• 한국의학물리학회지:의학물리
• /
• 제28권3호
• /
• pp.77-82
• /
• 2017

We proposed the objective ROC analysis method based on the setting of threshold value for evaluation of single photon emission computed tomography (SPECT) image. This proposed ROC analysis method uses the quantification computational threshold value to each signal on the SPECT image. The SPECT images for this study were acquired by using Monte Carlo n-particle extended simulation code (MCNPX, Ver. 2.6.0, Los Alamos National Laboratory, USA). The basic SPECT detectors and specific water phantom were realized in the simulation, and we could get the simulation results by the simulation operation. We tried to analyze the reconstructed images using threshold value application based objective ROC method. We can get the accuracy information of reconstructed region in the image. This proposed ROC technique can be helpful when we have to evaluate the weak signal for the NM image. In this study, the proposed threshold value based computational ROC analysis method can provide better objectivity than the conventional ROC analysis method.

• Communications for Statistical Applications and Methods
• /
• 제31권2호
• /
• pp.203-212
• /
• 2024

The area under the ROC curve (AUC) is one of the most common criteria used to measure the overall performance of binary classifiers for a wide range of machine learning problems. In this article, we propose a L₁-penalized AUC-optimization classifier that directly maximizes the AUC for high-dimensional data. Toward this, we employ the AUC-consistent surrogate loss function and combine the L₁-norm penalty which enables us to estimate coefficients and select informative variables simultaneously. In addition, we develop an efficient optimization algorithm by adopting k-means clustering and proximal gradient descent which enjoys computational advantages to obtain solutions for the proposed method. Numerical simulation studies demonstrate that the proposed method shows promising performance in terms of prediction accuracy, variable selectivity, and computational costs.

• Genomics & Informatics
• /
• 제16권4호
• /
• pp.26.1-26.12
• /
• 2018

Shigella spp. constitutes some of the key pathogens responsible for the global burden of diarrhoeal disease. With over 164 million reported cases per annum, shigellosis accounts for 1.1 million deaths each year. Majority of these cases occur among the children of the developing nations and the emergence of multi-drug resistance Shigella strains in clinical isolates demands the development of better/new drugs against this pathogen. The genome of Shigella flexneri was extensively analyzed and found 4,362 proteins among which the functions of 674 proteins, termed as hypothetical proteins (HPs) had not been previously elucidated. Amino acid sequences of all these 674 HPs were studied and the functions of a total of 39 HPs have been assigned with high level of confidence. Here we have utilized a combination of the latest versions of databases to assign the precise function of HPs for which no experimental information is available. These HPs were found to belong to various classes of proteins such as enzymes, binding proteins, signal transducers, lipoprotein, transporters, virulence and other proteins. Evaluation of the performance of the various computational tools conducted using receiver operating characteristic curve analysis and a resoundingly high average accuracy of 93.6% were obtained. Our comprehensive analysis will help to gain greater understanding for the development of many novel potential therapeutic interventions to defeat Shigella infection.

• Interdisciplinary Bio Central
• /
• 제3권4호
• /
• pp.15.1-15.11
• /
• 2011

Background: Histone deacetylase (HDAC) 8 is one of its family members catalyzes the removal of acetyl groups from N-terminal lysine residues of histone proteins thereby restricts transcription factors from being expressed. Inhibition of HDAC8 has become an emerging and effective anti-cancer therapy for various cancers. Application computational methodologies may result in identifying the key components that can be used in developing future potent HDAC8 inhibitors. Results: Facilitating the discovery of novel and potential chemical scaffolds as starting points in the future HDAC8 inhibitor design, quantitative structure-activity relationship models were generated with 30 training set compounds using genetic function approximation (GFA) and Bayesian algorithms. Six GFA models were selected based on the significant statistical parameters calculated during model development. A Bayesian model using fingerprints was developed with a receiver operating characteristic curve cross-validation value of 0.902. An external test set of 54 diverse compounds was used in validating the models. Conclusions: Finally two out of six models based on their predictive ability over the test set compounds were selected as final GFA models. The Bayesian model has displayed a high classifying ability with the same test set compounds and the positively and negatively contributing molecular fingerprints were also unveiled by the model. The effectively contributing physicochemical properties and molecular fingerprints from a set of known HDAC8 inhibitors were identified and can be used in designing future HDAC8 inhibitors.

• 지능정보연구
• /
• 제27권3호
• /
• pp.57-73
• /
• 2021

ICT 인프라의 이상탐지를 통한 유지보수와 장애 예방이 중요해지고 있다. 장애 예방을 위해서 이상탐지에 대한 관심이 높아지고 있으며, 지금까지의 다양한 이상탐지 기법 중 최근 연구들에서는 딥러닝을 활용하고 있으며 오토인코더를 활용한 모델을 제안하고 있다. 이는 오토인코더가 다차원 다변량에 대해서도 효과적으로 처리가 가능하다는 것이다. 한편 학습 시에는 많은 컴퓨터 자원이 소모되지만 추론과정에서는 연산을 빠르게 수행할 수 있어 실시간 스트리밍 서비스가 가능하다. 본 연구에서는 기존 연구들과 달리 오토인코더에 2가지 요소를 가미하여 이상탐지의 성능을 높이고자 하였다. 먼저 다차원 데이터가 가지고 있는 속성별 특징을 최대한 부각하여 활용하기 위해 멀티모달 개념을 적용한 멀티모달 오토인코더를 적용하였다. CPU, Memory, network 등 서로 연관이 있는 지표들을 묶어 5개의 모달로 구성하여 학습 성능을 높이고자 하였다. 또한, 시계열 데이터의 특징을 데이터의 차원을 늘리지 않고 효과적으로 학습하기 위하여 조건부 오토인코더(conditional autoencoder) 구조를 활용하는 조건부 멀티모달 오토인코더(Conditional Multimodal Autoencoder, CMAE)를 제안하였다. 제안한 CAME 모델은 비교 실험을 통해 검증했으며, 기존 연구들에서 많이 활용된 오토인코더와 비교하여 AUC, Accuracy, Precision, Recall, F1-score의 성능 평가를 진행한 결과 유니모달 오토인코더(UAE)와 멀티모달 오토인코더(Multimodal Autoencoder, MAE)의 성능을 상회하는 결과를 얻어 이상탐지에 있어 효과적이라는 것을 확인하였다.