• International Journal of Advanced Culture Technology
• /
• v.5 no.1
• /
• pp.83-89
• /
• 2017

In this paper, we propose an extended method of one-hot Long Short-Term Memory (LSTM) and evaluate the performance on spam filtering task. Most of traditional methods proposed for spam filtering task use word occurrences to represent spam or non-spam messages and all syntactic and semantic information are ignored. Major issue appears when both spam and non-spam messages share many common words and noise words. Therefore, it becomes challenging to the system to filter correct labels between spam and non-spam. Unlike previous studies on information filtering task, instead of using only word occurrence and word context as in probabilistic models, we apply a neural network-based approach to train the system filter for a better performance. In addition to one-hot representation, using term weight with attention mechanism allows classifier to focus on potential words which most likely appear in spam and non-spam collection. As a result, we obtained some improvement over the performances of the previous methods. We find out using region embedding and pooling features on the top of LSTM along with attention mechanism allows system to explore a better document representation for filtering task in general.

• Journal of Life Science
• /
• v.29 no.5
• /
• pp.564-569
• /
• 2019

Heavy drinking disrupts the nervous system by activation of GABA receptors and inhibition of glutamate receptors, thereby preventing short-term memory formation. Degradation of cognition by alcohol induces blackouts, and it can lead to alcoholic dementia if repeated. Therefore, drugs need to be developed to prevent alcohol-induced blackout. In this study, we confirmed the effect of an ethanol extract of Cassia obtusifolia seeds (COE) on alcohol-induced memory impairment. The effects of COE and ethanol on cognitive functions mice were examined using the passive avoidance and Y-maze tests. The manner in which alcohol affects long-term potentiation (LTP) in relation to the learning and memory was confirmed by electrophysiology performed on mouse hippocampal slices. We also measured N-methyl-D-aspartate (NMDA) receptor-mediated field excitatory synapses (fEPSPs), which have a known association with cognitive impairment caused by ethanol. Ethanol caused memory impairments in passive avoidance and Y-maze tests. COE prevented these ethanol-induced memory impairments in these tests. Ethanol also blocked LTP induction in the mouse hippocampus, and COE prevented this ethanol-induced LTP deficit. Ethanol decreased NMDA receptor-mediated fEPSPs in the mouse hippocampus, and this decrease was prevented by COE. These results suggest that COE might be useful in preventing alcohol-induced neurological dysfunctions, including blackouts.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.55 no.8
• /
• pp.399-403
• /
• 2006

An accurate constant output voltage provided by the analog memory cell may be used by the low power oscillator to generate an accurate low frequency output signal. This accurate low frequency output signal may be used to maintain long-term timing accuracy in host devices during sleep modes of operation when an external crystal is not available to provide a clock signal. Further, incorporation of the analog memory cell in the low power oscillator is fully implementable in a 0.35um Samsung standard CMOS process. Therefore, the analog memory cell incorporated into the low power oscillator avoids the previous problems in a oscillator by providing a temperature-stable, low power consumption, size-efficient method for generating an accurate reference clock signal that can be used to support long sleep mode operation.

• International journal of advanced smart convergence
• /
• v.7 no.3
• /
• pp.1-7
• /
• 2018

This study relates to a composition for improvement of memory or prevention and treatment of cognitive impairment using waste resources rich in beneficial substances. This study makes good effects to inhibit the activity of acetylcholinesterase in brain tissue and to improve the cognitive functions in a simulation model of cognitive impairment induced by scopolamine, so it can be available in the promotion of memory and the prevention and treatment of cognitive impairment. The composition uses the extract of fruit peels, which have long been used without causing toxicity in a wide range of food applications; therefore, it can be used safely without a risk of side effects, even in the case of a long-term administration for the preventive purpose. Furthermore, this research is a very beneficial invention in the environment-friendly aspect in association with the recycling of resources, as it is based on the novel efficacies of fruit peels, which have been conventionally disposed as a refuse of fruits due to their poor sensory qualities despite the content of beneficial substances.

• Molecules and Cells
• /
• v.39 no.7
• /
• pp.515-523
• /
• 2016

Long-term immunity to many viral and bacterial pathogens requires$ CD8^+$ memory T cell development, and the induction of long-lasting$ CD8^+$ memory T cells from a $na{\ddot{i}}ve$, undifferentiated state is a major goal of vaccine design. Formation of the memory$ CD8^+$ T cell compartment is highly dependent on the early activation cues received by $na{\ddot{i}ve}$ $CD8^+$ T cells during primary infection. This review aims to highlight the cellularity of various niches within the lymph node and emphasize recent evidence suggesting that distinct types of T cell activation and differentiation occur within different immune contexts in lymphoid organs.

