• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.29 no.1
• /
• pp.117-125
• /
• 2019

Deep Learning, which is being used in various fields recently, is being threatened with Adversarial Attack. In this paper, we experimentally verify that the classification accuracy is lowered by adversarial samples generated by malicious attackers in image classification models. We used MNIST dataset and measured the detection accuracy by injecting adversarial samples into the Autoencoder classification model and the CNN (Convolution neural network) classification model, which are created using the Tensorflow library and the Pytorch library. Adversarial samples were generated by transforming MNIST test dataset with JSMA(Jacobian-based Saliency Map Attack) and FGSM(Fast Gradient Sign Method). When injected into the classification model, detection accuracy decreased by at least 21.82% up to 39.08%.

• The Journal of Korea Robotics Society
• /
• v.17 no.3
• /
• pp.365-372
• /
• 2022

The accurate object recognition is important for the precise and accurate manipulation. To enhance the recognition performance, we can use various types of sensors. In general, acquired data from sensors have a high sampling rate. So, in the past, the RNN-based model is commonly used to handle and analyze the time-series sensor data. However, the RNN-based model has limitations of excessive parameters. CNN-based model also can be used to analyze time-series input data. However, CNN-based model also has limitations of the small receptive field in early layers. For this reason, when we use a CNN-based model, model architecture should be deeper and heavier to extract useful global features. Thus, traditional methods like RN N -based and CN N -based model needs huge amount of learning parameters. Recently studied result shows that Fast Fourier Convolution (FFC) can overcome the limitations of traditional methods. This operator can extract global features from the first hidden layer, so it can be effectively used for feature extracting of sensor data that have a high sampling rate. In this paper, we propose the algorithm to recognize objects using tactile sensor data and the FFC model. The data was acquired from 11 types of objects to verify our posed model. We collected pressure, current, position data when the gripper grasps the objects by random force. As a result, the accuracy is enhanced from 84.66% to 91.43% when we use the proposed FFC-based model instead of the traditional model.

• Nuclear Engineering and Technology
• /
• v.55 no.11
• /
• pp.3925-3934
• /
• 2023

The importance of fast neutron detection for nuclear safeguards purposes has increased due to its potential advantages such as reasonable cost and higher precision for larger sample masses of nuclear materials. Pulse-shape discrimination (PSD) is inevitably used to discriminate neutron- and gamma-ray- induced signals from organic scintillators of very high gamma sensitivity. The light output (LO) threshold corresponding to several MeV of recoiled proton energy could be necessary to achieve fine PSD performance. However, this leads to neutron count losses and possible distortion of results obtained by neutron multiplicity counting (NMC)-based nuclear material accountancy (NMA). Moreover, conventional PSD techniques are not effective for counting of neutrons in a high-gamma-ray environment, even under a sufficiently high LO threshold. In the present work, PSD performance (figure-of-merit, FOM) according to LO bands was confirmed using a conventional charge comparison method (CCM) and compared with results obtained by convolution neural network (CNN)-based PSD algorithms. Also, it was attempted, for the first time ever, to reject fake neutron signals from distorted PSD regions where neutron-induced signals are normally detected. The overall results indicated that higher neutron detection efficiency with better accuracy could be achieved via CNN-based PSD algorithms.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.6
• /
• pp.63-74
• /
• 2007

Fast Fourier transform(FFT) is powerful, fast computation framework for convolution in many image restoration application. However, an actually observed image acquired with finite aperture of the acquisition device from the infinite background and it lost data outside the cropped region. Because of these the boundary artifacts are produced. This paper reviewed and summarized the up to date the techniques that have been applied to reduce of the boundary artifacts. Moreover, we propose a new block-based fast image restoration using combined extrapolation and edge-tapering without boundary artifacts with reduced computational loads. We apply edgetapering to the inner blocks because they contain outside information of boundary. And outer blocks use half-convolution extrapolation. For this process it is possible that fast image restoration without boundary artifacts.

• Journal of Broadcast Engineering
• /
• v.14 no.5
• /
• pp.543-550
• /
• 2009

The most widely used block convolution method is the overlap save algorithm (OSA), where a block of M data to be convolved with a filter is concatenated with the previous block and 2M-point FFT and multiplications are performed for this overlapped block. By discarding half of the results, we obtain linear convolution results from the circular convolution. This paper proposes a new transform which reduces the block size to only M for the block convolution. The proposed transform can be implemented as the M multiplications followed by M-point FFT Hence, existing efficient FFT libraries and hardware can be exploited for the implementation of proposed method. Since the required transform size is half that of the conventional method, the overall computational complexity is reduced. Also the reduced transform size results in the reduction of data access time and cash miss-hit ratio, and thus the overall CPU time is reduced. Experiments show that the proposed method requires less computation time than the conventional OSA.

