http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
공유형 자율차의 이동 경로 예측에 관한 기초연구: AI 분석 기반의 택시 경로와 도시 빅데이터를 이용하여
김건욱(KIM, Keunwook),김정화(KIM, Junghwa),김우진(KIM, Woojin),이승현(LEE, Seunghyeon) 대한교통학회 2021 대한교통학회지 Vol.39 No.6
본 연구에서는 대구광역시를 대상으로 택시 DTG(Digital Tachograph) 데이터를 활용하여 공유형 자율주행차의 주행 형태가 택시와 유사하다는 대전제를 두고 공유형 자율주행차의 주요 이동 경로를 예측하기 위한 기초연구를 수행하였다. 택시 이용자의 승차지점부터 하차지점까지의 주행 경로가 기록된 DTG 데이터를 바탕으로 승차율 분석결과를 히트맵으로 시각화하였으며, 주행 경로를 통해 택시의 주요 이동 경로를 확인하였다. 승차율 분석결과 대구광역시 도심인 동성로, 부도심인 동대구역의 승차율이 높게 나타났으며 그다음으로 승차율이 높은 지역은 상업시설의 비중이 높은 지역으로 확인되었다. 분석결과 지하철 노선과 도시 내 주간선도로와 유사한 축으로 나타나는 경향을 확인하였으며 이를 통해 향후 간선도로와 보조 간선도로, 지하철 노선을 따라 공유형 자율주행차를 위한 인프라가 구축 될 수 있다는 시사점을 얻을 수 있었다. 또한 머신러닝 기법인 랜덤포레스트와 LightGBM를 활용한 회귀모형을 구축하여 택시 이동 경로 결정에 영향을 주는 인자를 분석하였다. 모형 추정결과 2개의 회귀모형 모두 적합성을 확보한 것으로 나타났으며, 지하철역까지의 거리, 간선도로, 유동인구 수, 카드매출 실적, 버스정류장까지의 거리가 주요 변수로 도출되었다. 장래 이와 같은 영향변수들의 변화를 고려하여 공유형 자율주행차의 주행과 관련된 필수 인프라를 전략적으로 배치하는 데 에 본 연구의 결과가 활용 될 수 있을 것으로 기대한다. This study analyzed the main travel routes of shared autonomous vehicles by using DTG(Digital Tachograph) data of Daegu Metropolitan City under the premise that the driving mode of the shared autonomous vehicle is similar to that of a taxi. Based on the taxi DTG data that recorded the driving route from the taxi user’s entry point to the exit point, the ride rate analysis result was visualized as a heat map. The main travel routes of taxi users were found through the driving route. As a result of the loading rate analysis, it was confirmed that the loading rate was high in Dongseong-ro, the downtown area of Daegu Metropolitan City, and the nearby area of Dongdaegu Station. The following high passenger rate region was analyzed as the region with a high proportion of commercial facilities. In the case of movement routes, it was confirmed that they appear mainly along the same axis as subway lines and major arterial roads in the city. And through this, it was possible to obtain the implication that the infrastructure for shared autonomous vehicles could be built along arterial roads, auxiliary arterial roads, and subway lines in the future. In addition, this study established a regression model using machine learning techniques such as Random forest and LightGBM to confirm the variables that affect the user’s choice to use a taxi. The distance to the subway station, the arterial road, the number of floating population, the card sales performance, and the distance to the bus stop were derived as the primary variables that affect the decision of the taxi route. As a result of model estimation, it was found that both regression models secured the fit. It is expected that the results of this study can be utilized to strategically arrange essential infrastructure related to the driving of shared autonomous vehicles in consideration of changes in these influencing variables.
Kim Seungsik,Gu Nami,Moon Jeongin,Kim Keunwook,Hwang Yeongeun,Lee Kyeongjun 한국통계학회 2023 Communications for statistical applications and me Vol.30 No.5
This study aimed to predict the number of meals served in a group cafeteria using machine learning methodology. Features of the menu were created through the Word2Vec methodology and clustering, and a stacking ensemble model was constructed using Random Forest, Gradient Boosting, and CatBoost as sub-models. Re-sults showed that CatBoost had the best performance with the ensemble model showing an 8% improvement in performance. The study also found that the date variable had the greatest influence on the number of diners in a cafeteria, followed by menu characteristics and other variables. The implications of the study include the potential for machine learning methodology to improve predictive performance and reduce food waste, as well as the removal of subjective elements in menu classification. Limitations of the research include limited data cases and a weak model structure when new menus or foreign words are not included in the learning data. Future studies should aim to address these limitations.
