• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2008년도 하계종합학술대회
• /
• pp.679-680
• /
• 2008

When we apply the neural networks to applications, we need to select proper architecture of the network and the activation function of the network is one of most important characteristics. In this research, we propose a method to make a network using multiple activation functions. The performance of the proposed method is investigated through the computer simulations on various regression problems.

• 대한물리의학회지
• /
• 제5권3호
• /
• pp.395-404
• /
• 2010

Purpose : The object of this study was to examine the effect of motor learning on brain activation depending on the method of motor learning. Methods : The brain activation was measured in 9 men by fMRI. The subjects were divided into the following groups depending on the method of motor learning: actually practice (AP, n=3) group, action observation (AO, n=3) group and motor imagery (MI, n=3) group. In order to examine the effect of motor learning depending on the method of motor learning, the brain activation data were measured during learning. For the investigation of brain activation, fMRI was conducted. Results : The results of brain activation measured before and during learning were as follows; (1) During learning, the AP group showed the activation in the following areas: primary motor area located in precentral gyrus, somatosensory area located in postcentral gyrus, supplemental motor area and prefrontal association area located in precentral gyrus, middle frontal gyrus and superior frontal gyrus, speech area located in superior temporal gyrus and middle temporal gyrus, Broca's area located in inferior parietal lobe and somatosensory association area of precuneus; (2) During learning, the AD groups showed the activation in the following areas: primary motor area located in precentral gyrus, prefrontal association area located in middle frontal gyrus and superior frontal gyrus, speech area and supplemental motor area located in superior temporal gyrus and middle temporal gyrus, Broca's area located in inferior parietal lobe, somatosensory area and primary motor area located in precentral gyrus of right cerebrum and left cerebrum, and somatosensory association area located in precuneus; and (3) During learning, the MI group showed activation in the following areas: speech area located in superior temporal gyrus, supplemental area, and somatosensory association area located in precuneus. Conclusion : Given the results above, in this study, the action observation was suggested as an alternative to motor learning through actual practice in serial reaction time task of motor learning. It showed the similar results to the actual practice in brain activation which were obtained using activation of mirror neuron. This result suggests that the brain activation occurred by the activation of mirror neuron, which was observed during action observation. The mirror neurons are located in primary motor area, somatosensory area, premotor area, supplemental motor area and somatosensory association area. In sum, when we plan a training program through physiotherapy to increase the effect during reeducation of movement, the action observation as well as best resting is necessary in increasing the effect of motor learning with the patients who cannot be engaged in actual practice.

• Reproductive and Developmental Biology
• /
• 제29권4호
• /
• pp.229-234
• /
• 2005

Methods for activation of reconstructed oocytes were examined for the production of nuclear transfer (NT) rat embryos using fetal neural stem cells as donor. Neural stem cells were isolated from Day 14.5 rat fetuses, and the oocytes for recipient cytoplasm were recovered from 4-week old Sprague Dawley rats. After enucleation and nuclear injection, the reconstructed oocytes were immediately exposed to activation medium consisting of 10 mM $SrCl_2$ for 4 h (immediate activation after injection; IAI), or cultured in vitro for $2\~3$ h before activation treatment (injection before activation; IBA). Pre-activated oocytes were also used for NT to test reprogramming potential of artificially activated oocytes. The oocytes were grouped as IIA (immediate injection after activation) and ABI (activation $2\~3$ h before injection). Following NT, the oocytes were cultured in vitro. Development of the NT embryos was monitored at 44 and 119 h after activation. The embryos in groups IAI, mA, and IIA were cleaved to the 2-cell stage at the rates of $36.6\%\;(15/41),\;39.5\%\;(17/43)\;and\;46.3\%$ (25/54), respectively. However, in the ABI group, only one embryo ($1.8\%$, 1/55) was cleaved after activation. After in vitro culture, two NT embryos from IAI group had developed to the morula stage $(4.9\%\cdot2/41)$. However, no morula or blastocyst was obtained in the other groups. These results suggest that immediate activation after injection (IAI) method may be used for the production of rat somatic cell NT embryos.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제9권7호
• /
• pp.213-220
• /
• 2020

