http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
페라이트상을 갖는 마르텐사이트계 내열강의 미세조직 및 기계적성질에 미치는 시효처리의 영향
이상명(S. M. Lee),이정규(J. K,Lee),조영관(Y. K. Jo),이종문(J. M. Lee),강창룡(C. Y. Kang) 한국동력기계공학회 2009 한국동력기계공학회 학술대회 논문집 Vol.2009 No.11
This study was carried out to investigate the specimen in martensite structure added Mo and not added Mo on the aging treatment were generated each changes of microstructure and mechanical properties, the result obtained from this study are as follow: In case of specimen added Mo, A Laves phase was mainly precipitated in ferrite phase, while carbide and Laves phase were overlapping precipitated in martensite phase, on the interface of martensite and ferrite phase. In case of specimen not add Mo, A microstructure was unchanged in ferrite phase, while carbide was precipitated in martensite phase. In case of specimen added Mo, A ferrite hardness change existed two peaks with the increase of the aging time, while a martensite hardness change which a hardness in the early stage of aging was low, but a re-hardening was occurred with the increase of the aging time. In case of specimen not add Mo, While a ferrite hardness change of aging was not largely changed, a ferrite hardness change was sharply decreased in the early stage of aging, but it was slowly decreased with the increase of the aging time. The tensile strength of specimen added Mo was eventually decreased with the increase of the aging time, while the tensile strength of specimen not added Mo was slowly decreased. A Sharpy impact's value of specimen added Mo was largely decreased by the aging treatment, while it was slowly increased with the increase of the aging time, in case of specimen not add Mo was gradually increased in the early stage of aging, the aging time was long, then it was quickly increased. The Disk shape of Laves phase precipitation in the early stage of aging was largely decreased the impact toughness .
임의의 잡음 신호 추가를 활용한 적대적으로 생성된 이미지 데이터셋 탐지 방안에 대한 연구
황정환,윤지원 한국정보전자통신기술학회 2019 한국정보전자통신기술학회논문지 Vol.12 No.6
In Deep Learning models derivative is implemented by error back-propagation which enables the model to learn the error and update parameters. It can find the global (or local) optimal points of parameters even in the complex models taking advantage of a huge improvement in computing power. However, deliberately generated data points can ‘fool’ models and degrade the performance such as prediction accuracy. Not only these adversarial examples reduce the performance but also these examples are not easily detectable with human’s eyes. In this work, we propose the method to detect adversarial datasets with random noise addition. We exploit the fact that when random noise is added, prediction accuracy of non-adversarial dataset remains almost unchanged, but that of adversarial dataset changes. We set attack methods (FGSM, Saliency Map) and noise level (0-19 with max pixel value 255) as independent variables and difference of prediction accuracy when noise was added as dependent variable in a simulation experiment. We have succeeded in extracting the threshold that separates non-adversarial and adversarial dataset. We detected the adversarial dataset using this threshold. 여러 분야에서 사용되는 이미지 분류를 위한 딥러닝(Deep Learning) 모델은 오류 역전파 방법을 통해 미분을 구현하고 미분 값을 통해 예측 상의 오류를 학습한다. 엄청난 계산량을 향상된 계산 능력으로 해결하여, 복잡하게 설계된 모델에서도 파라미터의 전역 (혹은 국소) 최적점을 찾을 수 있다는 것이 장점이다. 하지만 정교하게 계산된 데이터를 만들어내면 이 딥러닝 모델을 ‘속여’ 모델의 예측 정확도와 같은 성능을 저하시킬 수 있다. 이렇게 생성된 적대적 사례는 딥러닝을 저해할 수 있을 뿐 아니라, 사람의 눈으로는 쉽게 발견할 수 없도록 정교하게 계산되어 있다. 본 연구에서는 임의의 잡음 신호를 추가하는 방법을 통해 적대적으로 생성된 이미지 데이터셋을 탐지하는 방안을 제안한다. 임의의 잡음 신호를 추가하였을 때 일반적인 데이터셋은 예측 정확도가 거의 변하지 않는 반면, 적대적 데이터셋의 예측 정확도는 크게 변한다는 특성을 이용한다. 실험은 공격 기법(FGSM, Saliency Map)과 잡음 신호의 세기 수준(픽셀 최댓값 255 기준 0-19) 두 가지 변수를 독립 변수로 설정하고 임의의 잡음 신호를 추가하였을 때의 예측 정확도 차이를 종속 변수로 설정하여 시뮬레이션을 진행하였다. 각 변수별로 일반적 데이터셋과 적대적 데이터셋을 구분하는 탐지 역치를 도출하였으며, 이 탐지 역치를 통해 적대적 데이터셋을 탐지할 수 있었다.