• 제목/요약/키워드: Pre-evaluation for prediction

검색결과 63건 처리시간 0.034초

다중목표 대화형 추천시스템을 위한 사전 학습된 언어모델들에 대한 성능 평가 (Performance Evaluation of Pre-trained Language Models in Multi-Goal Conversational Recommender Systems)

  • 김태호;장형준;김상욱
    • 스마트미디어저널
    • /
    • 제12권6호
    • /
    • pp.35-40
    • /
    • 2023
  • 본 연구는 대화형 추천 시스템인 다중 목표 대화형 추천 시스템(MG-CRS)에서 사용되는 다양한 사전 학습된 언어 모델들을 고찰하고, 각 언어모델의 성능을 비교하고 분석한다. 특히, 언어 모델의 크기가 다중 목표 대화형 추천 시스템의 성능에 어떤 영향을 미치는지에 대해 살펴본다. BERT, GPT2, 그리고 BART의 세 종류의 언어모델을 대상으로 하여, 대표적인 다중 목표 대화형 추천 시스템 데이터셋인 DuRecDial 2.0에서 '타입 예측'과 '토픽 예측'의 정확도를 측정하고 비교한다. 실험 결과, 타입 예측에서는 모든 모델이 뛰어난 성능을 보였지만, 토픽예측에서는 모델 간에 혹은 사이즈에 따라 성능 차이가 관찰되었다. 이러한 결과를 바탕으로 다중 목표 대화형 추천 시스템의 성능 향상을 위한 방향을 제시한다.

공조시스템 유기 격실 소음 예측 프로그램 개발 (Development of Cabin Noise Prediction Program Induced by HVAC System)

  • 김병희;권종현;조대승
    • 한국소음진동공학회:학술대회논문집
    • /
    • 한국소음진동공학회 2004년도 추계학술대회논문집
    • /
    • pp.554-558
    • /
    • 2004
  • In this paper, we introduce noise prediction program of HVAC system to assist low-noisy design of ship's cabin. The developed program calculates sound power levels at HVAC components considering primary and secondary noise generated by fan and duct element, duct element noise attenuation, and duct break-in noise based on the authentic empirical method suggested by NEBB and acoustic power balancing method. Sound pressure level at cabin with or without ceiling system is evaluated by the diffuse-field theory considering diffuser and duct break-out sound powers. Moreover, the program provides intuitive pre- and post-processors using modem GUI functions to help efficient modeling and evaluation of cabin and HVAC component noise. To validate the accuracy and convenience of the program, noise prediction for a HVAC system is demonstrated.

  • PDF

Effects of Oxidation and Hot Corrosion on the Erosion of Silicon Nitride

  • Kim, Jong Jip
    • Corrosion Science and Technology
    • /
    • 제4권4호
    • /
    • pp.136-139
    • /
    • 2005
  • The effect of oxidation and hot corrosion on the solid particle erosion was investigated for hot-pressed silicon nitride using as-polished, pre-oxidized and pre-corroded specimens by molten sodium sulfates. Erosion tests were performed at 22, 500 and $900^{\circ}C$ using angular silicon carbide particles of mean diameter $100{\mu}m$. Experimental results show that solid particle erosion rate of silicon nitride increases with increasing temperature for as-polished or pre-oxidized specimens in consistent with the prediction of a theoretical model. Erosion rate of pre-oxidized specimens is lower than that of as-polished specimens at $22^{\circ}C$, but it is higher at $900^{\circ}C$. Lower erosion rate at $22^{\circ}C$ in the pre-oxidized specimens is attributed due to the blunting of surface flaws, and the higher erosion rate at $900^{\circ}C$ is due to brittle lateral cracking. Erosion rate of pre-corroded specimens decreases with increasing temperature. Less erosion at $900^{\circ}C$ than at $22^{\circ}C$ is associated with the liquid corrosion products sealing off pores at $900^{\circ}C$ and the absence of inter-granular crack propagation observed at $22^{\circ}C$.