합성곱 신경망은 합성곱층과 완전연결층으로 구성되어 있다. 합성곱층과 완전연결층의 각 층에서는 비선형 활성함수를 사용하고 있다. 활성함수는 뉴런 간에 신호를 전달할 때 입력신호가 일정 기준 이상이면 신호를 전달하고 기준에 도달하지 못하면 신호를 보내지 않을 수 있는 뉴런의 정보전달 방법을 모사하는 함수이다. 기존의 활성함수는 손실함수와 관계성을 가지고 있지 않아 최적해를 찾아가는 과정이 늦어지는 점을 개선하기 위해 활성함수를 일반화한 민첩한 활성함수를 제안하였다. 민첩한 활성함수의 매개변수는 역전파 과정에서, 매개변수에 대한 손실함수의 1차 미분계수를 이용한 학습과정을 통해 최적의 매개변수를 선택하는 방법으로 손실함수를 감소시킴으로써 심층신경망의 성능을 향상시킬 수 있다. MNIST 분류문제를 통하여 민첩한 활성함수가 기존의 활성함수에 비해 우월한 성능을 가짐을 확인하였다.

• 한국수소및신에너지학회논문집
• /
• 제8권4호
• /
• pp.161-171
• /
• 1997

$AB_2$ type Zr-based Laves phase alloys have been studied for potential application as negative electrode in Ni/MH batteries. However, They have a serious disadvantage of poor activation behavior in KOH solution. In this work, a new method of alloy design method was tried for improving Zr-based alloy activation. this method has focused on phase controlling to make multi-phase microstructure. In the case of multi-phase Zr-V-Mn-Ni shows good performance in activation, but activation mechanism has not been known. So, we were in search of elucidating this mechanism, Using morphological and electrochemical analysis, we could find that surface morphology and electocatalytic activity of the alloy change during immersion in KOH solution. V-rich second phases are selectively corroded and dissolved and then become Ni-rich phases. Resulting from these surface reaction in KOH solution, self-hydrogen charging occurs through Ni-rich phase. However, the alloy has poor cyclic durability because of such a corrosion mechanism. Therefore, finally we developed durable alloys by substitution of other alloying element.

• 동아시아식생활학회지
• /
• 제20권1호
• /
• pp.54-59
• /
• 2010

Basil is known to contain six types of polyphenols that engage in physiological activation; protocateuic acid, caffeic acid, chlorogenic acid, courmaric acid, rosmarinic acid and quercetin. In this study, the antioxidants in eight types of basil were evaluated. Specifically, the antioxidative activation of basil was evaluated based on the relationship between active oxygen scavenging (DPPH radical-scavenging), which was used as an index showing the content and functionality of the polyphenol compounds in basil, and $Fe^{2+}$/ascorbate (FTC) and 2-thiobarbituric acid (TBA). The total polyphenol content of the different types basil occurred in the following order: Dark-opal> Lettuce> Bush> Greek> Lemon> Sweet> Geno> Holy. The highest content Dark-opal was 173.3 mg, which was about three times greater than the lowest content Holy, which was 49.85 mg. In addition, DPPH radical-scavenging by Dark-opal 51%, which was the highest scavenging activity occurred in the folling order: Dark-opal> Lettuce> Bush> Greek> Lemon> Sweet> Geno> Holy. The antioxidative activation values measured using the FTC and TBA were the same as the value obtained using the DPPH method. Finally, the level of antioxidative activation measured using FTC, TBA and DPPH methods showed that the higher the content of polyphenol substance was, the stronger the antioxidative activation became.

• 한국안전학회지
• /
• 제18권3호
• /
• pp.81-87
• /
• 2003

Latex polymers are widely used for adhesive, binder, paint etc. Especially the PVAc(Polyvinyl acetate) latex which manufactured by vinyl acetate and vinyl alcohol as protective colloid is a useful environmentally friendly adhesive. To increase useful property of PVAc latex, this study was carried out for checking thermal characteristics and physical condition of PVAc latex by DSC, FT-IR, Pyrolyzer GC-MS. The activation energies of thermal decomposition for 40, 48, 56, 64% solid content of PVAc latex were found as 28.1-36.0kcal/mol by Kissinger's method and 17.2-22.0kcal/mol by DSC method. Actually, reasonable solid content could be consiered as 56% because of activation energy and adhesive characteristics. According to the effect of protective colloid for 4, 10, 15, 20wt%, the activation energy shows same tendency to both method and in case of l5wt% has been found as the highest activation energy. The mechanism of thermal decomposition was mainly estimated by main chain scission, not by side group on FT-IR analysis. Main component of Pyrolzer GC-MS result were consisted of $CH_3COOH$, $CH_3$, $H_2O$ and light gases(CO, $CO_2$, $CH_4$ etc).

