http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Seo Young Kwon,Gyujin Seo,Mirae Jang,Hanbyul Shin,Wooseok Choi,You Bin Lim,Min-Sup Shin,Bung-Nyun Kim 대한정신약물학회 2024 CLINICAL PSYCHOPHARMACOLOGY AND NEUROSCIENCE Vol.22 No.1
Objective: To examine the effect of mobile neurofeedback training on the clinical symptoms, attention abilities, and execution functions of children with attention deficit hyperactivity disorder (ADHD). Methods: The participants were 74 children with ADHD aged 8−15 years who visited the Department of Child and Adolescent Psychiatry at Seoul National University Children’s Hospital. The participants were randomly assigned to the mobile neurofeedback (n = 35) or control (sham; n = 39) group. Neurofeedback training was administered using a mobile app (equipped with a headset with a 2-channel electroencephalogram [EEG] sensor) for 30 min/day, 3 days/week, for 3 months. Children with ADHD were individually administered various neuropsychological tests, including the continuous performance test, Children’s Color Trails Test-1 and 2, and Stroop Color and Word Tests. The effects of mobile neurofeedback were evaluated at baseline and at 3 and 6 months after treatment initiation. Results: Following treatment, both mobile neurofeedback-only and sham-only groups showed significant improvements in attention and response inhibition. In the visual continuous performance test, omission errors decreased to the normal range in the mobile neurofeedback-only group after training, suggesting that mobile neurofeedback effectively reduced inattention in children with ADHD. In the advanced test of attention, auditory response times decreased in the mobile neurofeedback + medication group after training, but increased in the sham+medication group. Overall, there were no significant between-group differences in other performance outcomes. Conclusion: Mobile neurofeedback may have potential as an additional therapeutic option alongside medication for children with ADHD.
발목의 해부학적 회전구조를 구현하고 제어하는 발목재활로봇
김인우(Inwoo Kim),김규진(Gyujin Kim),권경민,장재용(Jeung Jang),임서균(Yim Seo Gyun),조대기(Daeki Cho),이수홍(Soo-Hong Lee),Gapsun Kim 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.11
When physical ability is reduced due to aging and various diseases, the athletic ability of the ankle, which plays a very important role in walking, is significantly weakened. Various ankle exoskeleton robots have been developed to assist and rehabilitate ankles with reduced mobility, but most do not implement the correct rotation axis of the ankle, making it difficult to transmit the correct force and increase the risk of injury. In this study, the anatomical axis of rotation of the ankle joint was structurally implemented. Two linear actuators were installed in the embodied structure to control the two rotation axes of the robot. By installing a load cell between the linear actuator and the structure, the magnitude of the torque transmitted by the robot to the human can be calculated inverse dynamic. The ankle exoskeleton robot can effectively simulate and support all movements of the human ankle.
( Donghyun Lee ),( Hosung Park ),( Soonshin Seo ),( Hyunsoo Son ),( Gyujin Kim ),( Ji-hwan Kim ) 한국인터넷정보학회 2021 KSII Transactions on Internet and Information Syst Vol.15 No.3
Recurrent neural network (RNN) architectures have been used for language modeling (LM) tasks that require learning long-range word or character sequences. However, the RNN architecture is still suffered from unstable gradients on long-range sequences. To address the issue of long-range sequences, an attention mechanism has been used, showing state-of-the-art (SOTA) performance in all LM tasks. A differentiable neural computer (DNC) is a deep learning architecture using an attention mechanism. The DNC architecture is a neural network augmented with a content-addressable external memory. However, in the write operation, some information unrelated to the input word remains in memory. Moreover, DNCs have been found to perform poorly with low numbers of weight parameters. Therefore, we propose a robust memory deallocation method using a limited retention vector. The limited retention vector determines whether the network increases or decreases its usage of information in external memory according to a threshold. We experimentally evaluate the robustness of a DNC implementing the proposed approach according to the size of the controller and external memory on the enwik8 LM task. When we decreased the number of weight parameters by 32.47%, the proposed DNC showed a low bits-per-character (BPC) degradation of 4.30%, demonstrating the effectiveness of our approach in language modeling tasks.