• Journal of Radiation Protection and Research
• /
• v.44 no.1
• /
• pp.1-7
• /
• 2019

Background: To monitor proton beam in proton therapy, prompt gamma imaging systems are being developed by several research groups, and these systems are expected to improve the quality of the treatment and the patient safety. To apply the prompt gamma imaging systems into spot scanning proton therapy, the systems should be able to monitor the proton beam range of a spot with a small number of protons ( <$10^8$ protons), which is quite often not the case due to insufficient prompt gamma statistics. Materials and Methods: In the present study, we propose to improve prompt gamma statistics by merging the prompt gamma distributions of several individual spots into a new distribution. This proposal was tested by Geant4 Monte Carlo simulations for a multi-slit prompt gamma camera which has been developed to measure the proton beam range in the patient. Results and Discussion: The results show that the proposed method clearly enhance the statistical precision of beam range measurement. The accuracy of beam range verification is improved, within ~1.4 mm error, which is not achievable before applying the developed method. Conclusion: In this study, we tried to improve the statistics of the prompt gamma statistics by merging the prompt gamma distributions of multiple spots, and it was found that the merged distribution provided sufficient prompt gamma statistics and the proton beam range was determined accurately.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.77-81
• /
• 2022

최근 사전학습 언어모델에 내재된 지식을 최대한으로 활용하고자 태스크에 대한 설명을 입력으로 주는 manual prompt tuning 방법과 자연어 대신 학습가능한 파라미터로 태스크에 대한 이해를 돕는 soft prompt tuning 방법론이 자연어처리 분야에서 활발히 연구가 진행되고 있다. 이에 본 연구에서는 페르소나 대화 생성 태스크에서 encoder-decoder 구조 기반의 사전학습 언어모델 BART를 활용하여 manual prompt tuning 및 soft prompt tuning 방법을 고안하고, 파인튜닝과의 성능을 비교한다. 전체 학습 데이터에 대한 실험 뿐 아니라, few-shot 세팅에서의 성능을 확인한다.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.269-274
• /
• 2023

질의응답 (Question Answering)은 주어진 질문을 이해하여 그에 맞는 답변을 생성하는 자연어 처리 분야의 핵심적인 기계 독해 작업이다. 현재 대다수의 자연어 이해 작업은 사전학습 언어 모델에 미세 조정 (finetuning)하는 방식으로 학습되고, 질의응답 역시 이러한 방법으로 진행된다. 하지만 미세 조정을 통한 전이학습은 사전학습 모델의 크기가 커질수록 전이학습이 잘 이루어지지 않는다는 단점이 있다. 게다가 많은 양의 파라미터를 갱신한 후 새로운 가중치들을 저장하여야 한다는 용량의 부담이 존재한다. 본 연구는 최근 대두되는 deep prompt tuning 방법론을 한국어 추출형 질의응답에 적용하여, 미세 조정에 비해 학습시간을 단축시키고 적은 양의 파라미터를 활용하여 성능을 개선했다. 또한 한국어 추출형 질의응답에 최적의 prompt 길이를 최적화하였으며 오류 분석을 통한 정성적인 평가로 deep prompt tuning이 모델 예측에 미치는 영향을 조사하였다.

• Journal of the Korea Safety Management & Science
• /
• v.11 no.4
• /
• pp.77-83
• /
• 2009

The purpose of this study was to examine the relative effectiveness of immediate feedback and informational prompt on safe sitting behaviors that may cause VDT syndromes. Participants were three white color workers and an ABCB within-subject design was adopted. Safety Posture System was developed specifically for the present study. The system could detect participants' unsafe sitting postures using sensors and provide feedback and prompt on the computer monitors. The results indicated that both immediate feedback and informational prompt considerably increased safe sitting behaviors. More importantly, the immediate feedback was more effective than the informational prompt in increasing safe sitting behaviors.

• Advances in Energy Research
• /
• v.3 no.2
• /
• pp.125-131
• /
• 2015

Prompt neutron lifetime calculations have been performed for the NIRR-1 reactor HEU and LEU cores using the 1/v insertion and the Adjoint flux weighing methods. Results of calculations obtained for the HEU and LEU cores are respectively $57.3{\pm}0.8$ and $47.5{\pm}0.7$ for the 1/v insertion and $56.9{\pm}0.3$ and $46.3{\pm}0.5$ for the Adjoint flux. There is a good agreement seen between the two methods for both cores. The prompt neutron lifetime was observed to be shorter in the LEU than for the HEU as expected. However, the Adjoint flux weighing method seemed to be the easiest method in calculating the prompt neutron lifetime for NIRR-1.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.05a
• /
• pp.595-597
• /
• 2023

대규모 언어 모델에 기반한 AI 챗봇인 ChatGPT 의 사용과 출력 품질을 개선하는 데 있어 Prompt Engineering 의 중요성이 확대되고 있다. Prompt Engineering 은 AI 모델에서 보다 정확하고 관련성 높은 응답을 생성하기 위해 프롬프트의 요소를 선택하고 구성하는 작업을 포함한다. 본 논문에서는 ChatGPT 에서 정보나 답변을 효과적으로 추출하는 데 사용할 수 있는 다양한 Prompt Engineering 기법을 소개하고 이러한 기법이 실제 시나리오에 어떻게 적용될 수 있는지에 대한 예를 제공한다.

