학습률(Step-Size)변화에 따른 디지털 신호의 기계학습 방법 개선

지상민,박지은 한국디지털정책학회 2020 디지털융복합연구 Vol.18 No.2

Machine learning is achieved by making a cost function from a given digital signal data and optimizing the cost function. The cost function here has local minimums in the cost function depending on the amount of digital signal data and the structure of the neural network. These local minimums make a problem that prevents learning. Among the many ways of solving these methods, our proposed method is to change the learning step-size. Unlike existed methods using the learning rate (step-size) as a fixed constant, the use of multivariate function as the cost function prevent unnecessary machine learning and find the best way to the minimum value. Numerical experiments show that the results of the proposed method improve about 3%(88.8%→91.5%) performance using the proposed method rather than the existed methods. 기계학습은 주어진 디지털 신호 Data로부터 비용함수를 만들고, 그 비용함수를 최소화함으로 학습이 이루어진다. 비용함수는 디지털 신호 Data의 양과 인공신경망의 구조에 따라 비용함수에 부분 최솟값(local minimum)들이 생기게 된다. 비용함수의 부분 최솟값들은 학습을 방해하는 요소가 된다. 이러한 방법을 해결하는 여러 방법 중 우리의 제안 방법은 학습률(Step-size)을 변화시키는 방법이다. 학습률을 고정된 상수로 이용하는 기존의 방법과는 다르게 비용함수를 이용한 다변수함수를 이용함으로써 불필요한 기계학습이 이루어지는 것을 방지할 수 있으며, 최솟값으로 가는 최적의 길을 찾을 수 있다. 수치적 실험을 통하여 기존의 방법보다 우리가 제안하는 방법을 이용하여 약 3%(88.8%→91.5%)의 성능이 향상하는 결과를 얻었다.

비용함수와 파라미터를 이용한 효과적인 디지털 데이터 기계학습 방법론

지상민,박지은,Ji, Sangmin,Park, Jieun 한국디지털정책학회 2021 디지털융복합연구 Vol.19 No.10

Machine learning is the process of constructing a cost function using learning data used for learning and an artificial neural network to predict the data, and finding parameters that minimize the cost function. Parameters are changed by using the gradient-based method of the cost function. The more complex the digital signal and the more complex the problem to be learned, the more complex and deeper the structure of the artificial neural network. Such a complex and deep neural network structure can cause over-fitting problems. In order to avoid over-fitting, a weight decay regularization method of parameters is used. We additionally use the value of the cost function in this method. In this way, the accuracy of machine learning is improved, and the superiority is confirmed through numerical experiments. These results derive accurate values for a wide range of artificial intelligence data through machine learning. 기계학습은 학습에 이용되는 학습 데이터와 데이터를 예측할 인공신경망을 이용하여 비용함수를 만들고, 비용함수를 최소화시키는 파라미터들을 찾는 과정이다. 파라미터들은 비용함수의 그래디언트 기반 방법들을 이용하여 변화하게 된다. 디지털 신호가 복잡할수록, 학습하고자 하는 문제가 복잡할수록, 인공신경망의 구조는 더욱 복잡해지고 깊어진다. 복잡하고, 깊어지는 인공신경망 구조는 과적합(Over-fitting) 문제를 발생시킨다. 과적합 문제를 해결하기 위하여 파라미터의 가중치 감소 정규화 방법이 사용되고 있다. 우리는 이러한 방법에서 추가로 비용함수의 값을 이용한다. 이러한 방법으로 기계학습의 정확도가 향상되는 결과를 얻었으며 이는 수치 실험을 통하여 우수성이 확인된다. 이러한 결과는 기계학습을 통한 인공지능의 폭넓은 데이터에 대한 정확한 값을 도출한다.

P-1004 : Novel Photocatalysts for Oxidation and Reduction of Water under Visible Light: Substitution Effect of Lead in Perovskite-Related Layered Oxides

지상민,김현규,장점석,배상원,손효창,이재성 한국화학공학회 2007 화학공학의이론과응용 Vol.10 No.2

The photocatalytic water splitting using sunlight is one of the most promising methods to store solar energy as chemical energy. In search of new highly efficient photocatalysts for water splitting under visible light irradiation, we studied the lead substitution effect in perovskite related layered niobates and titanates in a systematic manner. Our strategy was to shift the photocatalytic activity from UV into the visible light range, by high or complete substitution of cations in UV-active perovskite material with lead ion. We investigated layered perovskite phases of the Aurivillius (Bi2O2)2+(An-1BnO3n+1)2-, Dion-Jacobson M[An- 1BnO3n+1] and Ruddlesden-Popper (An+1BnO3n+1) types. We found in all the three structure type examples, that a non-visible light absorbing, non-photocatalytically active lead-free phase can be converted by lead substitution, into a visible light absorbing, visible light photocatalytically active phase. In this communication we report on the importance of lead substituion in obtaining visibly active layered photocatalysts.

