• ETRI Journal
• /
• 제39권6호
• /
• pp.832-840
• /
• 2017

Recently, deep recurrent neural networks have achieved great success in various machine learning tasks, and have also been applied for sound event detection. The detection of temporally overlapping sound events in realistic environments is much more challenging than in monophonic detection problems. In this paper, we present an approach to improve the accuracy of polyphonic sound event detection in multichannel audio based on gated recurrent neural networks in combination with auditory spectral features. In the proposed method, human hearing perception-based spatial and spectral-domain noise-reduced harmonic features are extracted from multichannel audio and used as high-resolution spectral inputs to train gated recurrent neural networks. This provides a fast and stable convergence rate compared to long short-term memory recurrent neural networks. Our evaluation reveals that the proposed method outperforms the conventional approaches.

• 한국정밀공학회지
• /
• 제32권11호
• /
• pp.989-995
• /
• 2015

In the literature, various stochastic anomaly detection methods, such as limit checking and PCA-based approaches, have been applied to weld defect detection. However, it is still a challenge to identify meaningful defect patterns from very limited sensor signals of laser welding, characterized by intermittent, discontinuous, very short, and non-stationary random signals. In order to effectively analyze the physical characteristics of laser weld signals: plasma intensity, weld pool temperature, and back reflection, we first transform the raw data of laser weld signals into the form of event logs. This is done by multidimensional discretization and event-codification, after which the event logs are decoded to extract weld defect patterns by $Na{\ddot{i}}ve$ Bayes classifier. The performance of the proposed method is examined in comparison with the commercial solution of PRECITEC's LWM$^{TM}$ and the most recent PCA-based detection method. The results show higher performance of the proposed method in terms of sensitivity (1.00) and specificity (0.98).

• 한국통신학회논문지
• /
• 제41권8호
• /
• pp.914-916
• /
• 2016

The noise in a power signal degrades the detection rate of the power quality (PQ) event signals. We present a new wavelet de-noising technique for PQ event detection that employs the correlation-based thresholding instead of the wavelet-scale-based thresholding of existing schemes. The simulation results show that the proposed scheme is more robust to Gaussian and impulsive noisy conditions and has further improved detection ratio than existing schemes.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권9호
• /
• pp.4355-4374
• /
• 2018

Crowd escape event detection has become one of the hottest problems in intelligent surveillance filed. When the 'escape event' occurs, pedestrians will escape in a disordered way with different velocities and directions. Based on these characteristics, this paper proposes a Direction-Collectiveness Model to detect escape event in crowd scenes. First, we extract a set of trajectories from video sequences by using generalized Kanade-Lucas-Tomasi key point tracker (gKLT). Second, a Direction-Collectiveness Model is built based on the randomness of velocity and orientation calculated from the trajectories to express the movement of the crowd. This model can describe the movement of the crowd adequately. To obtain a generalized crowd escape event detector, we adopt an adaptive threshold according to the Direction-Collectiveness index. Experiments conducted on two widely used datasets demonstrate that the proposed model can detect the escape events more effectively from dense crowd.

• 한국통신학회논문지
• /
• 제38B권7호
• /
• pp.589-595
• /
• 2013

최근 보안 관제를 위한 인원부족 및 감시 능력의 한계로 자동화된 지능형 관제 시스템에 대한 요구가 증가하고 있다. 이 논문에서는 지능형 감시시스템의 구축을 위하여 자동화된 객체추출 및 추적 시스템, 그리고 이상행위를 인지하는 이상행위 검출 시스템을 설계하고 구현하였다. 각 모듈은 기존의 연구 결과를 바탕으로 실제 환경에서 적용되고 상용화가 가능하도록 알고리즘의 성능을 높였으며, 구현 후 다양한 테스트를 통해 그 성과를 검증하였다. 특히, 배회 또는 도주와 같은 이상행위의 경우 1초 이내에 검출할 수 있었다.

