• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.162-167
• /
• 2023

본 논문은 거대언어모델에 대한 한국어 상식추론 기반의 새로운 평가 방식을 제안한다. 제안하는 평가 방식은 한국어의 일반 상식을 기초로 삼으며, 이는 거대언어모델이 주어진 정보를 얼마나 잘 이해하고, 그에 부합하는 결과물을 생성할 수 있는지를 판단하기 위함이다. 기존의 한국어 상식추론 능력 평가로 사용하던 Korean-CommonGEN에서 언어 모델은 이미 높은 수준의 성능을 보이며, GPT-3와 같은 거대언어모델은 사람의 상한선을 넘어선 성능을 기록한다. 따라서, 기존의 평가 방식으로는 거대언어모델의 발전된 상식추론 능력을 정교하게 평가하기 어렵다. 더 나아가, 상식 추론 능력을 평가하는 과정에서 사회적 편견이나 환각 현상을 충분히 고려하지 못하고 있다. 본 연구의 평가 방법은 거대언어모델이 야기하는 문제점을 반영하여, 다가오는 거대언어모델 시대에 한국어 자연어 처리 연구가 지속적으로 발전할 수 있도록 하는 상식추론 벤치마크 구성 방식을 새롭게 제시한다.

• The Korean Journal of Food & Health Convergence
• /
• v.8 no.2
• /
• pp.1-6
• /
• 2022

Purpose: This study aims to examine the opportunities in the food industry in the context of non-fungible tokens (NFTs) and the metaverse. The metaverse expands offline experiences into a virtual space and provides users with the advantage of overcoming the limitations of physical space. Research design, data and methodology: With the development of various contents based on advanced technology, the metaverse is becoming popular among the MZ generation (comprising millennials and Gen Zs), who are skilled at navigating digital platforms. Moreover, the previously entertainment game focused service has expanded to the business field after COVID-19; activities on the metaverse are now making space for economic value creation. I examined the recent global trends in the context of NFTs and the metaverse as well as the existing cases in Korea. Results: Globally, gaming and performance platforms such as Roblox, Fortnite etc. are still concentrated in the entertainment industry. However, as shown by ZEPETO and ifland in Korea there are many opportunities in the retail and food industries. Conclusions: We analyzed the ecosystems of the retail and food and beverage sectors, where the metaverse will be useful in the future, and have suggested the implications of the opportunities provided by the metaverse.

• Korean Journal of Weed Science
• /
• v.18 no.2
• /
• pp.154-160
• /
• 1998

This study was conducted to develop a foliar applied herbicide mixture covering wide spectrum weeds in dry direct seeded rice. Uniform precision central composite design(UPCCD) was employed to evaluate the effect of herbicide mixture among cyhalofop-butyl(cyhalofop), bispyribenzoxim and pyrazosulfuron-ethyl(pyrazosulfuron) having different weed control spectrum and to determine the best application rates of the three herbicide mixtures. The partial additive effect was observed in the mixture of cyhalofop and bispyribenzoxim with pyrazosulfron against Echinochloa crus-galli Beauv var crus-galli L., attributed to the negative interaction between cyhalofop and bispyribenzoxim system. Additive effect was observed in the mixture of bispyribenzoxim and pyrazosulfuron against Cyperus serotinus R. and Aneilema keisak H. but cyhalofop had no effect on these weeds. The most appropriate rate of the mixture of cyhalofop with bispyribenzoxim and pyrazosulfuron was 100 : 12 : 10g ai/ha exhibiting $ED_{90}$ against three weeds such as E. crus-galli, C. serotinus and A. keisak.

