한미 지식재산권 협상의 역사적 고찰

임근영 세창출판사 2003 창작과 권리 Vol.- No.30

한국어의 PCT 국제공개어 채택을 위한 연구

임근영,Im, Geun-Yeong 한국발명진흥회 2007 發明特許 Vol.32 No.6

한국어의 PCT 국제공개어 채택을 위한 연구

임근영,Im, Geun-Yeong 한국발명진흥회 2007 發明特許 Vol.32 No.5

LED 항공등화 전원장치 고찰

임근영,이기환,장성규,이상철,신재흥 대한전기학회 2013 대한전기학회 학술대회 논문집 Vol.2013 No.10

신경회로망을 이용한 RF 스퍼터링 ZnO 박막 증착 프로세스 모델링

임근영,이상극,박춘배,Lim, Keun-Young,Lee, Sang-Keuk,Park, Choon-Bae 한국전기전자재료학회 2006 전기전자재료학회논문지 Vol.19 No.7

ZnO deposition parameters are not independent and have a nonlinear and complex property. To propose a method that could verify and predict the relations of process variables, neural network was used. At first, ZnO thin films were deposited by using RF magnetron sputtering process with various conditions. Si, GaAs, and Glass were used as substrates. The temperature, work pressure, and RF power of the substrate were $50\sim500^{\circ}C$, 15 mTorr, and $180\sim210W$, respectively : the purity of the target was ZnO 4 N. Structural properties of ZnO thin films were estimated by using XRD (0002) peak intensity. The structure of neural network was a form of 4-7-1 that have one hidden layer. In training a network, learning rate and momentum were selected as 0.2, 0.6 respectively. A backpropagation neural network were performed with XRD (0002) peak data. After training a network, the temperature of substrate was evaluated as the most important parameter by sensitivity analysis and response surface. As a result, neural network could capture nonlinear and complex relationships between process parameters and predict structural properties of ZnO thin films with a limited set of experiments.

세계특허통일화 과정에서 지역간·양자간 특허협력의 의미

임근영 세창출판사 2004 창작과 권리 Vol.- No.37

The idea of obtaining the same patent right for one invention worldwide was firstly conceived in Paris Convention for the Protection of Industrial Property(Paris Convention) in 1883, and the practical measure on procedural harmonization was taken with the sign of Patent Cooperation Treaty(PCT) in 1970. The multilateral approach to world patent harmonization has been specified in the consultation on PCT reform and Substantive Patent Law Treaty(SPLT) within World Intellectual Property Organization(WIPO) since 2000. Recently the regional and bilateral patent cooperation among United States Patent and Trademark Office(USPTO), Japan Patent Office(JPO), and European Patent Office(EPO) has been intensified. For example, the trilateral cooperation, which was embarked on to reduce the informationization cost in 1983, has been developed focusing on the exchange of database, search method, and examiners, to build up confidence on the search and examination results. And USPTO and JPO also initiated a joint pilot project for the practical exploitation of search and examination results in 2003, which was evaluated to be effective in reducing examination workload. In this context, this study throws a question on why the multilateral discussion on world patent harmonization has recently been heated up, and on why the regional and bilateral patent cooperation among developed countries such as US, Japan, and Europe, has ever been enhanced beyond the multilateral level. In order to answer it, this paper explains the brief history of multilateral patent cooperation, the background of and the movement towards world patent harmonization within WIPO. And it also investigates and analyzes the recent aspects of regional and bilateral patent cooperation, and its implication in the process of world patent harmonization.

미국의 21세기 전략계획과 미국 수수료 개정 논란

임근영 한국발명진흥회 2002 發明特許 Vol.27 No.12

임의 차원 데이터 대응 Dynamic RNN-CNN 멀웨어 분류기

임근영,조영복,Lim, Geun-Young,Cho, Young-Bok 한국정보통신학회 2019 한국정보통신학회논문지 Vol.23 No.5

본 연구는 본 연구는 Microsoft Malware Classification Challenge 데이터 셋을 사용해 임의의 길이 입력 데이터에 대응할 수 있는 멀웨어 분류 모델을 제안한다. 우리는 기존 연구의 멜웨어 데이터를 이미지화 시키는 것을 기반으로 한다. 제안 모델은 멀웨어 데이터가 큰 경우는 많은 이미지를 생성하고, 작은 데이터는 적은 이미지를 생성한다. 생성된 이미지를 시계열 데이터로 Dynamic RNN으로 학습시킨다. RNN의 출력 값은 Attention 기법을 응용해 가장 가중치가 높은 출력만 사용하고, RNN 출력값을 다시 Residual CNN으로 학습시켜 최종적으로 멀웨어를 분류한다. 제안모델을 실험한 결과 검증 데이터 셋에서 Micro-average F1 score 92%를 기록하였다. 실험 결과 특별한 특징 추출 및 차원 축소 없이 임의 길이의 데이터를 학습 및 분류할 수 있는 모델의 성능을 검증할 수 있었다. This study proposes a malware classification model that can handle arbitrary length input data using the Microsoft Malware Classification Challenge dataset. We are based on imaging existing data from malware. The proposed model generates a lot of images when malware data is large, and generates a small image of small data. The generated image is learned as time series data by Dynamic RNN. The output value of the RNN is classified into malware by using only the highest weighted output by applying the Attention technique, and learning the RNN output value by Residual CNN again. Experiments on the proposed model showed a Micro-average F1 score of 92% in the validation data set. Experimental results show that the performance of a model capable of learning and classifying arbitrary length data can be verified without special feature extraction and dimension reduction.

