• The KIPS Transactions:PartD
• /
• v.15D no.1
• /
• pp.31-40
• /
• 2008

We generally use lower-dimensional transformations to convert high-dimensional sequences into low-dimensional points in similar sequence matching. These traditional transformations, however, show different characteristics in indexing performance by the type of time-series data. It means that the selection of lower-dimensional transformations makes a significant influence on the indexing performance in similar sequence matching. To solve this problem, in this paper we propose a hybrid approach that integrates multiple transformations and uses them in a single multidimensional index. We first propose a new notion of hybrid lower-dimensional transformation that exploits different lower-dimensional transformations for a sequence. We next define the hybrid distance to compute the distance between the transformed sequences. We then formally prove that the hybrid approach performs the similar sequence matching correctly. We also present the index building and the similar sequence matching algorithms that use the hybrid approach. Experimental results for various time-series data sets show that our hybrid approach outperforms the single transformation-based approach. These results indicate that the hybrid approach can be widely used for various time-series data with different characteristics.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.13 no.5
• /
• pp.313-322
• /
• 2020

A large volume of data will improve the robustness of deep learning models and avoid overfitting problems. In automatic tongue segmentation, the availability of annotated tongue images is often limited because of the difficulty of collecting and labeling the tongue image datasets in reality. Data augmentation can expand the training dataset and increase the diversity of training data by using label-preserving transformations without collecting new data. In this paper, augmented tongue image datasets were developed using seven augmentation techniques such as image cropping, rotation, flipping, color transformations. Performance of the data augmentation techniques were studied using state-of-the-art transfer learning models, for instance, InceptionV3, EfficientNet, ResNet, DenseNet and etc. Our results show that geometric transformations can lead to more performance gains than color transformations and the segmentation accuracy can be increased by 5% to 20% compared with no augmentation. Furthermore, a random linear combination of geometric and color transformations augmentation dataset gives the superior segmentation performance than all other datasets and results in a better accuracy of 94.98% with InceptionV3 models.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.1762-1765
• /
• 1997

Some ill-conditioned processes are very sensitive to small element-wise uncertainties arising in classical element-by-element model identifications. For such processes, accurate identification of simgular values and right singular vectors are more important than theose of the elements themselves. Singular values and right singular vectors can be found by iteraive identification methods which implement the input and output transformations iteratively. Methods based on SVD decomposition, QR decomposition and LU decomposition are proposed and compared with the Kuong and Mac Gregor's method. Convergence proofs are given. These SVD and QR mehtods use normal matrices for the transformations which cannot be calculated analytically in general and so they are hoard to apply to dynamic processes, whereas the LU method used simple analyitc transformations and can be directly applied to dynamic processes.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.2 no.2
• /
• pp.115-121
• /
• 2002

This paper presents an evolutionary design of digital IIR filters using the genetic algorithm (GA) with modified genetic operators and real-valued encoding. Conventional digital IIR filter design methods involve algebraic transformations of the transfer function of an analog low-pass filter (LPF) that satisfies prescribed filter specifications. Other types of frequency-selective digital fillers as high-pass (HPF), band-pass (BPF), and band-stop (BSF) filters are obtained by appropriate transformations of a prototype low-pass filter. In the GA-based digital IIR filter design scheme, filter coefficients are represented as a set of real-valued genes in a chromosome. Each chromosome represents the structure and weights of an individual filter. GA directly finds the coefficients of the desired filter transfer function through genetic search fur given filter specifications of minimum filter order. Crossover and mutation operators are selected to ensure the stability of resulting IIR filters. Other types of filters can be found independently from the filter specifications, not from algebraic transformations.

• Journal of the Korean Mathematical Society
• /
• v.48 no.2
• /
• pp.289-300
• /
• 2011

Let T(X) denote the semigroup (under composition) of transformations from X into itself. For a fixed nonempty subset Y of X, let S(X, Y) = {${\alpha}\;{\in}\;T(X)\;:\;Y\;{\alpha}\;{\subseteq}\;Y$}. Then S(X, Y) is a semigroup of total transformations of X which leave a subset Y of X invariant. In this paper, we characterize when S(X, Y) is isomorphic to T(Z) for some set Z and prove that every semigroup A can be embedded in S($A^1$, A). Then we describe Green's relations for S(X, Y) and apply these results to obtain its group H-classes and ideals.