• Biomolecules & Therapeutics
• /
• v.28 no.6
• /
• pp.512-518
• /
• 2020

Stroke is a leading cause of long-term disability in ischemic survivors who are suffering from motor, cognitive, and memory impairment. Previously, we have reported suppressing LPA₅ activity with its specific antagonist can attenuate acute brain injuries after ischemic stroke. However, it is unclear whether suppressing LPA₅ activity can also attenuate chronic brain injuries after ischemic stroke. Here, we explored whether effects of LPA₅ antagonist, TCLPA5, could persist a longer time after brain ischemic stroke using a mouse model challenged with tMCAO. TCLPA5 was administered to mice every day for 3 days, starting from the time immediately after reperfusion. TCLPA5 administration improved neurological function up to 21 days after tMCAO challenge. It also reduced brain tissue loss and cell apoptosis in mice at 21 days after tMCAO challenge. Such long-term neuroprotection of TCLPA5 was associated with enhanced neurogenesis and angiogenesis in post-ischemic brain, along with upregulated expression levels of vascular endothelial growth factor. Collectively, results of the current study indicates that suppressing LPA₅ activity can provide long-term neuroprotection to mice with brain ischemic stroke.

• Journal of Information Processing Systems
• /
• v.14 no.6
• /
• pp.1508-1520
• /
• 2018

In this paper, we propose an improved model to provide users with a better long-term prediction of waterworks operation data. The existing prediction models have been studied in various types of models such as multiple linear regression model while considering time, days and seasonal characteristics. But the existing model shows the rate of prediction for demand fluctuation and long-term prediction is insufficient. Particularly in the deep running model, the long-short-term memory (LSTM) model has been applied to predict data of water purification plant because its time series prediction is highly reliable. However, it is necessary to reflect the correlation among various related factors, and a supplementary model is needed to improve the long-term predictability. In this paper, convolutional neural network (CNN) model is introduced to select various input variables that have a necessary correlation and to improve long term prediction rate, thus increasing the prediction rate through the LSTM predictive value and the combined structure. In addition, a multiple linear regression model is applied to compile the predicted data of CNN and LSTM, which then confirms the data as the final predicted outcome.

• Animal cells and systems
• /
• v.2 no.1
• /
• pp.65-69
• /
• 1998

The hippocampus and prefrontal cortex are regarded as the highest-order association cortices. The hippocampus has been proposed to store "cognitive maps" of external environments, and the prefrontal cortex is known to be involved in the planning of behavior, among other functions. Considering the prominent functional roles played by these structures, it is not surprising to find direct monosynaptic projections from the hippocampus to the prefrontal cortex. Rhythmic stimulation of this projection patterned after the hippocampal EEG theta rhythm induced stable long-term potentiation of field potentials in the prefrontal cortex. Comparison of behavioral correlates of hippocampal and prefrontal cortical neurons during an a-arm radial maze, working memory task shows a striking contrast. Hippocampal neurons exhibit clear place-specific firing patterns, whereas prefrontal cortical neurons do not show spatial selectivity, but are correlated to different stages of the behavioral task. These data lead to the hypothesis that the role of hippocampal projection to the prefrontal cortex is not to impose spatial representations upon prefrontal activity, but to provide a mechanism for learning the spatial context in which particular behaviors are appropriate.propriate.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.559-560
• /
• 2018

순환 신경망(RNN) 기반의 Long Short-Term Memory(LSTM)는 자연어처리 분야에서 우수한 성능을 보이는 모델이다. 음성을 문자로 변환해주는 Speech to Text (STT)를 이용해 자막을 생성하고, 생성된 자막을 다른 언어로 동시에 번역을 해주는 서비스가 활발히 진행되고 있다. STT를 사용하여 자막을 추출하는 경우에는 마침표가 없이 전부 연결된 문장이 생성되기 때문에 정확한 번역이 불가능하다. 본 논문에서는 영어자막의 자동 번역 시, 정확도를 높이기 위해 텍스트를 문장으로 분할하여 마침표를 생성해주는 방법을 제안한다. 이 때, LSTM을 이용하여 데이터를 학습시킨 후 테스트한 결과 62.3%의 정확도로 마침표의 위치를 예측했다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06b
• /
• pp.321-323
• /
• 2012

스마트폰 사용의 보편화와 센서기술의 발달로 이를 응용하는 다양한 연구가 진행되고 있다. 특히 가속도, GPS, 조도, 방향센서 등의 센서들이 스마트폰에 부착되어 출시되고 있어서, 이를 이용한 상황인지, 행동인식 등의 관련 연구들이 활발하다. 하지만 다양한 클래스를 분류하면서 높은 인식률을 유지하는 것은 어려운 문제이다. 본 논문에서는 인식률 향상을 위해 계층적 구조의 순환 신경망을 이용하여 제스처를 인식한다. 스마트폰의 가속도 센서를 이용하여 사용자의 제스처 데이터를 수집하고 BLSTM(Bidirectional Long Short-Term Memory) 구조의 순환신경망을 계층적으로 사용하여, 20가지 사용자의 제스처와 비제스처를 분류한다. 약 24,850개의 시퀀스 데이터를 사용하여 실험한 결과, 기존 BLSTM은 평균 89.17%의 인식률을 기록한 반면 계층적 BLSTM은 평균 91.11%의 인식률을 나타내었다.