• Geomechanics and Engineering
• /
• v.31 no.3
• /
• pp.265-279
• /
• 2022

Tunnel convergence prediction is essential for the safe construction and design of tunnels. This study proposes five machine learning models of deep neural network (DNN), K-nearest neighbors (KNN), Gaussian process regression (GPR), support vector regression (SVR), and decision trees (DT) to predict the convergence phenomenon during or shortly after the excavation of tunnels. In this respect, a database including 650 datasets (440 for training, 110 for validation, and 100 for test) was gathered from the previously constructed tunnels. In the database, 12 effective parameters on the tunnel convergence and a target of tunnel wall convergence were considered. Both 5-fold and hold-out cross validation methods were used to analyze the predicted outcomes in the ML models. Finally, the DNN method was proposed as the most robust model. Also, to assess each parameter's contribution to the prediction problem, the backward selection method was used. The results showed that the highest and lowest impact parameters for tunnel convergence are tunnel depth and tunnel width, respectively.

• Journal of Medicine and Life Science
• /
• v.20 no.4
• /
• pp.141-157
• /
• 2023

Stroke is a leading cause of disability and death. The condition requires prompt diagnosis and treatment. The quality of care provided to patients with stroke can vary depending on the availability of medical resources, which in turn, can affect prognosis. Recently, there has been growing interest in using machine learning (ML) to support stroke diagnosis and treatment decisions based on large medical data sets. Current ML applications in stroke care can be divided into two categories: analysis of neuroimaging data and clinical information-based predictive models. Using ML to analyze neuroimaging data can increase the efficiency and accuracy of diagnoses. Commercial software that uses ML algorithms is already being used in the medical field. Additionally, the accuracy of predictive ML models is improving with the integration of radiomics and clinical data. is expected to be important for improving the quality of care for patients with stroke.

• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.47 no.3
• /
• pp.68-74
• /
• 2024

This study develops a model to determine the input rate of the chemical for coagulation and flocculation process (i.e. coagulant) at industrial water treatment plant, based on real-world data. To detect outliers among the collected data, a two-phase algorithm with standardization transformation and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is applied. In addition, both of the missing data and outliers are revised with linear interpolation. To determine the coagulant rate, various kinds of machine learning models are tested as well as linear regression. Among them, the random forest model with min-max scaled data provides the best performance, whose MSE, MAPE, R² and CVRMSE are 1.136, 0.111, 0.912, and 18.704, respectively. This study demonstrates the practical applicability of machine learning based chemical input decision model, which can lead to a smart management and response systems for clean and safe water treatment plant.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Journal of Computing Science and Engineering
• /
• v.7 no.2
• /
• pp.99-111
• /
• 2013

Approximating a divergence between two probability distributions from their samples is a fundamental challenge in statistics, information theory, and machine learning. A divergence approximator can be used for various purposes, such as two-sample homogeneity testing, change-point detection, and class-balance estimation. Furthermore, an approximator of a divergence between the joint distribution and the product of marginals can be used for independence testing, which has a wide range of applications, including feature selection and extraction, clustering, object matching, independent component analysis, and causal direction estimation. In this paper, we review recent advances in divergence approximation. Our emphasis is that directly approximating the divergence without estimating probability distributions is more sensible than a naive two-step approach of first estimating probability distributions and then approximating the divergence. Furthermore, despite the overwhelming popularity of the Kullback-Leibler divergence as a divergence measure, we argue that alternatives such as the Pearson divergence, the relative Pearson divergence, and the $L^2$-distance are more useful in practice because of their computationally efficient approximability, high numerical stability, and superior robustness against outliers.

• Communications for Statistical Applications and Methods
• /
• v.15 no.3
• /
• pp.343-351
• /
• 2008

By supervised learning with p predictors, we frequently obtain a prediction function of the form $y\;=\;f(x_1,...,x_p)$. When $p\;{\geq}\;3$, it is not easy to understand the inner structure of f, except for the case the function is formulated as additive. In this study, we propose to use p simple graphs for visual understanding of complex prediction functions produced by several supervised learning engines such as LOESS, neural networks, support vector machines and random forests.

• Journal of Information Processing Systems
• /
• v.13 no.2
• /
• pp.204-214
• /
• 2017

Artificial intelligence, especially deep learning technology, is penetrating the majority of research areas, including the field of bioinformatics. However, deep learning has some limitations, such as the complexity of parameter tuning, architecture design, and so forth. In this study, we analyze these issues and challenges in regards to its applications in bioinformatics, particularly genomic analysis and medical image analytics, and give the corresponding approaches and solutions. Although these solutions are mostly rule of thumb, they can effectively handle the issues connected to training learning machines. As such, we explore the tendency of deep learning technology by examining several directions, such as automation, scalability, individuality, mobility, integration, and intelligence warehousing.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.3
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.

• Journal of Information Technology Application
• /
• v.1
• /
• pp.173-211
• /
• 1999

The task of classification permeates all walks of life, from business and economics to science and public policy. In this context, nonlinear techniques from artificial intelligence have often proven to be more effective than the methods of classical statistics. The objective of knowledge discovery and data mining is to support decision making through the effective use of information. The automated approach to knowledge discovery is especially useful when dealing with large data sets or complex relationships. For many applications, automated software may find subtle patterns which escape the notice of manual analysis, or whose complexity exceeds the cognitive capabilities of humans. This paper explores the utility of a collaborative learning approach involving integrated models in the preprocessing and postprocessing stages. For instance, a genetic algorithm effects feature-weight optimization in a preprocessing module. Moreover, an inductive tree, artificial neural network (ANN), and k-nearest neighbor (kNN) techniques serve as postprocessing modules. More specifically, the postprocessors act as second0order classifiers which determine the best first-order classifier on a case-by-case basis. In addition to the second-order models, a voting scheme is investigated as a simple, but efficient, postprocessing model. The first-order models consist of statistical and machine learning models such as logistic regression (logit), multivariate discriminant analysis (MDA), ANN, and kNN. The genetic algorithm, inductive decision tree, and voting scheme act as kernel modules for collaborative learning. These ideas are explored against the background of a practical application relating to financial fraud management which exemplifies a binary classification problem.

• International Journal of Internet, Broadcasting and Communication
• /
• v.8 no.4
• /
• pp.19-25
• /
• 2016

There are many free applications that need users to sign up before they can use the applications nowadays. It is difficult to choose a suitable password for your account. If the password is too complicated, then it is hard to remember it. However, it is easy to be intruded by other users if we use a very simple password. Therefore, biometric-based approach is one of the solutions to solve the issue. The biometric-based approach includes keystroke dynamics on keyboard, mice, or mobile devices, gait analysis and many more. The approach can integrate with any appropriate machine learning algorithm to learn a user typing behavior for authentication system. Preprocessing phase is one the important role to increase the performance of the algorithm. In this paper, we have proposed ensemble-by-session (EBS) method which to operate the preprocessing phase before the training phase. EBS distributes the dataset into multiple sub-datasets based on the session. In other words, we split the dataset into session by session instead of assemble them all into one dataset. If a session is considered as one day, then the sub-dataset has all the information on the particular day. Each sub-dataset will have different information for different day. The sub-datasets are then trained by a machine learning algorithm. From the experimental result, we have shown the improvement of the performance for each base algorithm after the preprocessing phase.