• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.167-175
• /
• 2021

Semi-supervised learning uses both labeled data and unlabeled data. Recently consistency regularization is very popular in semi-supervised learning. Unsupervised data augmentation (UDA) that uses unlabeled data augmentation is also based on the consistency regularization. The Kullback-Leibler divergence is used for the loss of unlabeled data and cross-entropy for the loss of labeled data through UDA learning. UDA uses techniques such as training signal annealing (TSA) and confidence-based masking to promote performance. In this study, we propose to use Jensen-Shannon divergence instead of Kullback-Leibler divergence, reverse-TSA and not to use confidence-based masking for performance improvement. Through experiment, we show that the proposed technique yields better performance than those of UDA.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.11
• /
• pp.1649-1654
• /
• 2021

In this paper, we introduce a regularization of long short-term memory (LSTM) based fall detection system using TensorFlow that can detect falls that can occur in the elderly. Fall detection uses data from a 3-axis acceleration sensor attached to the body of an elderly person and learns about a total of 7 behavior patterns, each of which is a pattern that occurs in daily life, and the remaining 3 are patterns for falls. During training, a normalization process is performed to effectively reduce the loss function, and the normalization performs a maximum-minimum normalization for data and a L2 regularization for the loss function. The optimal regularization conditions of LSTM using several falling parameters obtained from the 3-axis accelerometer is explained. When normalization and regularization rate λ for sum vector magnitude (SVM) are 127 and 0.00015, respectively, the best sensitivity, specificity, and accuracy are 98.4, 94.8, and 96.9%, respectively.

• Smart Structures and Systems
• /
• v.31 no.3
• /
• pp.229-245
• /
• 2023

The absence of excitation measurements may pose a big challenge in the application of structural damage identification owing to the fact that substantial effort is needed to reconstruct or identify unknown input force. To address this issue, in this paper, an iterative strategy, a synergy of Tikhonov regularization method for force identification and modified Jaya algorithm (M-Jaya) for stiffness parameter identification, is developed for damage identification with partial output-only responses. On the one hand, the probabilistic clustering learning technique and nonlinear updating equation are introduced to improve the performance of standard Jaya algorithm. On the other hand, to deal with the difficulty of selection the appropriate regularization parameters in traditional Tikhonov regularization, an improved L-curve method based on B-spline interpolation function is presented. The applicability and effectiveness of the iterative strategy for simultaneous identification of structural damages and unknown input excitation is validated by numerical simulation on a 21-bar truss structure subjected to ambient excitation under noise free and contaminated measurements cases, as well as a series of experimental tests on a five-floor steel frame structure excited by sinusoidal force. The results from these numerical and experimental studies demonstrate that the proposed identification strategy can accurately and effectively identify damage locations and extents without the requirement of force measurements. The proposed M-Jaya algorithm provides more satisfactory performance than genetic algorithm, Gaussian bare-bones artificial bee colony and Jaya algorithm.

• Korea-Australia Rheology Journal
• /
• v.20 no.2
• /
• pp.51-58
• /
• 2008

This paper describes a two-step Tikhonov regularization procedure for converting the steady shear data generated by parallel-disk viscometers, in the presence of wall slip, into a shear stress-shear rate function and a wall shear stress-slip velocity functions. If the material under test has a yield stress or a critical wall shear stress below which no slip is observed the method will also provide an estimate of these stresses. Amplification of measurement noise is kept under control by the introduction of two separate regularization parameters and Generalized Cross Validation is used to guide the selection of these parameters. The performance of this procedure is demonstrated by applying it to the parallel disk data of an oil-in-water emulsion, of a foam and of a mayonnaise.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2007.11a
• /
• pp.446-451
• /
• 2007

An Improved finite element model updating method to address the numerical difficulty associated with ill-conditioning and rank-deficiency. These difficulties frequently occur in model updating problems, when the identification of a larger number of physical parameters is attempted than that warranted by the information content of the experimental data. Based on the standard Bounded Variables Least-squares (BVLS) method, which incorporates the usual upper/lower-bound constraints, the proposed method is equipped with new constraints based on the correlation coefficients between the sensitivity vectors of updating parameters. The effectiveness of the proposed method is investigated through the numerical simulation of a simple framed structure by comparing the results of the proposed method with those obtained via pure BVLS and the regularization method. The comparison indicated that the proposed method and the regularization method yield approximate solutions with similar accuracy.

• Smart Structures and Systems
• /
• v.10 no.4_5
• /
• pp.459-470
• /
• 2012

This paper summarizes a study for the modal analysis and model updating conducted using the monitoring data obtained from the Canton Tower of 610 m tall, which was established as an international benchmark problem by the Hong Kong Polytechnic University. Modal properties of the tower were successfully identified using frequency domain decomposition and stochastic subspace identification methods. Finite element model updating using the measurement data was further performed to reduce the modal property differences between the measurements and those of the finite element model. Over-fitting during the model updating was avoided by using an optimization scheme with a regularization term.

• Journal of the Korean Mathematical Society
• /
• v.48 no.2
• /
• pp.311-327
• /
• 2011

This paper is devoted to simplified Tikhonov regularization for two kinds of parabolic equations, i.e., a sideways parabolic equation, and a two-dimensional inverse heat conduction problem. The measured data are assumed to be known approximately. We concentrate on the convergence rates of the simplified Tikhonov approximation of u(x, t) and its derivative $u_x$(x, t) of sideways parabolic equations at 0 $\leq$ x < 1, and that of two-dimensional inverse heat conduction problem at 0 < x $\leq$ 1, respectively.

• Journal of the Chungcheong Mathematical Society
• /
• v.29 no.1
• /
• pp.59-71
• /
• 2016

To obtain more accurate approximation of the true images in the deblurring problems, the weighted global generalized cross validation(GCV) function to the inverse problem with multiple right-hand sides is suggested as an efficient way to determine the regularization parameter. We analyze the experimental results for many test problems and was able to obtain the globally useful range of the weight when the preconditioned global conjugate gradient linear least squares(Gl-CGLS) method with the weighted global GCV function is applied.

• International journal of advanced smart convergence
• /
• v.9 no.2
• /
• pp.173-178
• /
• 2020

A common problem with neural network learning is that it is too suitable for the specificity of learning. In this paper, various methods were compared to avoid overfitting: regularization, drop-out, different numbers of data and different types of neural networks. Comparative studies of the above-mentioned methods have been provided to evaluate the test accuracy. I found that the more data using method is better than the regularization and dropout methods. Moreover, we know that deep convolutional neural networks outperform multi-layer neural networks and simple convolution neural networks.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.6 no.2
• /
• pp.9-24
• /
• 2002

Total Variation(TV) regularization method is effective for reconstructing "blocky", discontinuous images from contaminated image with noise. But TV is represented by highly nonlinear integro-differential equation that is hard to solve. There have been much effort to obtain stable and fast methods. C. Vogel introduced "the Fixed Point Lagged Diffusivity Iteration", which solves the nonlinear equation by linearizing. In this paper, we apply multigrid(MG) method for cell centered finite difference (CCFD) to solve system arise at each step of this fixed point iteration. In numerical simulation, we test various images varying noises and regularization parameter $\alpha$ and smoothness $\beta$ which appear in TV method. Numerical tests show that the parameter ${\beta}$ does not affect the solution if it is sufficiently small. We compute optimal $\alpha$ that minimizes the error with respect to $L^2$ norm and $H^1$ norm and compare reconstructed images.