128 research outputs found
Convergence and Implicit Bias of Gradient Descent on Continual Linear Classification
No full textWe study continual learning on multiple linear classification tasks by sequentially running gradient descent (GD) for a fixed budget of iterations per each given task. When all tasks are jointly linearly separable and are presented in a cyclic/random order, we show the directional convergence of the trained linear classifier to the joint (offline) max-margin solution. This is surprising because GD training on a single task is implicitly biased towards the individual max-margin solution for the task, and the direction of the joint max-margin solution can be largely different from these individual solutions. Additionally, when tasks are given in a cyclic order, we present a non-asymptotic analysis on cycle-averaged forgetting, revealing that (1) alignment between tasks is indeed closely tied to catastrophic forgetting and backward knowledge transfer and (2) the amount of forgetting vanishes to zero as the cycle repeats. Lastly, we analyze the case where the tasks are no longer jointly separable and show that the model trained in a cyclic order converges to the unique minimum of the joint loss function
Glutamate Permeability of Chicken Best1
No full textBestrophin-1 (Best1) is a calcium (Ca(2+))-activated chloride (Cl(-)) channel which has a phylogenetically conserved channel structure with an aperture and neck in the ion-conducting pathway. Mammalian mouse Best1 (mBest1) has been known to have a permeability for large organic anions including gluconate, glutamate, and D-serine, in addition to several small monovalent anions, such as Cl(-), bromine (Br(-)), iodine (I(-)), and thiocyanate (SCN(-)). However, it is still unclear whether non-mammalian Best1 has a glutamate permeability through the ion-conducting pathway. Here, we report that chicken Best1 (cBest1) is permeable to glutamate in a Ca(2+)-dependent manner. The molecular docking and molecular dynamics simulation showed a glutamate binding at the aperture and neck of cBest1 and a glutamate permeation through the ion-conducting pore, respectively. Moreover, through electrophysiological recordings, we calculated the permeability ratio of glutamate to Cl(-) (P(Glutamate)/P(Cl)) as 0.28 based on the reversal potential shift by ion substitution from Cl(-) to glutamate in the internal solution. Finally, we directly detected the Ca(2+)-dependent glutamate release through cBest1 using the ultrasensitive two-cell sniffer patch technique. Our results propose that Best1 homologs from non-mammalian (cBest1) to mammalian (mBest1) have a conserved permeability for glutamate
Genome editing methods in animal models
No full text© 2020 The Author(s). Genetically engineered animal models that reproduce human diseases are very important for the pathological study of various conditions. The development of the clustered regularly interspaced short palindromic repeats (CRISPR) system has enabled a faster and cheaper production of animal models compared with traditional gene-targeting methods using embryonic stem cells. Genome editing tools based on the CRISPR-Cas9 system are a breakthrough technology that allows the precise introduction of mutations at the target DNA sequences. In particular, this accelerated the creation of animal models, and greatly contributed to the research that utilized them. In this review, we introduce various strategies based on the CRISPR-Cas9 system for building animal models of human diseases and describe various in vivo delivery methods of CRISPR-Cas9 that are applied to disease models for therapeutic purposes. In addition, we summarize the currently available animal models of human diseases that were generated using the CRISPR-Cas9 system and discuss future directions11Nsciescopuskc
Automotive cluster UI design based on time and path from the perspective of the 'Passenger'
No full textBackground One important design issue is the examination of how the user interface (UI) supports the new user role in future mobility. However, there are few design studies on the passenger's cognitive needs and behavior in autonomous vehicles (AVs) based on empirical data. There is no doubt that autonomous mobility technologies are growing. The technology is already aiding the driving experience, and it will change the mobility culture and the transition of 'driver' into 'passenger.' This study is based on the premise that future AVs are capable of performing all driving tasks and proposes a set of passenger-centered automotive cluster UI designs for future mobility employing two factors: time and path. A collection of empirical data is provided to understand the passenger's perspective. Methods In this study, a solid set of empirical data on the cognitive needs of passengers is collected. Human cognitive characteristics and driving tasks are investigated from various viewpoints to understand the passenger's perspective. The cognitive relationship in the driving environment is analyzed through a literature review on situation awareness (SA) and structuring of the data flow framework. The framework is further explored by connecting the technological role transformation to the passenger. Three sets of user tests and in-depth interviews were undertaken to construct the empirical database on