• /
• pp.645-647
• /
• 2021

본 논문에서는 목조건물의 Crack만을 움직여 Data set을 증강하는 기법을 제안한다. 이 기법은 이미지 내 Crack Detection의 학습 데이터를 만들기 위해 이미지의 전체적인 값으로 Flip, Rotation, Shift, Rescale 등의 변환을 통해 Data Augmentation을 진행하는 대신 Crack이라는 하나의 Object만을 가지고 새로운 데이터를 생성한다. 이때 Object는 관심 영역 내에서만 연산되어 기존의 방법보다 더욱 많은 데이터를 얻을 수 있으며, Crack이 관심 영역 밖으로 이동하지 않기 때문에 이상치 혹은 결측치가 존재하지 않는 데이터를 얻을 수 있다. 또한 Crack이 존재하지 않는 이미지에도 임의적으로 Crack을 생성하여 새로운 데이터를 만들 수 있다. 결론적으로 본 논문에서는 Crack Detection의 학습을 위하여 기존 방법보다 우수한 성능의 Data Augmentation을 제안하였다.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.694-696
• /
• 2021

본 논문에서는 CNN 모델에서 학습에 사용할 이미지 데이터를 늘리기 위해 사용되는 Mixup 알고리즘을 MLP 모델에 사용하는 데이터셋에 적용하여 data augmentation 효과를 얻을 수 있는 지에 대한 테스트를 수행했다. 테스트 결과 MLP 모델에 사용할 데이터셋에도 Mixup 알고리즘으로 data augmentation 효과를 기대할 수 있음을 보여준다.

• Journal of Information Processing Systems
• /
• 제20권2호
• /
• pp.159-172
• /
• 2024

The number of healthcare products available for pets has increased in recent times, which has prompted active research into wearable devices for pets. However, the data collected through such devices are limited by outliers and missing values owing to the anomalous and irregular characteristics of pets. Hence, we propose pet behavior recognition based on a hybrid one-dimensional convolutional neural network (CNN) and long short- term memory (LSTM) model using pet wearable devices. An Arduino-based pet wearable device was first fabricated to collect data for behavior recognition, where gyroscope and accelerometer values were collected using the device. Then, data augmentation was performed after replacing any missing values and outliers via preprocessing. At this time, the behaviors were classified into five types. To prevent bias from specific actions in the data augmentation, the number of datasets was compared and balanced, and CNN-LSTM-based deep learning was performed. The five subdivided behaviors and overall performance were then evaluated, and the overall accuracy of behavior recognition was found to be about 88.76%.

• International journal of advanced smart convergence
• /
• 제13권2호
• /
• pp.195-204
• /
• 2024

In this paper, we investigate the comparative performance of two leading object detection architectures, YOLOv7 and Detection Transformer (DETR), across varying levels of data augmentation using CycleGAN. Our experiments focus on chest scan images within the context of biomedical informatics, specifically targeting the detection of abnormalities. The study reveals that YOLOv7 consistently outperforms DETR across all levels of augmented data, maintaining better performance even with 75% augmented data. Additionally, YOLOv7 demonstrates significantly faster convergence, requiring approximately 30 epochs compared to DETR's 300 epochs. These findings underscore the superiority of YOLOv7 for object detection tasks, especially in scenarios with limited data and when rapid convergence is essential. Our results provide valuable insights for researchers and practitioners in the field of computer vision, highlighting the effectiveness of YOLOv7 and the importance of data augmentation in improving model performance and efficiency.

• 대한수학회보
• /
• 제45권1호
• /
• pp.101-103
• /
• 2008

The structure of $I_G^n/I_G^{n+1}$ for finite abelian group G is investigated.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• 제39권4호
• /
• pp.156-160
• /
• 2013

Objectives: Interest in bone graft material has increased with regard to restoration in cases of bone defect around the implant. Autogenous tooth bone graft material was developed and commercialized in 2008. In this study, we evaluated the results of vertical and horizontal ridge augmentation with autogenous tooth bone graft material. Materials and Methods: This study targeted patients who had vertical or horizontal ridge augmentation using AutoBT from March 2009 to April 2010. We evaluated the age and gender of the subject patients, implant stability, adjunctive surgery, additional bone graft material and barrier membrane, post-operative complication, implant survival rate, and crestal bone loss. Results: We performed vertical and horizontal ridge augmentation using powder- or block-type autogenous tooth bone graft material, and implant placement was performed on nine patients (male: 7, female: 2). The average age of patients was $49.88{\pm}12.98$ years, and the post-operative follow-up period was $35{\pm}5.31$ months. Post-operative complications included wound dehiscence (one case), hematoma (one case), and implant osseointegration failure (one case; survival rate: 96%); however, there were no complications related to bone graft material, such as infection. Average marginal bone loss after one-year loading was $0.12{\pm}0.19$ mm. Therefore, excellent clinical results can be said to have been obtained. Conclusion: Excellent clinical results can be said to have been obtained with vertical and horizontal ridge augmentation using autogenous tooth bone graft material.

• 대한두개안면성형외과학회지
• /
• 제19권2호
• /
• pp.94-101
• /
• 2018

Background: Atrophy of muscle and fat often contributes to temporal hollowing after pterional craniotomy. However, the main cause is from the bony defect. Several methods to prevent temporal hollowing have been introduced, all with specific limitations. Autologous bone grafts are most ideal for cranial defect reconstruction. The authors investigated the effectiveness of bony defect coverage and temporal augmentation using pterional craniotomy bone flap. Methods: This study was conducted in 100 patients who underwent brain tumor excision through pterional approach from 2015 to 2016. Group 1 underwent pterional craniotomy with temporal augmentation and group 2 without temporal augmentation. In group 1, after splitting the calvarial bone at the diploic space, the inner table was used for covering the bone defect and as an onlay graft for temporal augmentation. The outcome is evaluated by computed tomography at 1-year follow-up. Results: The mean operative time for temporal augmentation was 45 minutes. The mean follow-up was 12 months. The ratio of temporal thickness of operated side to non-operated side was 0.99 in group 1 and 0.44 in group 2, which was statistically different. The mean visual analogue scale score was 1.77 in group 1 and 6.85 in group 2. Conclusion: This study demonstrated a surgical technique using autologous bone graft for successfully preventing the temporal hollowing and improved patient satisfaction.