Fatigue Indicator Sensor의 형상에 따른 균열진전 특성의 비교 연구 (A Comparative Study on the Crack Propagation Characteristics According to the Pre-Notch Shapes of Fatigue Indicator Sensor)

  • 김재현;김슬기;조영근;여승훈;김경수;김성찬;이장현
    • 대한조선학회논문집
    • /
    • 제47권4호
    • /
    • pp.565-572
    • /
    • 2010
  • It is difficult to predict the accurate fatigue life of the ship structure because of load uncertainty and load redistribution at the ship structure members. As one of studies for accurate evaluation and prediction of fatigue life, it is a promising way to detect the crack previously by attaching the Fatigue Indicator Sensor (FIS) at the crack prediction region. In order to predict the fatigue life of the ship structure by using FIS, it is required to know previously the crack propagation characteristics according to pre-notch shapes. In this study, we obtained the stress distribution phase, stress concentration factors and stress intensity factor of various pre-notch shapes through FEA. Additionally, we conducted the fatigue test and obtained the characteristics of crack propagation according to the pre-notch shapes through comparison between the fatigue test and the FEA. Consequently, we classified the pre-notch shape into 3 categories: Long, Medium, and Short life type. On the basis of the numerical and experimental results, the FIS can be developed.

스틸코드 롤러교정공정의 압하량에 따른 교정도 평가 (Evaluation of the Product Quality According to Intermesh of the Roller Straightening Process of Steel Cord)

  • 배기현;이종섭;허훈;이준우;이병호
    • 한국소성가공학회:학술대회논문집
    • /
    • 한국소성가공학회 2009년도 추계학술대회 논문집
    • /
    • pp.271-274
    • /
    • 2009
  • This paper deals with an evaluation of the product quality according to intermesh of the roller straightening process of a steel cord. To perform the experiments, a single-layered steel cord with three wires is selected as a target. Intermeshes at inlet and outlet of the roller straightening device are selected as a respective design parameter. According to two intermeshes of the roller straightening device, a design table is generated and experiments were performed. Three assessment items of the product quality, such as the residual torsion, the arc-height and the pre-forming ratio, are measured in each experimental case for the quantitative evaluation of a steel cord. Based on the measured data, the sensitivity of two intermeshes was analyzed and the prediction equation for the product quality of a steel cord was also constructed from the regression analysis.

  • PDF

Aerodynamic Shape Optimization using Discrete Adjoint Formulation based on Overset Mesh System

  • Lee, Byung-Joon;Yim, Jin-Woo;Yi, Jun-Sok;Kim, Chong-Am
    • International Journal of Aeronautical and Space Sciences
    • /
    • 제8권1호
    • /
    • pp.95-104
    • /
    • 2007
  • A new design approach of complex geometries such as wing/body configuration is arranged by using overset mesh techniques under large scale computing environment. For an in-depth study of the flow physics and highly accurate design, several special overlapped structured blocks such as collar grid, tip-cap grid, and etc. which are commonly used in refined drag prediction are adopted to consider the applicability of the present design tools to practical problems. Various pre- and post-processing techniques for overset flow analysis and sensitivity analysis are devised or implemented to resolve overset mesh techniques into the design optimization problem based on Gradient Based Optimization Method (GBOM). In the pre-processing, the convergence characteristics of the flow solver and sensitivity analysis are improved by overlap optimization method. Moreover, a new post-processing method, Spline-Boundary Intersecting Grid (S-BIG) scheme, is proposed by considering the ratio of cell area for more refined prediction of aerodynamic coefficients and efficient evaluation of their sensitivities under parallel computing environment. With respect to the sensitivity analysis, discrete adjoint formulations for overset boundary conditions are derived by a full hand-differentiation. A smooth geometric modification on the overlapped surface boundaries and evaluation of grid sensitivities can be performed by mapping from planform coordinate to the surface meshes with Hicks-Henne function. Careful design works for the drag minimization problems of a transonic wing and a wing/body configuration are performed by using the newly-developed and -applied overset mesh techniques. The results from design applications demonstrate the capability of the present design approach successfully.