the passenger. The user tests were designed employing the Wizard of Oz method, and the results were summarized using descriptive and exploratory analysis. Based on these insights, a set of UI designs from the perspective of the passenger was proposed, and usability tests were conducted to verify its effectiveness and usability. Results The results of the tests demonstrate that a major percentage of the information request was related to time (current time and duration) and path (vehicle location and surroundings). Based on the data, a UI framework was built. Two usage scenarios were designed, time-full and time-less, for better in-situation comprehension. Time- and route-based UI were proposed to flow with the scenarios. A usability test was conducted, and a passenger's cognitive framework was defined. There are two aspects to this study: the data flow frameworks of the driver/passenger, and the UI design proposal for the future AVs. The detailed data relationship between the user, the vehicle, and the road is determined as a framework. Also, the driver's data perceiving flow is analyzed to understand the data flow change due to AVs. Furthermore, the cognitive framework of the passenger was proposed based on the data. Conclusions This study provides a solid understanding of drivers' emerging needs when they are relieved of the cognitive burden of driving tasks. The UI features for AVs are introduced based on the empirical data and research related to the provision of better situation awareness, focusing on time and location. This study contributes to the extant literature by observing the perspective of passengers in autonomous vehicles based on a qualitative study. The proposed UI design will be further explored as a communication method between the system and the passive user in future mobility.N
Ever-Changing Mind
No full textEverything around us is changing fast. We sometimes find ourselves feeling confused and fragile in this world. The purpose of the design was to reflect our ever-changing minds. We created various images of the textile designs that express different emotional statuses, and then displayed them in a blank dress.</p
오피스빌딩 퍼실리티 매니지먼트의 균형 잡힌 성과 측정 모델
No full text학위논문 (박사) -- 서울대학교 대학원 : 공과대학 건축학과, 2020. 8. 이현수.Facility management is gaining more importance as office building becomes high-rise and large, its facilities and systems get more sophisticated, and performance requirements on office building increase. As financial market evolves, office building has been recognized as one of the preferred alternative assets, and the importance of facility management has been increasingly emphasized due to its mission to fulfill the expectations and requirements of tenants on environment and services office building provides under financial constraints imposed by building owner, usually a financial investor in case of commercial office building. Furthermore, facility management has expanded to the core business area that affects the profits of office building.
Performance measurement of office building facility management should consider the conflicts generated from the different objectives of building owner as investor and tenant as user as well as the characteristics of building as physical property. Thus, it requires the use of a balanced, holistic, and multi-faceted performance measurement model covering various perspectives.
The author derived five performance factors that comprise Financial, Function, Organization, Safety-Health-Environment, and Satisfaction. They represent the perspectives of building owner and tenants, and reflect the role of facility management organization that drives the facility management performance as well as the characteristics of the facility management activities.
The author found that prioritization of five factors can be different depending on building grade and investigated the relationship of five factors. As a result, five factors cycle structure is presented, which can progressively improve facility management performance by resolving the conflicts that exist amongst the stakeholders.
To derive Key Performance Indicators (KPIs) for measuring the performance of five factors, a Delphi survey was conducted. The Delphi panels were composed of facility management experts and building owners. Thirty KPIs commonly recognized as important by both groups were identified.
The author developed FFOSS model that measures office building facility management performance, based on the thirty KPIs which were categorized into five performance factors. FFOSS model is a balanced and full-fledged performance measurement model that covers various facets of facility management and presents a cycle structure amongst the five factors.
The cost-effective expenditure made to the facilities of building is evaluated as important financial performance, which is a salient feature of this model distinguishing it from existing model. Investment made by building owner on buildings function and facility management organization leads to improved performance of the buildings physical facilities, and together with rising capability of the organization, it enhances Safety-Health-Environment performance. Then, tenant satisfaction rises, and eventually a better financial performance is achieved through rising rents paid by satisfied tenants.
