• Title/Summary/Keyword: Zero-shot

Search Result 41, Processing Time 0.035 seconds

A Novel Control Method of Resistance Spot Welding Inverter using Dynamic Resistance Characteristics for Weld Quality Improvement (용접품질 향상을 위한 저항 스폿 용접용 인버터의 동저항 특성을 이용한 새로운 제어기법)

  • Kang, Sung-Kwan;Jung, Jae-Hun;Nho, Eui-Cheol
    • The Transactions of the Korean Institute of Power Electronics
    • /
    • v.20 no.6
    • /
    • pp.491-497
    • /
    • 2015
  • This study proposes a new control method for a resistance spot welding inverter to improve weld quality. The proposed method is based on the dynamic resistance characteristics of steel sheets to be welded. A point in the second peak value of the dynamic resistance occurs during one shot of the welding current flow. A constant voltage control is applied from zero to the peak point, and a constant current control is adopted from the peak point to the end of the shot. The mixed mode control of the constant voltage and current guarantees high weld quality. Experiments are conducted with a 5 kA power supply and 0.5 mm steel sheets to compare quality. Experimental results show that weld quality is improved more than 10 times that of the conventional control method.

Impact of FWM on manchester coded DPSK WDM communication systems (Manchester coded DPSK WIDM 통신 시스템에서 FWM의 영향)

  • 이호준
    • Korean Journal of Optics and Photonics
    • /
    • v.5 no.2
    • /
    • pp.319-325
    • /
    • 1994
  • The performance of Manchester-coded DPSK optical wavelength division multiplexing (WDM) systems using a stochastic approach is evaluated taking into account the shot noise and the four-wave mixing (FWM) caused by fiber nonlinearities. The result of Manchester-coded system is compared to conventional non-return-to-zero (NRZ) systems for DPSK modulation formats. Further, the dynamic range, defined as the ratio of the maximum input power (limited by the FWM), to the minimum input power (limited by receiver sensitivity), is evaluated. For $1.55.{\mu}m$16 channel WDM systems, the dynamic range of DPSK Manchester coded systems shows a 2.1 dB improvement with respect to the NRZ. This result holds true for both dispersion-shifted fiber and conventional fiber; it has been obtained for 10 GHz channel spacing, 1 Gbps/channel bit rate.t rate.

  • PDF

Impact of Four Wave Mixing on Manchester Coded ASK Multichannel Optical Communication System (Manchester Coded ASK 다중채널 광통신 시스템의 Four Wave Mixing 에 대한 영향)

  • Lee, Ho-Joon;Leonid G. Kazovsky
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.18 no.11
    • /
    • pp.1607-1617
    • /
    • 1993
  • The performance of Manchester-coded ASK optical wavelength division multiplexing(WDM) systems is evaluated laking into account the shot noise and the four wave mixing(FWM) caused by fiber nonlinearities. The result is compared to conventional non-return-to-zero(NRZ) systems for ASK modulation formats. Further, the dynamic range, defined as the ratio of the maximum input power(limited by the FWM), to the minimum input power(limited by receiver sensitivity), is evaluated. For 1.55 rm 16 channel WDM systems, the dynamic range of ASK Manchester coded systems shows a 2.0 dB improvement with respect to the NRZ. This result holds true for both dispersion-shifted fiber and conventional fiber it has been obtained for 10 GHz channel spacing, 1 Gbps/channel bit rate.