• Restorative Dentistry and Endodontics
• /
• v.29 no.5
• /
• pp.423-429
• /
• 2004

The purpose of this study was to evaluate the effect of multiple application of all-in-one dentin adhesive system on microtensile bond strength using confocal laser scanning microscope and microtensile bond strength test. Flat occlusal dentin surfaces were prepared using low-speed diamond saw. In group I, Scotchbond Multipurpose (SM) was applied by manufacturer's recommendation. In group II, after Adper Prompt L-Pop was applied for 15s and light cured for 10s. the second coat was re-applied and light-cured. In group III, after light-curing the second layer. the third coat was re-applied and light-cured. Specimens bonded with a resin-composite were sectioned into resin-dentin stick for measuring the adhesive layer thickness by confocal laser scanning microscope and evaluating micro-tensile bond strength. The adhesive layers of three-step dentin adhesive system. 3 coats of Adper Prompt L-Pop had significantly thicker than SM. 2 coats of Adper Prompt L-Pop (p < 0.05). However. there was no significant differences in bond strengths between SM and 3 coats of Adper Prompt L-Pop (p > 0.05). And SM. 3 coats of Adper Prompt L-Pop had significantly higher than 2 coats of Adper Prompt L-Pop in bond strengths (p < 0.05).

• Proceedings of the Korean Nuclear Society Conference
• /
• 1997.05a
• /
• pp.184-189
• /
• 1997

A prompt fission gas release model incorporating the resistance to gas flow in the gap was developed and the effects of gap resistance and failure location on prompt fission gas release from the cladding breach were assessed. The process of prompt fission gas release from the plenum and gap into the coolant was modeled in accordance with three major phenomena: (1) transient gas flow in the gap, (2) the growth of the fission gas bubble while it is still attached to the breach, and (3) the detachment of the fission gas bubble from the breach and mixing with the coolant. The cumulative mass release fraction by the present model was calculated for the case of Young-Gwang 3 & 4 nuclear fuel rod as a typical example. The results showed that the release behavior of prompt fission gas with time was different from the frictionless model which has frequently been used in a simplified approach, and that the location of cladding failure was another key factor for the prompt fission gas release process due to the resistance in the gap.

• Nuclear Engineering and Technology
• /
• v.54 no.9
• /
• pp.3422-3428
• /
• 2022

The prompt gamma imaging (PGI) technique is considered as one of the most promising approaches to estimate the range of proton beam in the patient and unlock the full potential of proton therapy. In the PGI technique, a dedicated algorithm is required to estimate the range of the proton beam from the prompt gamma (PG) distribution acquired by a PGI system. In the present study, a new range estimation algorithm was developed for a multi-slit prompt-gamma camera, one of PGI systems, to estimate the range of proton beam with high accuracy. The performance of the developed algorithm was evaluated by Monte Carlo simulations for various beam/phantom combinations. Our results generally show that the developed algorithm is very robust, showing very high accuracy and precision for all the cases considered in the present study. The range estimation accuracy of the developed algorithm was 0.5-1.7 mm, which is approximately 1% of beam range, for 1×10⁹ protons. Even for the typical number of protons for a spot (1×10⁸), the range estimation accuracy of the developed algorithm was 2.1-4.6 mm and smaller than the range uncertainties and typical safety margin, while that of the existing algorithm was 2.5-9.6 mm.

• Progress in Medical Physics
• /
• v.28 no.4
• /
• pp.207-217
• /
• 2017

For effective patient treatment in proton therapy, it is therefore important to accurately measure the beam range. For measuring beam range, various researchers determine the beam range by measuring the prompt gammas generated during nuclear reactions of protons with materials. However, the accuracy of the beam range determination can be lowered in heterogeneous phantoms, because of the differences with respect to the prompt gamma production depending on the properties of the material. In this research, to improve the beam range determination in a heterogeneous phantom, we derived a formula to correct the prompt-gamma distribution using the ratio of the prompt gamma production, stopping power, and density obtained for each material. Then, the prompt-gamma distributions were acquired by a multi-slit prompt-gamma camera on various kinds of heterogeneous phantoms using a Geant4 Monte Carlo simulation, and the deduced formula was applied to the prompt-gamma distributions. For the case involving the phantom having bone-equivalent material in the soft tissue-equivalent material, it was confirmed that compared to the actual range, the determined ranges were relatively accurate both before and after correction. In the case of a phantom having the lung-equivalent material in the soft tissue-equivalent material, although the maximum error before correction was 18.7 mm, the difference was very large. However, when the correction method was applied, the accuracy was significantly improved by a maximum error of 4.1 mm. Moreover, for a phantom that was constructed based on CT data, after applying the calibration method, the beam range could be generally determined within an error of 2.5 mm. Simulation results confirmed the potential to determine the beam range with high accuracy in heterogeneous phantoms by applying the proposed correction method. In future, these methods will be verified by performing experiments using a therapeutic proton beam.