FFOSS model provides a balanced evaluation framework to office building owners and can be used by customizing five factors weightings. Weighting of five factors is a strategic method of tailored performance evaluation that can reflect the varying characteristics by building grades. The author proposes Financial and Function priority strategy as a standard for general building and Satisfaction and Safety priority strategy for prime grade office building. It also allows building owners to effectively communicate their objectives and intentions to facility management organization.
The author verified the applicability, adequacy and effectiveness of FFOSS model through experts evaluation. The evaluation scores showed the validity of model. In order to increase the objectivity, a comparison model was also selected and evaluated at the same time. But it failed to show as much validity as FFOSS model as it obtained low evaluation score and was assessed to have poor applicability. Also, the performance measurement of facility management of six office buildings was conducted with FFOSS model. The measurement score for each building was verified as valid, representing the current overall facility management performance in view of the grade of building.퍼실리티 매니지먼트 (Facility management)는 건축물이 고층, 대형화되고 설비와 시스템이 정교해지며, 성능에 대한 요구사항이 증대됨에 따라 중요도가 높아지고 있다. 금융시장의 발전에 따라 오피스빌딩은 투자자산으로서 선호되고 있으며, 퍼실리티 매니지먼트는 상업용 오피스빌딩의 투자 수익을 좌우하는 핵심적인 비즈니스 영역으로 확장되었다. 이에 투자자의 수익을 위한 재정적인 제약 가운데에서 사용자인 기업 테넌트 (Tenant)의 업무 환경과 서비스에 대한 다양한 필요와 기대를 충족시켜야 하므로 그 역할이 더욱 강조되고 있다.
오피스빌딩 퍼실리티 매니지먼트의 성과 측정은 건축물로서의 물리적인 특성을 반영하고, 퍼실리티 매니지먼트 역할과 책임의 넓은 범위를 포괄하며, 소유자인 투자자의 특성과 사용자인 기업 테넌트의 이해 충돌을 인정하는 균형 잡힌 성과 측정 모델의 사용이 요구된다.
저자는 오피스빌딩 퍼실리티 매니지먼트 조직에 요구되는 다양한 성과의 영역을 구분하는 5개 인자 (Factor)를 도출하고, 그 우선순위와 상호 관계를 규명하여, 5개 인자의 순환 구조에 기반한 성과 측정 모델을 개발하였다.
퍼실리티 매니지먼트 성과의 5개 인자는 재정 (Financial), 기능 (Function), 조직 (Organization), 안전-건강-환경 (Safety-Health-Environment), 만족 (Satisfaction)으로 도출되었다. 이는 성과 평가 주체인 빌딩 소유자 및 사용자인 테넌트의 관점과 함께, 성과를 이끌어내는 퍼실리티 매니지먼트 조직의 역할과 업무의 특성을 반영한 것이다.
빌딩 소유자는 빌딩 비용의 절감을 통하여 재정적 성과를 극대화하고자 한다. 그러나 오피스빌딩은 지속적인 유지관리 노력과 비용이 투입되지 않으면 노후화가 빠르게 진행되는 건축물로서의 물리적 특성을 갖고 있기 때문에, 이러한 접근은 테넌트의 만족도를 저해하여 오히려 수익을 손상시킬 수 있다. 이에 본 저자는 빌딩의 수익 증대에 기여하는 유지관리 비용의 효용적 지출을 퍼실리티 매니지먼트의 중요한 성과로 평가함으로서, 균형 잡힌 재정적 성과 평가의 관점을 제시하였다.
빌딩 소유자의 빌딩 기능과 퍼실리티 매니지먼트 조직 역량에 대한 적절한 재정적 투자가 빌딩의 기능을 향상시키고, 테넌트가 필요로 하는 안전-건강-환경 성과를 제고한다. 이러한 성과를 바탕으로 테넌트 만족이 증진될 수 있으며, 만족한 테넌트가 지불하는 임대료를 통하여 빌딩 소유자가 추구하는 재정적 성과가 성취된다. 이러한 순환은 빌딩 오너와 테넌트의 이해 충돌을 발전적으로 해소하면서 퍼실리티 매니지먼트 성과를 총체적으로 증진한다.
