• Geomechanics and Engineering
• /
• v.39 no.1
• /
• pp.27-41
• /
• 2024

Accurately estimating the performance of tunnel boring machines (TBMs) is crucial for mitigating the substantial financial risks and complexities associated with tunnel construction. Machine learning (ML) techniques have emerged as powerful tools for predicting non-linear time series data. In this research, six advanced meta-heuristic optimization algorithms based on long short-term memory (LSTM) networks were developed to predict TBM penetration rate (TBM-PR). The study utilized 1125 datasets, partitioned into 20% for testing, 70% for training, and 10% for validation, incorporating six key input parameters influencing TBM-PR. The performances of these LSTM-based models were rigorously compared using a suite of statistical evaluation metrics. The results underscored the profound impact of optimization algorithms on prediction accuracy. Among the models tested, the LSTM optimized by the particle swarm optimization (PSO) algorithm emerged as the most robust predictor of TBM-PR. Sensitivity analysis further revealed that the orientation of discontinuities, specifically the alpha angle (α), exerted the greatest influence on the model's predictions. This research is significant in that it addresses critical concerns of TBM manufacturers and operators, offering a reliable predictive tool adaptable to varying geological conditions.

• Smart Structures and Systems
• /
• v.33 no.5
• /
• pp.325-332
• /
• 2024

This study proposes a novel long short-term memory (LSTM)-based approach for predicting carbonation depth, with the aim of enhancing the durability evaluation of concrete structures. Conventional carbonation depth prediction relies on statistical methodologies using carbonation influencing factors and in-situ carbonation depth data. However, applying in-situ data for predictive modeling faces challenges due to the lack of time-series data. To address this limitation, an LSTM-based carbonation depth prediction technique is proposed. First, training data are generated through random sampling from the distribution of carbonation velocity coefficients, which are calculated from in-situ carbonation depth data. Subsequently, a Bayesian theorem is applied to tailor the training data for each target bridge, which are depending on surrounding environmental conditions. Ultimately, the LSTM model predicts the time-dependent carbonation depth data for the target bridge. To examine the feasibility of this technique, a carbonation depth dataset from 3,960 in-situ bridges was used for training, and untrained time-series data from the Miho River bridge in the Republic of Korea were used for experimental validation. The results of the experimental validation demonstrate a significant reduction in prediction error from 8.19% to 1.75% compared with the conventional statistical method. Furthermore, the LSTM prediction result can be enhanced by sequentially updating the LSTM model using actual time-series measurement data.

• Journal of the Korean Solar Energy Society
• /
• v.39 no.5
• /
• pp.41-52
• /
• 2019

The purpose of the work is to develop a simple solar irradiance prediction model using a deep learning method, the LSTM (long term short term memory). Other than existing prediction models, the proposed one uses only the cloudiness among the information forecasted from the national meterological forecast center. The future cloudiness is generally announced with four categories and for three-hour intervals. In this work, a daily irradiance pattern is used as an input vector to the LSTM together with that cloudiness information. The proposed model showed an error of 5% for learning and 30% for prediction. This level of error has lower influence on the load prediction in typical building cases.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.8
• /
• pp.307-324
• /
• 2018

Recently, recurrent neural networks have been attracting attention in solving prediction problem of sequential data through structure considering time dependency. However, as the time step of sequential data increases, the problem of the gradient vanishing is occurred. Long short-term memory models have been proposed to solve this problem, but there is a limit to storing a lot of data and preserving it for a long time. Therefore, research on memory-augmented neural network (MANN), which is a learning model using recurrent neural networks and memory elements, has been actively conducted. In this paper, we describe the structure and characteristics of MANN models that emerged as a hot topic in deep learning field and present the latest techniques and future research that utilize MANN.

• The Korean Journal of Physiology and Pharmacology
• /
• v.18 no.6
• /
• pp.457-460
• /
• 2014

At central synapses, activity-dependent synaptic plasticity has a crucial role in information processing, storage, learning, and memory under both physiological and pathological conditions. One widely accepted model of learning mechanism and information processing in the brain is Hebbian Plasticity: long-term potentiation (LTP) and long-term depression (LTD). LTP and LTD are respectively activity-dependent enhancement and reduction in the efficacy of the synapses, which are rapid and synapse-specific processes. A number of recent studies have a strong focal point on the critical importance of another distinct form of synaptic plasticity, non-Hebbian plasticity. Non-Hebbian plasticity dynamically adjusts synaptic strength to maintain stability. This process may be very slow and occur cell-widely. By putting them all together, this mini review defines an important conceptual difference between Hebbian and non-Hebbian plasticity.

