http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
박형준,홍충선,박성배,송현제 한국정보과학회 2022 정보과학회 컴퓨팅의 실제 논문지 Vol.28 No.3
Multi-domain response generation is the task of generating responses to utterances that cover more than one domain. As the number of domains increases, multi-domain response generation must be able to respond appropriately to newly added domains as well as previously learned domains. However, previous studies based on fine-tuning with a new domain dataset have a problem that involves an inadequate response generation to pre-learned domains or a generation of domainindependent general responses. To solve this problem, the proposed model adopts a memory-replay continual learning to evolve a single-domain response generator to a multi-domain response generator. The model first learns a specific domain and retains some training instances in this domain for the next step. Then, the model is trained again with training instances for a new domain as well as the remaining instances. Since the remaining instances should represent the previous domains, the proposed method selects representative instances using clustering specialized to response generation. Through intensive experiments, the proposed method outperforms the baselines. 다중 도메인 응답 생성은 하나 이상의 도메인에 대해 입력되는 발화에 응답을 생성하는 문제이다. 여러 도메인을 대응해야 하는 환경에서 다중 도메인 응답 생성은 기존에 학습된 도메인뿐만 아니라 새로이 추가되는 도메인에 대해서도 적절히 응답할 수 있어야 한다. 기존 연구들은 기학습된 응답 생성 모델에 새로이 추가되는 도메인의 데이터로 미세 조정하였다. 하지만, 위 방법들은 기학습된 도메인에 대해 적절히 응답을 생성하지 못하거나 도메인과 무관한 일반적인 대답을 생성하는 문제가 있다. 본 논문에서는 기억재현 지속학습 기반 다중 도메인 응답 생성 모델을 구축하는 방법을 제안한다. 기억재현 지속학습 기반 응답 생성 모델은 이전 도메인 발화-응답 데이터 중 대표 데이터를 메모리에 저장해두고, 새로운 도메인의 데이터와 메모리에 저장된 데이터를 함께 활용하여 학습한다. 기학습된 도메인 중 일부분을 메모리에 저장 및 학습에 활용하여 새로운 도메인을 위해 모델을 갱신할지라도 도메인에 특화된 대답을 생성할 수 있다. 도메인에 특화된 대표 데이터를 생성하기 위해 본 논문에서는 k-평균 군집화를 활용한다. 실험에서 총 5개 도메인에 대해 순차적으로 학습하고 모든 도메인에 대해 응답을 제대로 생성하는지를 평가하였다. 실험 결과 제안한 응답 생성 모델이 모든 도메인에서 기존 비교 모델보다 우수한 성능을 보였다. 제안한 방법은 각 도메인별 성능 비교에서 비교 모델 대비 최대 4.45 BLEU 성능 향상을 얻었다. 뿐만 아니라 정성적 평가를 통해 제안한 방법이 비교 모델에 비해 도메인에 특화된 응답을 생성하는 것을 확인할 수 있었다.
박형준,최영철 대한신경과학회 2021 대한신경과학회지 Vol.39 No.-
With the rapid increase in the number of clinical trials in myopathy over the past decade, there is an increasing need for clinical scales to reflect patient’s clinical status. This article outlines the process of identifying possible measures. Detailed consideration has been given to key measures of muscle strength, function, and disability. As well as the usual assessment of the validity and reliability of the measures, three key characteristics were identified as necessary to the assessment of clinical scales used in health care: 1) the type of scale; 2) the clinical significance of the property being measured; and 3) the mathematical properties of the data. Consideration of such aspects facilitates the choice of clinical scales and the interpretation of data
박형준,윤재현 대한전기학회 2006 전기학회논문지 D Vol.55 No.2(D)
- The electroencephalogram generated by light stimulation in human body of dark adaption state was measured and analyzed in the cases that electromagnetic wave was put in and not put in, respectively. Shieldroom being able to interrupt the light and the electromagnetic wave of outer was constructed, and the experimental system being able to apply any light and any electromagnetic wave was designed. When the electromagnetic wave was applied to body or not, the variation characteristics of each component in the electroencephalogram were as follows. The δ wave was increased and the α wave and the β wave were decreased in the case that the electromagnetic wave was applied, and the variation range of the θ wave was small. And the influence of electromagnetic wave on human body was that the appearance time of the β wave was late, and it meaned that the time of visual recognition was delayed.
박형준 대한근전도전기진단의학회 2015 대한근전도 전기진단의학회지 Vol.17 No.2
The major nerves of the lower extremities are sciatic, femoral, obturator, lateral femoral cutaneous, superior gluteal and inferior gluteal nerves. These nerves are often entrapped at areas of anatomic restriction by various causes including trauma, compression and surgical/medical procedures. A well-designed electrodiagnostic study can help us localize a nerve lesion, assess severity and evaluate for the differential diagnosis of other peripheral nerve lesions including plexopathy or radiculopathy.