또한 저자는 퍼실리티 매니지먼트 5개 인자의 성과를 가장 효과적으로 측정하는 주요 성과 지표들 (Key performance indicators)을 도출하여 성과 측정 모델을 완성하였다. 전문가 델파이 그룹을 빌딩 소유자 측 전문가 7인과 빌딩 퍼실리티 매니지먼트 조직 팀장급 7인으로 구성하여, 3차에 걸친 설문을 통하여 빌딩 소유자와 퍼실리티 매니지먼트 조직이 모두 중요성을 인정한 주요 성과 지표 30개를 규명하여 성과 측정 기준과 함께 제시하였다.
FFOSS모델은 각 인자들의 우선순위와 평가 비중을 평가자가 직접 설정할 수 있으며, 저자는 오피스 빌딩 등급의 특성을 반영한 두 가지의 평가 전략을 제시하였다. 재정과 기능을 우선시하여 높은 비중으로 평가하는 일반 등급 빌딩을 위한 표준 전략과 함께, 테넌트 만족과 안전을 우선시하는 프라임 등급 빌딩을 위한 전략이 제시되었다. 평가 비중의 설정을 통해 빌딩 소유자는 빌딩 운영 우선순위를 퍼실리티 매니지먼트 조직에 효과적으로 전달할 수 있다.
FFOSS 모델의 5개 인자와 30개 주요 성과 지표를 사용한 성과 측정은 저성과의 요인을 지적하는 동시에 해당 저성과의 영향 범위가 제한적인지 아니면 광범위한지를 알려 준다. 따라서 빌딩 소유자가 한정된 자원을 성과 개선에 효과적으로 배분하기 위한 의사 결정 도구로서도 효과적이다.
모델의 검증을 위하여 선행연구들 중 완성도가 가장 높은 비교모델을 선정하여 빌딩 소유자 측 전문가들을 통하여 두 모델을 동일한 기준으로 평가하도록 하였다. 평가 결과 FFOSS 모델의 적용성, 적정성, 효과성이 모두 우수함이 입증되었으며. 비교모델보다 높은 평가 점수로 상대적인 우수성도 입증되었다.
또한 6개 오피스빌딩 퍼실리티 매니지먼트의 성과를 두 모델을 사용하여 측정하도록 하였으며, 비교모델은 실제로 평가가 불가한 지표들이 많아서 종합 점수가 산출되지 못하였다. 그러나 FFOSS 모델은 모든 빌딩들에서 종합 점수가 산출되었으며, 해당 빌딩의 등급과 상태를 감안할 때 현행 성과에 부합하는 점수임에, 실제로 성과 측정과 적용이 우수한 모델임이 입증되었다.Chapter 1. Introduction 1
1.1 Research Background 1
1.2 Problem Statement 4
1.3 Research Objective 6
1.4 Research Methodology 8
1.5 Dissertation Outline 11
Chapter 2. Preliminary Research 13
2.1 Performance Measurement Definition 13
2.2 Researches on FM Performance Measurement 16
2.2.1 Models from Other Industry 19
2.2.2 Models Focused to Building Facilities 24
2.2.3 Models Reflecting FM Context 29
2.3 Identification of FM Stakeholders 39
2.4 Summary 42
Chapter 3. Model Components 45
3.1 Features of FM Performance 46
3.2 Identification of Five Factors 49
3.2.1 FM Stakeholders Perspectives 49
3.2.2 Idenfitication of Five Factors 52
3.3 Derivation of Potential KPIs 59
3.4 Derivation of Longlist KPIs 62
3.5 Summary 67
Chapter 4. Model Development 69
4.1 Five Factors Priority and Relationship 70
4.1.1 Hypothesis of Five Factors Priority 70
4.1.2 Five Factors Relationship 78
4.1.3 Questionnaire Survey for Five Factors 82
4.2 Five Factors Cycle and Strategy 88
4.2.1 Five Factors Cycle 88
4.2.2 Strategy for Prime Grade Office Building 90
4.3 Delphi Survey for KPIs Derivation 92
4.3.1 Delphi Survey Design 92
4.3.2 Delphi Questionnaire and Analysis 95
4.3.3 Final KPIs 110
4.4 Description of KPIs 112
4.4.1 KPIs of Financial Factor 112
4.4.2 KPIs of Function Factor 115
4.4.3 KPIs of Organization Factor 119
4.4.4 KPIs of Safety-Health-Environment Factor 122
4.4.5 KPIs of Satisfaction Factor 126
4.5 FFOSS Model 129
4.5.1 Features of FFOSS Model 129
4.5.2 Simulation of FFOSS Model 142
4.6 Summary 148
Chapter 5. Model Validation 151
5.1 Evaluation of Model 152
5.1.1 Evaluation Design 152
5.1.2 Evaluation Results 156
5.2 Application of Model 159
5.3 Summary 168
Chapter 6. Conclusions 171
6.1 Research Results and Discussion 171
6.2 Contributions 181
6.3 Future Research 183
Bibliography 185
Appendices 191
Abstract (Korean) 203Docto
User Sensory Emotional Factor Support System and User Interface Guideline for Non-expert Designer
No full textAdministration of a Sol-Gel Formulation of Phenylephrine Using Low-Temperature Hollow Microneedle for Treatment of Intermittent Fecal Incontinence