  • PDF

Improving visual relationship detection using linguistic and spatial cues

  • Jung, Jaewon;Park, Jongyoul
    • ETRI Journal
    • /
    • v.42 no.3
    • /
    • pp.399-410
    • /
    • 2020
  • Detecting visual relationships in an image is important in an image understanding task. It enables higher image understanding tasks, that is, predicting the next scene and understanding what occurs in an image. A visual relationship comprises of a subject, a predicate, and an object, and is related to visual, language, and spatial cues. The predicate explains the relationship between the subject and object and can be categorized into different categories such as prepositions and verbs. A large visual gap exists although the visual relationship is included in the same predicate. This study improves upon a previous study (that uses language cues using two losses) and a spatial cue (that only includes individual information) by adding relative information on the subject and object of the extant study. The architectural limitation is demonstrated and is overcome to detect all zero-shot visual relationships. A new problem is discovered, and an explanation of how it decreases performance is provided. The experiment is conducted on the VRD and VG datasets and a significant improvement over previous results is obtained.

The Vertical Deviation of the Impact Point from and Aiming Point at an Inclined shooting Ranges (소총의 상하향 사격시 탄도학상의 수직편차)

  • ;;Chang, Won Hong
    • Transactions of the Korean Society of Mechanical Engineers
    • /
    • v.2 no.2
    • /
    • pp.47-51
    • /
    • 1978
  • In order to improve the hitting rate in the shot of rifles, it is required that the analysis of exterior ballistics and the line of sight. One of the important factors influenced a marksman using a rifle obtained the zero-setting of a rifle at the horizontal range, is the deviation of the impact point from the aiming point when the shooting is performed in an inclined ranges. The deviation usually cccurs from the reaction force along the bore line, the characteristics of exterior ballistics, and the error of a shooting range judgement by the inclined range. This study is concerned with the problem of the vertical difference between the impact and aiming point in the inclined shooting ranges. The computing method to find the vertical difference is represented. This method is applied for and experimental rifle in three cases, (1) hofizontal shooting ranges, (2) upper inclined shooting ranges, and (3) lower inclined shooting ranges.

Diffusion Tensor MRI and Fiber Tractography: Evaluation of Developmental CNS Anomaly: Preliminary Results

  • Lee, Seung-Koo;Kim, Dong-Ik
    • Proceedings of the KSMRM Conference
    • /
    • 2002.11a
    • /
    • pp.86-86
    • /
    • 2002
  • Purpose: To evaluate the white matter fiber configuration in various developmental CNS anomaly 대상 및 방법: Jubert Syndrome, congenital schizencephaly, callosal agenesis and hemiplegic cerebral palsy patients were evaluated by diffusion tensor MRI. All studies were performed using a 1.5T Philips Gyroscan Intern system. Diffusion weighted imaging was performed using single-shot echo planar imaging, with navigator echo phase correction and SENSE. Diffusion weighting was performed along six independent axes, using diffusion weighting of b=600s/$\textrm{mm}^2$. 128 matrix/zero filled to 256, 23cm FOV, 3mm slice thickness were used for imaging parameters. Data were processed on a Window-2000 PC equipped with IDL and PRIDE (Philips Medical System).

  • PDF

Zero-Shot Readability Assessment of Korean ESG Reports using BERT (BERT를 활용한 한국어 지속가능경영 보고서의 제로샷 가독성 평가)

  • Son, Guijin;Yoon, Naeun;Lee, Kaeun
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2022.05a
    • /
    • pp.456-459
    • /
    • 2022
  • 본 연구는 최근 자연어 인공지능 연구 동향에 발맞추어 사전 학습된 언어 인공지능을 활용한 의미론적 분석을 통해 국문 보고서의 가독성을 평가하는 방법론 두 가지를 제안한다. 연구진은 연구 과정에서 사전 학습된 언어 인공지능을 활용해 추가 학습 없이 문장을 임의의 벡터값으로 임베딩하고 이를 통해 1. 의미론적 복잡도 와 2. 내재적 감정 변동성 두 가지 지표를 추출한다. 나아가, 앞서 발견한 두 지표가 국문 보고서의 가독성과 정(+)의 상관관계에 있음을 확인하였다. 본 연구는 통사론적 분석과 레이블링 된 데이터에 크게 의존하던 기존의 가독성 평가 방법론으로 부터 탈피해, 별도의 학습 없이 기존 가독성 지표에 근사한다는 점에서 의미가 있다.