• 한국산학기술학회논문지
• /
• 제18권10호
• /
• pp.75-80
• /
• 2017

본 논문에서는 딥러닝에서 활용되는 정규화(regularization) 및 항등사상(identity mapping)이 활성함수(activation function) 성능에 미치는 영향에 대해 설명한다. 딥러닝에서 활성함수는 비선형 변환을 위해 사용된다. 초기에는 sigmoid 함수가 사용되었으며, 기울기가 사라지는 기존의 활성함수의 문제점을 극복하기 위해 ReLU(Rectified Linear Unit), LReLU(Leaky ReLU), PReLU(Parametric ReLU), ELU(Exponetial Linear Unit)이 개발되었다. 활성함수와의 연구와는 별도로 과적합(Overfitting)문제를 해결하기 위해, Dropout, 배치 정규화(Batch normalization) 등의 정규화 방법들이 개발되었다. 추가적으로 과적합을 피하기 위해, 일반적으로 기계학습 분야에서 사용되는 data augmentation 기법이 활용된다. 딥러닝 구조의 측면에서는 기존에 단순히 컨볼루션(Convolution) 층을 쌓아올리는 구조에서 항등사상을 추가하여 순방향, 역방향의 신호흐름을 개선한 residual network가 개발되었다. 위에서 언급된 활성함수들은 각기 서로 다른 특성을 가지고 있으나, 새로운 정규화 및 딥러닝 구조 연구에서는 가장 많이 사용되는 ReLU에 대해서만 검증되었다. 따라서 본 논문에서는 정규화 및 항등사상에 따른 활성함수의 성능에 대해 실험적으로 분석하였다. 분석을 통해, 정규화 및 항등사상 유무에 따른 활성함수 성능의 경향을 제시하였으며, 이는 활성함수 선택을 위한 교차검증 횟수를 줄일 수 있을 것이다.

• PNF and Movement
• /
• 제19권2호
• /
• pp.269-278
• /
• 2021

Purpose: This study aimed to investigate effective cueing methods for selective muscle activation of the serratus anterior. Methods: Based on the inclusion criteria, 26 healthy adults, both males and females, were recruited for the measurement of muscle activation of the upper trapezius and serratus anterior muscles while performing basic movements in knee push-up plus (KPP) and dynamic hug (DH) positions using five different cueing methods. An electromyogram was used to measure muscle activation, and both muscle activation and muscle ratio (serratus anterior/upper trapezius) were compared during the basic movements and different cueing methods. The cueing methods were trapezius verbal cueing, trapezius verbal cueing + trapezius tactile cueing, emphasis verbal cueing, serratus anterior tactile cueing, and trapezius verbal cueing + trapezius tactile cueing + serratus anterior tactile cueing. Results: The results of the study showed that there was a significant difference in the muscles for the two exercises (p < 0.05). There was also a significant difference between the cueing methods (p < 0.05). The correlative effect between the muscles and cueing methods was also significant (p < 0.05). The muscle ratio in trapezius verbal cueing + trapezius tactile cueing + serratus anterior tactile cueing during KPP and DH was higher than in basic movements and other cueing methods. This confirms that trapezius verbal cueing + trapezius tactile cueing + serratus anterior tactile cueing is an effective cueing method for selective activation of the serratus anterior during KPP and DH. This study also demonstrated that cueing by a therapist may both increase and decrease selective muscle activation. Conclusion: Through this study, an effective cueing method to selectively activate the serratus anterior may be suggested, and the results of this study may provide basic information regarding future studies and clinical practice.

• 한국환경과학회지
• /
• 제22권3호
• /
• pp.331-339
• /
• 2013

Efforts were made to determine the activation energy and the reaction order by adopting Kissinger and Flynn-Wall-Ozawa analysis methods. All the data were acquired from TGA thermograms for the mixed fuels with different temperature heating rates. It could be known that both the coal and the mixed fuels decomposed thermally at temperature ranges of $300{\sim}700^{\circ}C$. The temperature at the maximum reaction rate, Tp, could be determined by DTG method, which could be obtained by differentiation of TGA thermogram. Kissinger analysis showed the linear relationship with experimental data, showing the activation energy of $319.64{\pm}4$ kJ/mol. From Flynn-Wall-Ozawa analysis, it was shown that the activation energies and the reaction orders did not undergo any significant changes with both the conversions and the heating rates. It was considered from this facts that the combustion mechanism of the mixed fuels could not be affected by the extent of conversion and heating rate. In the present study, the activation energies showed different values according to the different analysis methods. The difference might be originated from the inconsistency of the mathematical data treatment method. In other words, while the activation energies obtained from the Kissinger method indicated the average values for overall reaction, that from Flynn-Wall-Ozawa method showed the average values for the each conversion around Tp.