• KIISE Transactions on Computing Practices
• /
• v.23 no.4
• /
• pp.268-273
• /
• 2017

Recently, deep learning encoder-based approach has been actively applied in the field of sentiment classification. However, Long Short-Term Memory network deep learning encoder, the commonly used architecture, lacks the quality of vector representation when the length of the documents is prolonged. In this study, for effective classification of the sentiment documents, we suggest the use of attention method-based deep learning encoder that generates document vector representation by weighted sum of the outputs of Long Short-Term Memory network based on importance. In addition, we propose methods to modify the attention method-based deep learning encoder to suit the sentiment classification field, which consist of a part that is to applied to window attention method and an attention weight adjustment part. In the window attention method part, the weights are obtained in the window units to effectively recognize feeling features that consist of more than one word. In the attention weight adjustment part, the learned weights are smoothened. Experimental results revealed that the performance of the proposed method outperformed Long Short-Term Memory network encoder, showing 89.67% in accuracy criteria.

• Science of Emotion and Sensibility
• /
• v.21 no.4
• /
• pp.37-42
• /
• 2018

This study was conducted to evaluate the effects of LED illumination and task difficulty on a person's long-term memory. Illumination levels of 400 lx and 1,000 lx were employed in this study, and task difficulty was set at learning 4 words (easy task) and 7 words (difficult task). The person's retention rate of the learned task was designated as a dependent variable. A total of 64 subjects participated in this study, and 16 participants assigned to each of the four sets of conditions. The results indicated that the retention rate for the difficult task under relatively dark 400 lx conditions was 68.49%, while and the retention rate was higher than 56.03% for 1,000 lx conditions. In addition, for the easy task, the retention rate was 67.97% and 56.55% for the 400 1x and the 1,000 1x conditions, respectively. However, the interaction between illumination and task difficulty was not statistically significant. The study results further suggested that long-term memory can be effective in relatively dark conditions and indirectly suggests that long-term memory may not follow the Yokers-Dodson law.

• Journal of Korean Institute of Industrial Engineers
• /
• v.23 no.3
• /
• pp.451-467
• /
• 1997

Tabu search (TS) is a useful strategy that has been successfully applied to a number of complex combinatorial optimization problems. By guiding the search using flexible memory processes and accepting disimproved solutions at some iterations, TS helps alleviate the risk of being trapped at a local optimum. In this article, we propose TS-based heuristics for solving a class of clustering problems, and compare the relative performances of the TS-based heuristic and the simulated annealing (SA) algorithm. Computational experiments show that the TS-based heuristic with a long-term memory offers a higher possibility of finding a better solution, while the TS-based heuristic without a long-term memory performs better than the others in terms of the combined measure of solution quality and computing effort required.

• Journal of Korean Biological Nursing Science
• /
• v.22 no.2
• /
• pp.111-118
• /
• 2020

Purpose: The purpose of this study was to understand the impact of long- and short-term energy drinks on anxiety-like, depressionlike, and cognitive behavior in adolescent rats. Methods: Adolescent rats (age six weeks) were randomly classified into a control group (CON), a long-term administration group (LT), and a short-term administration group (ST). The LT group was orally administered 1.5 mL/100 g (body weight) of energy drink twice daily for 14 days, the ST group was orally administered for one day, and the control group applied the same amount of normal saline. Later, an open-field test, a forced swim test, novel object recognition test, and an 8-arm radial maze test was conducted to assess the rats' anxiety, depression, and cognitive function. Results: There were different effects in the long- and short-term groups of energy drink administration. In the LT group, anxiety- and depressive-like behavior increased because of increased movement in the side corner and decrease of immobility time. Also, the time to explore novel objects decreased, and the number of correct responses was reduced, indicating a learning and memory function disorder. However, the ST group was not different from the control group. Conclusion: These results indicate that long-term consumption of energy drinks can increase anxiety-like, depression-like behavior, and this can lead to decrease in learning and memory functions. Thus, nurse and health care providers should understand the impact of energy drink consumption in adolescence to provide appropriate practices and education.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.5 no.4
• /
• pp.169-177
• /
• 2009

In come cases, people differently respond on the same joke or thoughtless behavior - sometimes like it and laugh, another time feel annoyed or angry. This fact is explained that experiences which we had in the past are remembered by emotional memory, so they cause different responses. When people face similar situation or feel similar emotion, they evoke the emotion experienced in the past and the emotional memory affects current emotion. This paper suggested the mechanism of the emotional memory using SOM through the similarity between the emotional memory and SOM learning algorithm. It was assumed that the mechanism of the emotional memory has also the characteristics of association memory, long-term memory and short-term memory in its process of remembering emotional experience, which are known as the characteristics of the process of remembering factual experience. And then these characteristics were applied. The mechanism of the emotional memory designed like this was applied to toy hammer game and I measured the change in the power of toy hammer caused by differently responding on the same stimulus. The mechanism of the emotional memory suggest in above is expected to apply to the fields of game, robot engineering, because the mechanism can express various emotions on the same stimulus.