• International journal of advanced smart convergence
• /
• 제9권2호
• /
• pp.20-27
• /
• 2020

Sound event detection in real-world environments suffers from the interference of non-stationary and time-varying noise. This paper presents an adaptive noise reduction method for sound event detection based on non-negative matrix factorization (NMF). In this paper, we proposed a deep learning model that integrates Convolution Neural Network (CNN) with Non-Negative Matrix Factorization (NMF). To improve the separation quality of the NMF, it includes noise update technique that learns and adapts the characteristics of the current noise in real time. The noise update technique analyzes the sparsity and activity of the noise bias at the present time and decides the update training based on the noise candidate group obtained every frame in the previous noise reduction stage. Noise bias ranks selected as candidates for update training are updated in real time with discrimination NMF training. This NMF was applied to CNN and Hidden Markov Model(HMM) to achieve improvement for performance of sound event detection. Since CNN has a more obvious performance improvement effect, it can be widely used in sound source based CNN algorithm.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 1998년도 추계종합학술대회 논문집
• /
• pp.875-878
• /
• 1998

In this paper, an adaptive event detection algorithm is proposed, for which we use the statistics of subblock image and adaptive threshold levels. The adaptive threshold level for a parameter binarization is taken by averaging the corresponding paramerter obtained from several input images. As simulation results, it is shown that the proposed algorithm is much more adaptive to the input images and effective in event detection rate than the conventional difference based algorithms.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제55권5호
• /
• pp.248-253
• /
• 2006

The purpose of this study is to develop a portable gait-event detection system which is necessary for the cycle-to-cycle FES(functional electrical stimulation) control of locomotion. To make the system portable, we made following modifications in the gait signal measurement system. That is, 1) to make the system wireless using Bluetooth communication, 2) to make the system small-sized and battery-powered by using low power consumption ${\mu}$ P(ATmega8535L). The gait-events were analyzed in off-line at the main computer using ANN(Artificial Neural Network). The Proposed system showed no mis-detection of the gait-events of normal subject and hemiplegia subjects. The performance of the system was better than the previous wired-system.

• International Journal of Computer Science & Network Security
• /
• 제23권12호
• /
• pp.107-114
• /
• 2023

The attack technique by the malware distribution form is a dangerous, difficult to detect and prevent attack method. Current malware detection studies and proposals are often based on two main methods: using sign sets and analyzing abnormal behaviors using machine learning or deep learning techniques. This paper will propose a method to detect malware on Endpoints based on Event IDs using deep learning. Event IDs are behaviors of malware tracked and collected on Endpoints' operating system kernel. The malware detection proposal based on Event IDs is a new research approach that has not been studied and proposed much. To achieve this purpose, this paper proposes to combine different data mining methods and deep learning algorithms. The data mining process is presented in detail in section 2 of the paper.

• 한국음향학회지
• /
• 제36권4호
• /
• pp.267-272
• /
• 2017

본 논문에서는 다채널 오디오 특징값을 게이트형 순환 신경망(Gated Recurrent Neural Networks, GRNN)에 적용한 효과적인 다성 사운드 이벤트 검출 방식을 제안한다. 실생활의 사운드는 여러 사운드 이벤트가 겹쳐있는 다성사운드로, 기존의 단일 채널 오디오 특징값으로는 다성 사운드에서 개별적인 이벤트의 검출이 어렵다는 한계가 있다. 이에 본 논문에서는 다채널 오디오 신호를 기반으로 추출된 특징값을 사용하여 다성 사운드 이벤트 검출에 적용하였다. 또한 본 논문에서는 현재 순환 신경망에서 가장 높은 성능을 보이는 장단기 기억 신경망(Long Short Term Memory, LSTM) 보다 간단한 GRNN을 분류에 적용하여 다성 사운드 이벤트 검출의 성능을 더욱 향상시키고자 하였다. 실험결과는 본 논문에서 제안한 방식이 기존의 방식보다 성능이 더 뛰어나다는 것을 보인다.