딥러닝과 Char2Vec을 이용한 문장 유사도 판별

임근영,조영복,Lim, Geun-Young,Cho, Young-Bok 한국정보통신학회 2018 한국정보통신학회논문지 Vol.22 No.10

본 연구는 자연어 처리 문제 중 하나인 문장 유사도 판별 문제를 딥러닝으로 해결하는 데에 있어 Char2Vec기반으로 문장을 전 처리하고 학습시켜 그 성능을 확인하고 대표적인 Word Embedding 모델 Word2Vec를 대체할 수 있는 가능성이 있는지 파악하고자 한다. 임의의 두 문장을 비교할 때 쓰는 딥러닝 구조로 Siamese Ma-STM 네트워크를 사용하였다. Word2Vec와 Char2Vec를 각각 기반으로 한 문장 유사도 판별 모델을 학습시키고 그 결과를 분석하였다. 실험 결과 Char2Vec를 기반으로 학습시킨 모델이 validation accuracy 75.1%을 보였고 Word2Vec를 기반으로 학습시킨 모델은 validation accuracy 71.6%를 보였다. 따라서 고 사양을 요구하는 Word2Vec대신 임베딩 레이어를 활용한 Char2Vec 기반의 전처리 모델을 활용함으로 분석 환경을 최적화 할 수 있다. The purpose of this study is to see possibility of Char2Vec as alternative of Word2Vec that most famous word embedding model in Sentence Similarity Measure Problem by Deep-Learning. In experiment, we used the Siamese Ma-LSTM recurrent neural network architecture for measure similarity two random sentences. Siamese Ma-LSTM model was implemented with tensorflow. We train each model with 200 epoch on gpu environment and it took about 20 hours. Then we compared Word2Vec based model training result with Char2Vec based model training result. as a result, model of based with Char2Vec that initialized random weight record 75.1% validation dataset accuracy and model of based with Word2Vec that pretrained with 3 million words and phrase record 71.6% validation dataset accuracy. so Char2Vec is suitable alternate of Word2Vec to optimize high system memory requirements problem.

동물 폐 표본에서의 해부학적 구조물과 병리적 병변: 미세 전산화단층촬영과 세절편 다중 검출기 전산화단층촬영에서의 발견능 비교

임근영,이현주,이창현,손규리,서준범,구진모,임정기 대한영상의학회 2006 대한영상의학회지 Vol.54 No.5

Purpose: We wanted to compare the capability of micro CT and the clinically available thin-slice multi-detector row CT (MDCT) for demonstrating fine anatomic structures and pathological lesions in formalin-fixed lung specimens. Materials and Methods: The porcine lung with shark liver oil-induced lipoid pneumonia and the canine lung with pulmonary paragonimiasis were fixed by ventilating them with formalin vapor, and they were then sliced into one-centimeter thick sections. Micro CT (section thickness, 18 micrometer) and MDCT (section thickness, 0.75 mm) images were acquired in four of the lung slices of the lipoid pneumonia specimen and in five of the lung slices of the paragonimiasis specimen. On 62 pairs of micro CT and MDCT images, 169 pairs of rectangular ROIs were manually drawn in the corresponding locations. Two chest radiologists recorded the detectability of three kinds of anatomic structures (lobular core structure, interlobular septum and small bronchiolar lumen) and two kinds of pathological lesions (ground-glass opacity and consolidation) with using a fivepoint scale. The statistical comparison was performed by using the Wilcoxon signed rank test. Interobserver agreement was evaluated with kappa statistics. Results: For all observers, all the kinds of anatomic structures and pathological lesions were detected better on the micro CT images than on the MDCT images (p<0.01). Agreement was fair between two observers (κ= 0.38, p<0.001). Conclusion: The fine anatomic structures and pathological lesions of the lung were more accurately demonstrated on micro CT than on thin-slice MDCT in the inflated and fixed lung specimens. 목적: 고정된 폐 표본을 이용하여 micro-CT에서 획득한 영상과 thin-slice MDCT에서 획득한 영상의 미세한 해부학적 구조물과 병변의 발견능을 비교하고자 한다. 대상과 방법: 리포이드 폐렴이 있는 돼지 폐 절편 4개와 폐흡충증이 있는 개 폐 5절편의 영상을 micro-CT와 MDCT를 이용하여 얻었다. 각각 대응하는 micro-CT 영상과 MDCT 영상으로 이루어진 62쌍의 영상에서 169쌍의 관심영역을 지정하였다. 각각의 관심영역에 대하여 해부학적 구조물과 병변을 두 명의 흉부 방사선과 의사가 평가 하였으며 그 결과를 Wilcoxon signed rank test를 이용하여 통계적 분석을 시행하였고 kappa statistics로 두 평가자 간의 일치도를 평가하였다. 결과: 모든 해부학적 구조물과 병변에서 thin-slice MDCT보다 micro-CT의 평가 척도 평균값이 통계적으로 유의하게 높았으며 (p < 0.01) 평가자 간의 일치도는 fair 였다(κ=0.38). 결론: 고정된 폐 표본의 미세한 해부학적 구조물과 병변에 있어서 micro-CT에서의 발견능이 MDCT에 비하여 더 우월하다.