• Proceedings of the KSRS Conference
• /
• 2008.10a
• /
• pp.406-409
• /
• 2008

Clustering data streams has an importance over many applications like sensor networks. Existing hierarchical methods follow a semi fuzzy clustering that yields duplicate clusters. In order to solve the problems, we propose an extended online divisive agglomerative clustering on data streams. It builds a tree-like top-down hierarchy of clusters that evolves with data streams using geometric time frame for snapshots. It is an enhancement of the Online Divisive Agglomerative Clustering (ODAC) with a pruning strategy to avoid duplicate clusters. Our main features are providing update time and memory space which is independent of the number of examples on data streams. It can be utilized for clustering sensor data and network monitoring as well as web click streams.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.1226-1228
• /
• 2022

FHD 이상을 넘어선 UHD급의 고해상도 동영상 콘텐츠의 수요 및 공급이 증가함에 따라 전반적인 산업 영역에서 네트워크 자원을 효율적으로 이용하여 동영상 콘텐츠를 제공하는 데에 관심을 두게 되었다. 기존 방법을 통한 bi-cubic, bi-linear interpolation 등의 방법은 딥 러닝 기반의 모델에 비교적 인풋 이미지의 특징을 잘 잡아내지 못하는 결과를 나타내었다. 딥 러닝 기반의 초 해상화 기술의 경우 기존 방법과 비교 시 연산을 위해 더 많은 자원을 필요로 하므로, 이러한 사용 조건에 따라 본 논문은 초 해상화가 가능한 딥 러닝 모델을 경량화 기법을 사용하여 기존에 사용된 모델보다 비교적 적은 자원을 효율적으로 사용할 수 있도록 연구 개발하는 데 목적을 두었다. 연구방법으로는 structure pruning을 이용하여 모델 자체의 구조를 경량화 하였고, 학습을 진행해야 하는 파라미터를 줄여 하드웨어 자원을 줄이는 연구를 진행했다. 또한, Residual Network의 개수를 줄여가며 PSNR, LPIPS, tOF등의 결과를 비교했다.

• The Transactions of the Korea Information Processing Society
• /
• v.3 no.3
• /
• pp.439-448
• /
• 1996

Extraction and selection of the informative features play a central role in pattern recognition. This paper describes a modified back-propagation algorithm that performs selection of the informative features and trains a neural network simultaneously. The algorithm is mainly composed of three repetitive steps : training, connection pruning, and input unit elimination. Afer initial training, the connections that have small magnitude are first pruned. Any unit that has a small number of connections to the hidden units is deleted,which is equivalent to excluding the feature corresponding to that unit.If the error increases,the network is retraned,again followed by connection pruning and input unit elimination.As a result,the algorithm selects the most im-portant features in the measurement space without a transformation to another space.Also,the selected features are the most-informative ones for the classification,because feature selection is tightly coupled with the classifi-cation performance.This algorithm helps avoid measurement of redundant or less informative features,which may be expensive.Furthermore,the final network does not include redundant parameters,i.e.,weights and biases,that may cause degradation of classification performance.In applications,the algorithm preserves the most informative features and significantly reduces the dimension of the feature vectors whiout performance degradation.

• Journal of Korea Society of Industrial Information Systems
• /
• v.22 no.5
• /
• pp.17-24
• /
• 2017

Recently Deep-learning, Technologies using Large or Deep Artificial Neural Networks, have Shown Remarkable Performance, and the Increasing Size of the Network Contributes to its Performance Improvement. However, the Increase in the Size of the Neural Network Leads to an Increase in the Calculation Amount, which Causes Problems Such as Circuit Complexity, Price, Heat Generation, and Real-time Restriction. In This Paper, We Propose and Test a Method to Reduce the Number of Network Connections by Effectively Pruning the Redundancy in the Connection and Showing the Difference between the Performance and the Desired Range of the Original Neural Network. In Particular, we Proposed a Simple Method to Improve the Performance by Re-learning and to Guarantee the Desired Performance by Allocating the Error Rate per Layer in Order to Consider the Difference of each Layer. Experiments have been Performed on a Typical Neural Network Structure such as FCN (full connection network) and CNN (convolution neural network) Structure and Confirmed that the Performance Similar to that of the Original Neural Network can be Obtained by Only about 1/10 Connection.

• Journal of Electrical Engineering and Technology
• /
• v.13 no.5
• /
• pp.1814-1820
• /
• 2018

State estimation of power systems has become vital in recent days of power operation and control. SCADA and EMS are intended for the state estimation and to communicate and monitor the systems which are operated at specified time. Although various methods are used we can achieve the better results by using PMU technique. On placing the PMU, operating time is reduced and making the performance reliable. In this paper, PMU placement is done in two ways. Those are 'optimal technique with pruning operation' and 'depth of unobservability' considering incomplete and complete observability of a network. By Depth of Unobservability Number of PMUs are reduced to attain Observability of the network. Proposed methods are tested on IEEE 14, 30, 57, SR-system and Sub systems (1, 2) with bus size of 270 and 444 buses. Along with achieving complete observability analysis, single PMU loss condition is also achieved.