흰점박이꽃무지(딱정벌레목 : 풍뎅이과) 3령 유충의 사육 밀도 조건에 따른 발육특성 연구

지상민,김선영,김소윤,송정훈 경상국립대학교 농업생명과학연구원 2023 농업생명과학연구 Vol.57 No.2

Rearing density experiments were performed in order to investigate the optimal rearing density of the third instar larvae of Protaetiabrevitarsis seulensis (Kolbe) with a long developmental period. For this study, 20, 30, and 40 third-instar larvae were reared in thesame size rearing boxes (3L), respectively, and developmental periods (from third instar to pupation, pupal duration, from third instarto adult), developmental characteristics and gender differences of adults (weight, body length, and width) were compared among densitytreatments. As a result, the developmental period (from third-instar to puaption, from third-instar to adult) of 30 larvae/box is shorterthan the other treatments (P < 0.05). The developmental characteristics of adults also showed statistically significant differences amongtreatment and 20 of them had the most weight, body length and width. The difference in developmental characteristics between malesand females was significant mainly in the 20-larvae density. The developmental characteristics of adults according to density were alsocompared by sexes, and although females were not affected by density, males were significantly affected by density in all characteristics,especially under 20-larvae density. In conclusion, third instar larvae are suitable to raise 30 larvae in 3-L boxes, which can significantlyshorten the developmental period from third instar to adult and reduce feed cost. 흰점박이꽃무지[Protaetia brevitarsis seulensis (Kolbe)]의 적정 사육 밀도를 구명하기 위해 상대적으로 발육기간이 긴 3령 유충을 대상으로밀도 실험을 수행하였다. 3령 유충기부터 번데기까지의 발육기간과 3령 유충부터 성충까지의 발육기간은 모두 30마리 밀도에서 가장 짧았다. 하지만, 번데기에서 성충까지 걸리는 시간은 유의미한 차이가 없었다. 성충의 외형적 특성도 밀도 간 통계적으로 유의미한 차이가 나타났는데,그중 20마리 밀도 조건에서 사육된 성충은 무게와 체장, 체폭이 가장 우량하였다. 암수의 외형적 특성 차이는 주로 20마리 조건에서 유의미하게나타났다. 암컷은 밀도 조건에 영향을 받지 않았지만, 수컷의 경우는 모든 외형적 특성에서 밀도에 영향을 받는 것으로 분석되었으며, 특히 20마리조건은 다른 조건보다 성충의 크기가 큰 것을 확인하였다. 결론적으로, 흰점박이꽃무지의 3령 유충은 3L 사육상자에 30마리 밀도 조건으로 사육하는것이 효율적이며, 이를 통해 3령에서 성충까지의 발육기간을 단축할 수 있고, 사료 비용도 절감할 수 있다

Novel Photocatalysts for Oxidation and Reduction of Water under Visible Light: Substitution Effect of Lead in Perovskite-Related Layered Oxides

지상민,김현규,장점석,배상원,손효창,이재성 한국공업화학회 2004 한국공업화학회 연구논문 초록집 Vol.2004 No.1

Prepatation of N-doped Strontium niobium oxide for the photocatalytic water splitting

지상민,김현규,장점석,배상원,이재성 한국화학공학회 2007 화학공학의이론과응용 Vol.10 No.1

A layered perovskite Sr2Nb2O7 was reported to be an efficient photocatalyst for water splitting under UV light irradiation. Nitrogen doping turned this material into a photocatalyst active for water splitting under visible light irradiation (>400 nm). Sr2Nb2O7-xNx catalysts were synthesized from aSr2Nb2O7 precursor at 973 ~ 1273 K for 3 hr under NH3 flow (100 sccm). The catalyst obtained by nitriding Sr2Nb2O7 at 1073 K for 3 hr had the highest activity under visible light irradiation. Amount of a doped nitrogen increased as nitridation temperature increased. An excess N-doping induced the collapse of the layered structure of parent oxide. The obtained new crystal structure, SrNbO2N (oxynitride) was less active as photocatalyst for water splitting under visible light irradiation. This the optimum structure of the catalyst showing the maximum hydrogen production rate was the one that maintaining the layered structure of Sr2Nb2O7, with a part of its oxygen was replaced with nitrogen.