• The KIPS Transactions:PartB
• /
• v.13B no.1 s.104
• /
• pp.27-34
• /
• 2006

In this study, we propose this new algorithm that generates score function in ICA(Independent Component Analysis) using entropy theory. To generate score function, estimation of probability density function about original signals are certainly necessary and density function should be differentiated. Therefore, we used kernel density estimation method in order to derive differential equation of score function by original signal. After changing formula to convolution form to increase speed of density estimation, we used FFT algorithm that can calculate convolution faster. Proposed score function generation method reduces the errors, it is density difference of recovered signals and originals signals. In the result of computer simulation, we estimate density function more similar to original signals compared with Extended Infomax and Fixed Point ICA in blind source separation problem and get improved performance at the SNR(Signal to Noise Ratio) between recovered signals and original signal.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.12
• /
• pp.63-73
• /
• 2009

This paper presents a vehicle detection method that uses convolution matching method based on a simple color information. An input image is expressed as 8 oriented color expression(Red, Green, Blue, White, Black, Cyan, Yellow, Magenta) considering an orientation of a pixel color vector. It makes the image very reliable and strong against changes of illumination condition or environment. This paper divides the vehicle detection into a hypothesis generation step and a hypothesis verification step. In the hypothesis generation step, the vehicle candidate region is found by vertical edge and shadow. In the hypothesis verification step, the convolution matching and the complexity of image edge are used to detect real vehicles. It is proved that the proposed method has the fast and high detection rate on various experiments where the illumination source and environment are changed.

• Proceedings of the IEEK Conference
• /
• 2004.08c
• /
• pp.897-899
• /
• 2004

There is scientific work containing the recurrence method of synthesis the new class of orthogonal circle m-invariant signals: designed effective algorithms of fast-direct computing m-convolution in time domain: engineer methods of design economic scheme of decoders for optimal receiving in aggregate of suggested signal.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.7 no.3
• /
• pp.265-276
• /
• 1995

In this Paper, the Performance of the convolution method has been investigated as an effort to develop a simple system of predicting wind-driven surface current on a real time basis. In this approach wind stress is assumed to be spatially uniform and the effect of atmospheric pressure is neglected. The discrete convolution weights are determined in advance at each point using a linear three-dimensional Galerkin model with linear shape functions(Galerkin-FEM model). Four directions of wind stress(e.g. NE, SW, NW, SE) with unit magnitude are imposed in the model calculation for the construction of data base for convolution weights. Given the time history of wind stress, it is then possible to predict with-driven currents promptly using the convolution product of finite length. An unsteady wind stress of arbitrary form can be approximated by a series of wind pulses with magnitude of 6 hour averaged value. A total of 12 pulses are involved in the convolution product To examine the accuracy of the convolution method a series of numerical experiments has been carried out in the idealized basin representing the scale of the Yellow Sea and the East China Sea. The wind stress imposed varies sinusoidally in time. It was found that the predicted surface currents and elevation fields were in good agreement with the results computed by the direct integration of the Galerkin model. A model with grid 1/8$^{\circ}$ in latitude, l/6$^{\circ}$ in longitude was established which covers the entire region of the Yellow Sea and the East China Sea. The numerical prediction in terms of the convolution product has been carried out with particular attention on the formation of upwind flow in the middle of the Yellow Sea by northerly wind.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.1
• /
• pp.115-126
• /
• 2023

Due to the fluctuating random and periodical nature of renewable energy generation power quality disturbances occurred more frequently in power generation transformation transmission and distribution. Various power quality disturbances may lead to equipment damage or even power outages. Therefore it is essential to detect and classify different power quality disturbances in real time automatically. The traditional PQD identification method consists of three steps: feature extraction feature selection and classification. However, the handcrafted features are imprecise in the feature selection stage, resulting in low classification accuracy. This paper proposes a deep neural architecture based on Convolution Neural Network and Long Short Term Memory combining the time and frequency domain features to recognize 16 types of Power Quality signals. The frequency-domain data were obtained from the Fast Fourier Transform which could efficiently extract the frequency-domain features. The performance in synthetic data and real 6kV power system data indicate that our proposed method generalizes well compared with other deep learning methods.