No full textLaparoscopic distal pancreatectomy of a solid pseudopapillary tumour in a teenager after size reduction: A case report
No full textA solid pseudopapillary tumour (SPT) of the pancreas is very rare in children. Despite its malignancy potential, SPT is known to have a good prognosis. No prior report has described the use of neoadjuvant chemoradiation with retinoid therapy for size reduction of an SPT before surgical excision.
An 11-year-old girl had intermittent abdominal pain and a huge mass occupying the whole abdomen. The plain abdominal radiograph and abdominal computed tomography scan showed a 16-cm, huge pancreatic mass across the midline. An ultrasonography-guided biopsy was performed, and pathological examination confirmed an SPT of the pancreas. As the mass was adjacent to large vessels, chemoradiotherapy was preferred to surgical resection. After 1 year of chemoradiotherapy, the tumour decreased to 14 cm. Three years of immunotherapy with retinoic acid reduced the tumour to 10 cm. Laparoscopic distal pancreatectomy was performed after the size of the tumour had decreased. The tumour was successfully excised, and the pathological examination confirmed an SPT of the pancreas without malignant characteristics. The patient's postoperative clinical course was uneventful; there was no evidence of recurrence or metastasis during the 16-monthfollow-up.
Surgical resection is a reliable treatment for SPTs; however, a second option needs to be used when operation is inappropriate. In our case, subsequent laparoscopic distal pancreatectomy was successful
STAT3-mediated MLST8 gene expression regulates cap-dependent translation in cancer cells
No full textSignal transducer and activator of transcription 3 (STAT3) regulates cell growth, cell survival, angiogenesis, metastasis of cancer cells, and cancer immune evasion by regulating gene expression as a transcription factor. However, the effect of STAT3 on translation is almost unknown. We demonstrated that STAT3 acts as a trans-acting factor for MLST8 gene expression and the protein level of mLST8, a core component of mechanistic target of rapamycin complex 1 and 2 (mTORC1/2), positively regulates the mTORC1/2 downstream pathways. Suppression of STAT3 by siRNA attenuated 4E-BP1 phosphorylation, cap-dependent translation, and cell proliferation in a variety of cancer cells. In HCT116 cells, STAT3 knockdown-induced decreases in 4E-BP1 and AKT phosphorylation levels were further attenuated by MLST8 knockdown or recovered by mLST8 overexpression. STAT3 knockdown-induced G2/M phase arrest was partially restored by co-knockdown of 4EBP1, and the attenuation of cell proliferation was enhanced by the expression of an mTORC1-mediated phosphorylation-defective mutant of 4E-BP1. ChIP and promoter mapping using a luciferase reporter assay showed that the -951 to -894 bp of MLST8 promoter seems to include STAT3-binding site. Overall, these results suggest that STAT3-driven MLST8 gene expression regulates cap-dependent translation through 4E-BP1 phosphorylation in cancer cells.ope
- …