곤충자원을 이용한 치유농업프로그램의 효과와 지속성

김소윤,지상민,김선영,송정훈 한국환경과학회 2022 한국환경과학회지 Vol.31 No.11

This study aimed to investigate the psychological effect of a healing agricultural program using insect resources and to confirm the persistence of the healing effect. The applied insect was silkworm Bombyx mori (Linnaeus) and 28 adolescents from local children’s centers were used as the subjects. The healing agricultural program was applied at the centers for 4 weeks, and stress was measured using salivary ɑ-amylase levels and a survey on life satisfaction, happiness, and insect preference. The results indicated that stress levels decreased, and life satisfaction and insect preference increased after participating in the the healing agricultural program. Furthermore, it was confirmed that this healing effect continued until 4 weeks after the end of the program.

식용곤충 갈색거저리의 장기저장 및 생육 조절 연구

김찬욱,지상민,장규동,김소윤,송정훈 한국환경과학회 2024 한국환경과학회지 Vol.33 No.1

We introduce a method for preserving yellow mealworm (Tenebrio molitor Linnaeus) larvae for an extended period and show that a high percentage of larvae can survive in good health under low-temperature storage conditions combined with specific diapause termination conditions. Our study revealed that we can regulate important development stages such as pupation, emergence, and larval duration by adjusting these termination conditions. When storing larvae aged 6–8 weeks for 140 days, the storage temperature can be varied based on our goals, giving us control over yellow mealworm production to meet specific demands. To produce adult beetles, storing larvae at 15 ℃ with wheat bran and ending diapause at 30 ℃ resulted in 90% pupation rate, with 60% becoming adults in 21 days. If our aim is larvae production, storing them at 10–12 ℃ with wheat bran and ending diapause at 25–30 ℃ allows the larvae to reach a suitable weight for processing. This straightforward approach ensures long-term storage of yellow mealworm larvae and provides a practical way to control their development, allowing efficient mass production tailored to market demands.

곤충치유프로그램의 운영 방식에 따른 치유효과 분석

김소윤,지상민,고혜진,박지연,김선영 한국환경과학회 2023 한국환경과학회지 Vol.32 No.11

Insect-healing program is designed for farm use, and this study attempted to scientifically investigate its healing effect based on application methods. The insects used in the program are Trypoxylus dichotomus, Papilio xuthus, and Bombyx mori. Insect-healing program was conducted on 61 adults of over 50 years old and at two different farms for 12 weeks. Stress, anxiety, insomnia, and cognitive function were measured through pre- and post-surveys. Result of the analysis showed that cognitive function changed positively in direct experienced group (those that encountered live insects) and indirect experience group (those that encountered insects only through photos and videos). However, psychological variables such as stress and anxiety decreased and sleep time increased significantly only in the direct experienced group. These results demonstrate that an agro-healing program that involve the use of insect must include activities such as raising and taking care of live insect to achieve positive psychological healing effects.

정서곤충을 이용한 치유프로그램의 효과 분석 - 누에나방 사례 -

김소윤,지상민,김원태,송정훈 한국응용곤충학회 2022 한국응용곤충학회지 Vol.61 No.2

: The study aimed to investigate the psychological effect of the emotional insect-assisted healing program on the stress experienced by 43 children in a community child center. Silkworm (Bombyx mori [Linnaeus]) was used in the healing programs, and the stress was measured using salivary alpha-amylase levels. Saliva samples and a survey on insect preference were collected before and after the healing programs, respectively. After the program, the satisfaction score for participation was evaluated. Results indicate that the emotional insect-assisted healing programs led to statistically significant lowering of salivary alpha-amylase levels and increased the preference for insects. The children’s satisfaction in the healing program activities using silkworms was 4.23 points out of 5. 본 연구에서는 정서곤충을 이용한 치유프로그램 활동이 지역아동센터를 이용하는 아동 43명의 스트레스에 미치는 영향을 조사하였다. 치유 프로그램에 이용된 정서곤충은 누에나방[Bombyx mori (Linnaeus)]이다. 스트레스의 정도는 타액 ɑ-아밀라아제 검사를 통해 측정하였다. 스트레스 수치를 평가하기 위한 타액 샘플 채취와 곤충 선호도에 대한 설문지는 곤충치유활동에 참여하기 전과 후에 측정되었고, 활동을 모두 마친 후에는 치유프로그램 참여 경험에 대한 만족도 점수를 평가하였다. 분석 결과, 곤충치유프로그램 활동은 참여 아동의 타액 ɑ-아밀라아제 수치를 통계적으로 낮추고, 곤충에 대한 선호도를 높여준다는 결과를 확인하였다. 누에를 이용한 치유프로그램 활동에 참가한 아동들의 만족도는 5점 만점에 4.23 점이었다.