Characterising Forages for Ruminant Feeding

  • Dynes, R.A.;Henry, D.A.;Masters, D.G.
    • Asian-Australasian Journal of Animal Sciences
    • /
    • 제16권1호
    • /
    • pp.116-123
    • /
    • 2003
  • Forages are the most important feed resource for ruminants worldwide, whether fed as pastures, forage crops or conserved hay, silage or haylage. There is large variability in the quality of forages so measurement and prediction of feeding value and nutritive value are essential for high levels of production. Within a commercial animal production system, methods of prediction must be inexpensive and rapid. At least 50% of the variation in feeding value of forages is due to variation in voluntary feed intake. Identification of the factors that constrain voluntary feed intake allows these differences to be managed and exploited in forage selection. Constraints to intake have been predicted using combinations of metabolic and physical factors within the animal while simple measurements such as the energy required to shear the plant material are related to constraints to intake with some plant material. Animals respond to both pre- and post-ingestive feedback signals from forages. Pre-ingestive signals may play a role in intake with signals including taste, odour and texture together with learned aversions to nutrients or toxins (post-ingestive feedback signals). The challenge to forage evaluation is identification of the factors which are most important contributors to these feedback signals. Empirical models incorporating chemical composition are also widely used. The models tend to be useful within the ranges of the datasets used in their development but none can claim to have universal application. Mechanistic models are becoming increasingly complex and sophisticated and incorporate both feed characteristics and use of biochemical pathways within the animal. Improvement in utilisation through the deliberate selection of pasture plants for high feeding value appears to have potential and has been poorly exploited. Use of Near Infrared Reflectance Spectroscopy is a simple method that offers significant potential for the preliminary screening of plants with genetic differences in feeding value. Near Infrared Reflectance Spectroscopy will only be as reliable as the calibration sets from which the equations are generated.

발전소 온배수에 의한 해양물리학적 평가기법 개선방안 연구 (Improvement Plan of Ocean Physics Assessment Technique for Power Plant Thermal Effluent)

  • 김명원;조광우;맹준호;강태순;김종규
    • 한국해양공학회지
    • /
    • 제28권3호
    • /
    • pp.245-253
    • /
    • 2014
  • This research analyzed the current situation and problems with an environmental impact assessment to provide a rational ocean physics assessment technique for power plant thermal effluent. This research also tried to create an improvement plan for heated effluent diffusion impact assessment by examining the reporting regulations for environmental impact assessment, national and international evaluation guidelines, etc. In the case of evaluating the oceanographic impact of heated effluent discharged from power plants, a pre-investigation is necessary before a full-scale presentence investigation, to accurately predict and minimize power plant construction effects on the surrounding environments. Before this presentence investigation, moreover, an integrated presentence plan, which agrees with the business plan, effect prediction, and post-investigation, needs to be established. A sufficient summit investigation must be made, which considers climate changes, and new and additional power plant construction. For accurate long-term oceanic environmental change prediction, the credibility of effect prediction must be elevated by presenting an evaluation method that is categorized by numerical organization models, verification methods, result presentation, and other things. Furthermore, unproductive conflicts between the people involved in heated effluent evaluation should be reduced by these improvement plans.

대형 컨테이너운반선의 공조 소음 해석사례 (Noise Analysis of Large Container Carrier Vessel on HVAC Noise)

  • 김문수;조대승;김병희;권종현
    • 대한조선학회 특별논문집
    • /
    • 대한조선학회 2006년도 특별논문집
    • /
    • pp.65-70
    • /
    • 2006
  • In this paper, we introduce prediction program of HVAC system, HJNOVAC Version 2.0. The developed program adopts both the authentic empirical method suggested by NEBB and acoustic power balancing method. The program provides intuitive pre- and post- processor using modern GUI function to help efficient modeling and evaluation of cabin and HVAC component noises. To verify the accuracy and convenience of the program, we carry out noise prediction of HVAC system for 8,100 TEU Container Carrier and measure the noise levels of cabins during sea trial.

  • PDF

Physical Model Investigation of a Compact Waste Water Pumping Station

  • Kirst, Kilian;Hellmann, D.H.;Kothe, Bernd;Springer, Peer
    • International Journal of Fluid Machinery and Systems
    • /
    • 제3권4호
    • /
    • pp.285-291
    • /
    • 2010
  • To provide required flow rates of cooling or circulating water properly, approach flow conditions of vertical pump systems should be in compliance with state of the art acceptance criteria. The direct inflow should be vortex free, with low pre-rotation and symmetric velocity distribution. Physical model investigations are common practice and the best tool of prediction to evaluate, to optimize and to document flow conditions inside intake structures for vertical pumping systems. Optimization steps should be accomplished with respect to installation costs and complexity on site. The report shows evaluation of various approach flow conditions inside a compact waste water pumping station. The focus is on the occurrence of free surface vortices and the evaluation of air entrainment for various water level and flow rates. The presentation of the results includes the description of the investigated intake structure, occurring flow problems and final recommendations.