AHN, EUN HEE,KIM, DAE WON,SHIN, MIN JEA,RYU, EUN JI,YONG, JI IN,CHUNG, SEOK YOUNG,CHA, HYUN JU,KIM, SANG JIN,CHOI, YEON JOO,KIM, DUK-SOO,CHO, SUNG-WOO,LEE, KEUNWOOK,CHO, YOON SHIN,KWON, HYEOK YIL,PARK UNKNOWN 2016 INTERNATIONAL JOURNAL OF MOLECULAR MEDICINE Vol.38 No.1
<P>Antioxidant 1 (ATOX1) functions as an antioxidant against hydrogen peroxide and superoxide, and therefore may play a significant role in many human diseases, including diabetes mellitus (DM). In the present study, we examined the protective effects of Tat-ATOX1 protein on streptozotocin (STZ)-exposed pancreatic insulinoma cells (RINm5F) and in a mouse model of STZ-induced diabetes using western blot analysis, immunofluorescence staining and MTT assay, as well as histological and biochemical analysis. Purified Tat-ATOX1 protein was efficiently transduced into RINm5F cells in a dose-and time-dependent manner. Additionally, Tat-ATOX1 protein markedly inhibited reactive oxygen species (ROS) production, DNA damage and the activation of Akt and mitogen activated protein kinases (MAPKs) in STZ-exposed RINm5F cells. In addition, Tat-ATOX1 protein transduced into mice pancreatic tissues and significantly decreased blood glucose and hemoglobin A1c (HbA1c) levels as well as the body weight changes in a model of STZ-induced diabetes. These results indicate that transduced Tat-ATOX1 protein protects pancreatic beta-cells by inhibiting STZ-induced cellular toxicity in vitro and in vivo. Based on these findings, we suggest that Tat-ATOX1 protein has potential applications as a therapeutic agent for oxidative stress-induced diseases including DM.</P>
DLK1 suppresses melanoma growth by shaping tumor immune microenvironment
Misu Kim,Hyejin Min,Seongwon Pak,Dohyeon Chung,Yejin Lee,Bikash Thapa,Keunwook Lee 한국실험동물학회 2021 한국실험동물학회 학술발표대회 논문집 Vol.2021 No.7
Immunoediting is a dynamic process that changes host immune system from immunosurveillance to immune escape during tumor progression. Despite the recent advances in cancer immunotherapy targeting immune checkpoints, it remains unclear how tumor cells acquire an ability to shape a favorable immune landscape within tumor microenvironment. To identify immunoediting factors that promote tumor progression, we performed a comparative transcriptomic analysis using tumor tissues derived from syngeneic mice transplanted with B16 melanoma manipulated to accelerate the tumor growth, in comparison with mice received with control B16 cells. Among the tumor-intrinsic factors that were differentially expressed and putative immune-associated genes, delta-like noncanonical Notch ligand 1 (DLK1) was further validated its functional impact on tumor growth and immune microenvironment. Forced expression of DLK1 attenuated melanoma growth whereas knockdown of DLK1 enhanced the tumor growth in the transplanted mice. Of note, DLK1 expression did not promote tumor growth intrinsically as neither ectopic expression nor knockdown of DLK1 affect proliferation or apoptosis of B16 cells. On the other hand, we observed a marked changed in the immunophenotype within tumor tissues including increased infiltration of effector CD4 T cells and NK cells, and reduced myeloid cells. In vitro coculture assay demonstrated that DLK1 expression in B16 upregulated the ability of melanoma cells to promote migration of CD4 T cells and NK cells toward the tumor. It is in progress to investigate the mechanism by which tumors modulate the expression of DLK1 in the tumor microenvironment during cancer progression. Collectively, our experimental data suggest DLK1 as a potent anti-tumor immunomodulating factor derived from tumor itself and provide a novel therapeutic venue for cancer treatment by reshaping the tumor immune microenvironment.
Bikash Thapa,Dong-Gyu Kim,Seongwon Pak,Dohyeon Chung,Jungwoo Shin,Keunwook Lee 한국실험동물학회 2021 한국실험동물학회 학술발표대회 논문집 Vol.2021 No.7
Activated B cells in the germinal center (GC) undergo clonal expansion along with immunoglobulin isotype switching and affinity maturation, leading to differentiation into plasma cells that produce appropriate humoral immunity against invading pathogen. Although PI3K/mTOR signaling is indispensable for survival and proliferation of B cells, little is known about how this signaling pathway integrates environmental milieu within the GCs to regulate GC reactions and to yield plasma cells. In this study, we dissected the impact of PI3K/mTOR signaling pathway on antibody responses by combination of the inducible deletion of Pten (a PI3K-antagonistic lipid phosphatase), Raptor (a subunit of mTORC1), and Rictor (a subunit of mTORC2). Rag2-/- mice were reconstituted with naïve B cells isolated from the inducible KO mice along with normal CD4 T cells and then immunized with NP-ovalbumin. Mice that received Pten depleted B cells exhibited impaired class switching to IgG1 and production of high affinity antibodies against the NP antigen in the sera. On the other hand, inactivation of mTORC2 by depleting Rictor under Pten loss in B cells rescued IgG1 class switching and affinity maturation of NP-specific antibodies whereas that of mTORC1 by depletion Raptor did not. To delineate the mechanism, we adapted a co-culture system to generate GC B cells and plasmablasts (PBs) ex vivo. B cells depleted Pten differentiated into PBs expressing predominantly IgM and less IgG1 in the co-cultivation which was comparable to the control PBs. Finally, we performed transcriptomic analysis of PB cells and mapped to the set of B cell function-associated genes under the regulation of the PI3K-PTEN-mTORC2 signaling network. It is in progress to determine whether this signaling modulation within B cell compartment is required for protective antibody responses against influenza virus. Collectively, out data provide a mechanistic insight by which PTEN and mTORC2 tune up the PI3K signaling pathway to promote class-switching and affinity maturation of antibodies during GC reaction.