HeavyRoBERTa: Pretrained Language Model for Heavy Industry (HeavyRoBERTa: 중공업 특화 사전 학습 언어 모델)

  • Lee, Jeong-Doo;Na, Seung-Hoon
    • Annual Conference on Human and Language Technology
    • /
    • 2021.10a
    • /
    • pp.602-604
    • /
    • 2021
  • 최근 자연어 처리 분야에서 사전 학습된 언어 모델은 다양한 응용 태스크에 적용되어 성능을 향상시켰다. 하지만 일반적인 말뭉치로 사전 학습된 언어 모델의 경우 중공업 분야처럼 전문적인 분야의 응용 태스크에서 좋은 성능을 나타내지 못한다. 때문에 본 논문에서는 이러한 문제점을 해결하기 위해 중공업 말뭉치를 이용한 RoBERTa 기반의 중공업 분야에 특화된 언어 모델 HeavyRoBERTa를 제안하고 이를 통해 중공업 말뭉치 상에서 Perplexity와 zero-shot 유의어 추출 태스크에서 성능을 개선시켰다.

  • PDF

A Study on Instruction Tuning for Large-scale Language Models (명령어 튜닝이 대규모 언어 모델의 문장 생성에미치는 영향력 분석)

  • Yohan Na;Dong-Kyu Chae
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2023.05a
    • /
    • pp.684-686
    • /
    • 2023
  • 최근 대규모 언어모델 (large language models) 을 활용하여 다양한 자연어처리 문제를 추가학습 없이 풀어내기 위한 zero-shot 학습에 대한 연구가 활발히 수행되고 있다. 특히 프롬프트 튜닝(prompt tuning)을 활용하여 적은 학습만으로도 효과적으로 다양한 태스크에 적응하도록 돕는 방법이 최근 대규모 언어모델의 성능을 향상시키고 있다. 본 논문은 명령어 튜닝 (instruction tuning) 이 언어모델에 끼치는 영향을 분석하였다. 명령어 튜닝된 모델이 기존 언어모델과 비교하여 변화된 문장 생성 특징, 생성된 문장의 품질 등에 대한 분석을 수행하고 결과를 제시한다.

Wanda Pruning for Lightweighting Korean Language Model (Wanda Pruning에 기반한 한국어 언어 모델 경량화)

  • Jun-Ho Yoon;Daeryong Seo;Donghyeon Jeon;Inho Kang;Seung-Hoon Na
    • Annual Conference on Human and Language Technology
    • /
    • 2023.10a
    • /
    • pp.437-442
    • /
    • 2023
  • 최근에 등장한 대규모 언어 모델은 다양한 언어 처리 작업에서 놀라운 성능을 발휘하고 있다. 그러나 이러한 모델의 크기와 복잡성 때문에 모델 경량화의 필요성이 대두되고 있다. Pruning은 이러한 경량화 전략 중 하나로, 모델의 가중치나 연결의 일부를 제거하여 크기를 줄이면서도 동시에 성능을 최적화하는 방법을 제시한다. 본 논문에서는 한국어 언어 모델인 Polyglot-Ko에 Wanda[1] 기법을 적용하여 Pruning 작업을 수행하였다. 그리고 이를 통해 가중치가 제거된 모델의 Perplexity, Zero-shot 성능, 그리고 Fine-tuning 후의 성능을 분석하였다. 실험 결과, Wanda-50%, 4:8 Sparsity 패턴, 2:4 Sparsity 패턴의 순서로 높은 성능을 나타냈으며, 특히 일부 조건에서는 기존의 Dense 모델보다 더 뛰어난 성능을 보였다. 이러한 결과는 오늘날 대규모 언어 모델 중심의 연구에서 Pruning 기법의 효과와 그 중요성을 재확인하는 계기가 되었다.

  • PDF