• Communications of the Korean Mathematical Society
• /
• v.20 no.4
• /
• pp.723-729
• /
• 2005

Let F be a non-trivial non-Archimedian field. The sequence spaces $\Gamma\;(F)\;and\;{\Gamma}^{\ast}(F)$ were defined and studied by Soma-sundaram[4], where these spaces denote the spaces of entire and analytic sequences defined over F, respectively. In 1997, these spaces were generalized by Mursaleen and Qamaruddin[1] by considering an arbitrary sequence $U\;=\;(U_k),\;U_k\;{\neq}\;0 \;(\;k\;=\;1,2,3,{\cdots})$. They characterized some classes of infinite matrices considering these new classes of sequences. In this paper, we further generalize the above mentioned spaces and define the spaces $\Gamma(u,\;F,\;{\Delta}),\;{\Gamma}^{\ast}(u,\;F,\;{\Delta}),\;l_p(u,\;F,\;{\Delta})$), and $b_v(u,\;F,\;{\Delta}$). We also study some matrix transformations on these new spaces.

• Proceedings of the Korean Society for Technology of Plasticity Conference
• /
• 2006.05a
• /
• pp.77-80
• /
• 2006

Structural phase transformations of silicon during nanoindentation were investigated in detail at the atomic level. The molecular dynamics simulations of nanoindentation on the (100), (110) and (111) surface of single crystalline silicon were simulated, and this supported the theoretical prediction of the anisotropic behavior of structural phase transformations. Simulations showed that microscopic aspects of phase transformation varied according to the crystallographic orientation of the contact surface and were directly linked to the slip system.

• Journal of Mechanical Science and Technology
• /
• v.16 no.5
• /
• pp.703-709
• /
• 2002

In the present paper, approximate coordinate transformations for simulation of turbulent flows with wall deformation, significantly reducing computational cost with little degradation in numerical accuracy, are presented. The Wavier-Stokes equations are coordinate-transformed with an approximation of Tailor-series truncation. The performance is evaluated by performing numerical simulations of a channel flow at Re$\sub$$\tau$/ = 140 with active wall motions of η$\sub$m/$\^$+/ $\leq$5. The approximate transformations provide flow structures as well as turbulence statistics in good agreement with those from a complete transformation [Phys. Fluids 12, 3301 (2000) ] and allow 25-30% savings in the CPU time as compared to the complete one.

• Journal of the Chungcheong Mathematical Society
• /
• v.28 no.2
• /
• pp.331-343
• /
• 2015

Invertible transformations over n-bit words are essential ingredients in many cryptographic constructions. When n is large such invertible transformations are usually represented as a composition of simpler operations such as linear functions, S-P networks, Feistel structures and T-functions. Among them we study T-functions which are probably invertible transformations and are very useful in stream ciphers. In this paper we study the number of single cycle T-functions satisfying some conditions and characterize single cycle T-functions on $(\mathbb{Z}_2)^n$ generated by some elements in $(\mathbb{Z}_2)^{n-1}$.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.34S no.8
• /
• pp.74-83
• /
• 1997

In this paper, we propose a new fractal image decoding algorithm with fast convergence speed by using the data dependence and the improved initial image estimation. Conventional method for fractal image decoding requires high-degrdd computational complexity in decoding process, because of iterated contractive transformations applied to whole range blocks. On proposed method, Range of reconstruction imagte is divided into referenced range and data dependence region. And computational complexity is reduced by application of iterated contractive transformations for the referenced range only. Data dependence region can be decoded by one transformations when the referenced range is converged. In addition, more exact initial image is estimated by using bound () function in case of all, and an initial image more nearer to a fixed point is estimated by using range block division estimation. Consequently, the convergence speed of reconstruction iamge is improved with 40% reduction of computational complexity.