• Journal of Microbiology and Biotechnology
• /
• v.20 no.1
• /
• pp.208-216
• /
• 2010

Fusarium verticillioides is an important pathogen of maize, being responsible for ear rots, stalk rots, and seedling blight worldwide. During the past decade, F. verticillioides has caused several severe epidemics of maize seedling blight in many areas of China, which lead to significant losses. In order to understand the molecular mechanisms regulating fungal development and pathogenicity in this pathogen, we isolated and characterized the gene fpk1 (GenBank Accession No. EF405959) encoding a homolog of the cAMP-dependent protein kinase catalytic subunit, which included a 1,854-bp DNA sequence from ATG to TAA, with a 1,680-bp coding region, and three introns (lengths: 66 bp, 54 bp, and 54 bp), and the predicated protein precursor had 559 aa. The mutant ${\Delta}fpk1$, which was disrupted of the fpkl gene, showed reduced vegetative growth, fewer and shorter aerial mycelia, strongly impaired conidiation, and reduced spore germination rate. After germinating, the fresh hypha was stubby and lacking of branch. When inoculated in susceptible maize varieties, the infection of the mutant ${\Delta}fpk1$ was delayed and the infection efficiency was reduced compared with that of the wild-type strain. AU this indicated that gene fpk1 participated in hyphal growth, conidiophore production, spore germination, and virulence in F. verticillioides.

• Journal of Radiation Protection and Research
• /
• v.44 no.4
• /
• pp.149-155
• /
• 2019

Background: Magnetic resonance (MR) image guided radiation therapy system, enables real time MR guided radiotherapy (RT) without additional radiation exposure to patients during treatment. However, MR image lacks electron density information required for dose calculation. Image fusion algorithm with deformable registration between MR and computed tomography (CT) was developed to solve this issue. However, delivered dose may be different due to volumetric changes during image registration process. In this respect, synthetic CT generated from the MR image would provide more accurate information required for the real time RT. Materials and Methods: We analyzed 1,209 MR images from 16 patients who underwent MR guided RT. Structures were divided into five tissue types, air, lung, fat, soft tissue and bone, according to the Hounsfield unit of deformed CT. Using the deep learning model (U-NET model), synthetic CT images were generated from the MR images acquired during RT. This synthetic CT images were compared to deformed CT generated using the deformable registration. Pixel-to-pixel match was conducted to compare the synthetic and deformed CT images. Results and Discussion: In two test image sets, average pixel match rate per section was more than 70% (67.9 to 80.3% and 60.1 to 79%; synthetic CT pixel/deformed planning CT pixel) and the average pixel match rate in the entire patient image set was 69.8%. Conclusion: The synthetic CT generated from the MR images were comparable to deformed CT, suggesting possible use for real time RT. Deep learning model may further improve match rate of synthetic CT with larger MR imaging data.

• Korean Journal of Weed Science
• /
• v.18 no.3
• /
• pp.197-204
• /
• 1998

This study was conducted to determine the activity of ALS, content of endogenous free amino acids and fatty acids affected by herbicide mixture of cyhalofop-butyl, pyribenzoxim and pyrazosulfuron-ethyl. $I_{50}$ values (concentration required for 50% inhibition of ALS activity) of pyribenzoxim herbicide on the activity of ALS in Echinochloa crus-galli and Cyperus serotinus in vitro were recorded at 4${\times}$100 nM and 5${\times}$10 nM, respectively, while $I_{50}$ values of pyrazosulfuron against E crus-galli and C. serotinus were 4.5${\times}$10 nM and 4${\times}$10 nM, respectively, and the mixture of two herbicides showed additive effect on ALS activity at the low application rate, and independent effect at the high application rates of two herbicides. The inhibition rates of the three herbicides mixture treatment on the three branch-chain amino acids such as valine, leucine and isoleucine were 74.6%, 66.6% and 57.9% in C. serotinus and 36.6% 51.1% and 48.1% in E. crus-galli, respectively. A little bit higher inhibitory effect on the three branch-chain amino acids in C. serotinus and E. crus-galli seedlings was observed in two herbicide mixture of pyribenzoxim with pyrazosulfuron than three herbicide mixture of cyhalofop with pyribenzoxim and pyrazosulfuron. The interaction among three herbicides showed non-antagonism on the amounts of endogenous free amino acids.