• Journal of Information Processing Systems
• /
• v.7 no.2
• /
• pp.241-260
• /
• 2011

In earlier days, most of the data carried on communication networks was textual data requiring limited bandwidth. With the rise of multimedia and network technologies, the bandwidth requirements of data have increased considerably. If a network link at any time is not able to meet the minimum bandwidth requirement of data, data transmission at that path becomes difficult, which leads to network congestion. This causes delay in data transmission and might also lead to packet drops in the network. The retransmission of these lost packets would aggravate the situation and jam the network. In this paper, we aim at providing a solution to the problem of network congestion in mobile ad hoc networks [1, 2] by designing a protocol that performs routing intelligently and minimizes the delay in data transmission. Our Objective is to move the traffic away from the shortest path obtained by a suitable shortest path calculation algorithm to a less congested path so as to minimize the number of packet drops during data transmission and to avoid unnecessary delay. For this we have proposed a protocol named as Congestion Aware Selection Of Path With Efficient Routing (CASPER). Here, a router runs the shortest path algorithm after pruning those links that violate a given set of constraints. The proposed protocol has been compared with two link state protocols namely, OSPF [3, 4] and OLSR [5, 6, 7, 8].The results achieved show that our protocol performs better in terms of network throughput and transmission delay in case of bulky data transmission.

• Journal of the Korea Computer Graphics Society
• /
• v.21 no.3
• /
• pp.17-26
• /
• 2015

In this paper, we explain a volume rendering system for client-server environment. A single GPU-equipped PC works as a server which is based on the ideas that only a few concurrent users use a volume rendering system in a small hospital. As the clients, we used Android mobile devices such as smart phones. User events are transformed to rendering requests by the client application. When the server receives a rendering request, it renders the volume using the GPU. The rendered image is compressed to JPEG or PNG format so that we can save network bandwidth and reduce transfer time. In addition, we perform an event pruning method while a user is dragging the touch to enhance latency. The server compensates the pruning by interpolating the touch positions. As the result, real-time volume rendering is possible for 5 concurrent users on single GPU-equipped commodity hardware.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2010.10a
• /
• pp.434-437
• /
• 2010

NIPS(Network Intrusion Prevention System) is in line at the end of the external and internal networks which performed two kinds of action: Signature-based filtering and anomaly detection and prevention-based on self-learning. Among them, a signature-based filtering is well known to defend against attacks. By using signature-based filtering, intrusion prevention system passing a payload of packets is compared with attack patterns which are signature. If match, the packet is discard. However, when there is packet delay, it will increase the required pattern matching time as the number of signature is increasing whenever there is delay occur. Therefore, to ensure the performance of IPS, we needed more efficient pattern matching algorithm for high-performance ISP. To improve the performance of pattern matching the most important part is to reduce the number of comparisons signature rules and the packet whenever the packets arrive. In this paper, we propose an improve signature hashing-based pattern matching method. We use tuple pruning algorithm with Bloom filters, which effectively remove unnecessary tuples. Unlike other existing signature hashing-based IPS, our proposed method to improve the performance of IPS.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.6
• /
• pp.850-858
• /
• 2022

To deploy Gate Recurrent Units (GRU) on resource-constrained embedded devices, this paper presents a reconfigurable FPGA-based GRU accelerator that enables structured compression. Firstly, a dense GRU model is significantly reduced in size by hybrid quantization and structured top-k pruning. Secondly, the energy consumption on external memory access is greatly reduced by the proposed reuse computing pattern. Finally, the accelerator can handle a structured sparse model that benefits from the algorithm-hardware co-design workflows. Moreover, inference tasks can be flexibly performed using all functional dimensions, sequence length, and number of layers. Implemented on the Intel DE1-SoC FPGA, the proposed accelerator achieves 45.01 GOPs in a structured sparse GRU network without batching. Compared to the implementation of CPU and GPU, low-cost FPGA accelerator achieves 57 and 30x improvements in latency, 300 and 23.44x improvements in energy efficiency, respectively. Thus, the proposed accelerator is utilized as an early study of real-time embedded applications, demonstrating the potential for further development in the future.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.12
• /
• pp.3904-3922
• /
• 2022

As a research hotspot, pedestrian detection has a wide range of applications in the field of computer vision in recent years. However, current pedestrian detection methods have problems such as insufficient detection accuracy and large models that are not suitable for large-scale deployment. In view of these problems mentioned above, a lightweight pedestrian detection and early warning method using a new model called you only look once (Yolov5) is proposed in this paper, which utilizing advantages of Yolov5s model to achieve accurate and fast pedestrian recognition. In addition, this paper also optimizes the loss function of the batch normalization (BN) layer. After sparsification, pruning and fine-tuning, got a lot of optimization, the size of the model on the edge of the computing power is lower equipment can be deployed. Finally, from the experimental data presented in this paper, under the training of the road pedestrian dataset that we collected and processed independently, the Yolov5s model has certain advantages in terms of precision and other indicators compared with traditional single shot multiBox detector (SSD) model and fast region-convolutional neural network (Fast R-CNN) model. After pruning and lightweight, the size of training model is greatly reduced without a significant reduction in accuracy, and the final precision reaches 87%, while the model size is reduced to 7,723 KB.