• ETRI Journal
• /
• 제44권5호
• /
• pp.746-758
• /
• 2022

This paper focuses on a mobile edge-computing-enabled heterogeneous network. A battery level-aware task-scheduling framework is proposed to improve the energy efficiency and prolong the operating hours of battery-powered mobile devices. The formulated optimization problem is a typical mixed-integer nonlinear programming problem. To solve this nondeterministic polynomial (NP)-hard problem, a decomposition-based task-scheduling algorithm is proposed. Using an alternating optimization technology, the original problem is divided into three subproblems. In the outer loop, task offloading decisions are yielded using a pruning search algorithm for the task offloading subproblem. In the inner loop, closed-form solutions for computational resource allocation subproblems are derived using the Lagrangian multiplier method. Then, it is proven that the transmitted power-allocation subproblem is a unimodal problem; this subproblem is solved using a gradient-based bisection search algorithm. The simulation results demonstrate that the proposed framework achieves better energy efficiency than other frameworks. Additionally, the impact of the battery level-aware scheme on the operating hours of battery-powered mobile devices is also investigated.

• Journal of Intelligence and Information Systems
• /
• 제4권1호
• /
• pp.79-87
• /
• 1998

정확한 태풍진로 예측은 동아시아 최대의 자연재해인 태풍의 피해를 최소화하는데 필수적이다. 기상역학에 기초를 둔 수치모델과 회귀분석등의 통계적 접근법이 사용되어왔다. 본 논문에서는 비선형 신경망모델인 다층퍼셉트론을 제안한다. 즉, 태풍진로예측을 이동경로, 속도, 기압 등의 변수로 이루어진 시계열의 예측으로 본다. 1945년부터 1989년까지 한반도에 접근한 태풍 데이터를 이용하여 제안된 신경망을 학습한 후, 94, 95년도에 접근한 태풍의 진로를 예측하였다. 신경망의 예측성능은 수치모델의 성능보다 조금 우수하거나 비슷하였다. 신경망의 성능은 충분히 더 향상될 수 있는 여지가 있다. 또한, 고가의 슈퍼컴퓨터로 여러 시간 계산을 해야하는 수치모델에 비하여 PC상에서 수초만에 계산을 할 수 있는 신경망 모델은 비용 면에서도 장점이 있다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 한국방송∙미디어공학회 2022년도 추계학술대회
• /
• pp.123-125
• /
• 2022

본 논문에서는 판별자를 활용하여 Image to Image translation(I2I) 분야에서 사용되는 적대적 생성 신경망(GAN)을 압축하는 방법을 제시한다. 우선, 잘 학습된 판별자와 생성자 사이의 adversarial loss 를 활용하여 생성자 내 필터들의 중요도 점수를 매겨준다. 그리고 생성자 내의 필터들을 중요도 점수를 기준으로 나열한 후 점수가 낮은 필터들을 제거하는 필터 프루닝을 한번 수행하여 적은 시간 비용으로 생성자를 압축한다. 마지막으로 지식 증류를 활용해 압축된 생성자를 학습시켜 기존의 생성자와 유사한 성능을 보이도록 하였다. 이 과정들을 통해 효과적이고 빠르게 GAN 모델을 압축할 수 있음을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 한국방송∙미디어공학회 2020년도 하계학술대회
• /
• pp.675-679
• /
• 2020

최근 이미지 분류의 성능 향상을 위해 깊은 레이어와 넓은 채널을 가지는 모델들이 제안되어져 왔다. 높은 분류 정확도를 보이는 모델을 제안하는 것은 과한 컴퓨팅 파워와 계산시간을 요구한다. 본 논문에서는 이미지 분류 기법에서 사용되는 딥 뉴럴 네트워크 모델에 있어, 프루닝 방법을 통해 상대적으로 불필요한 가중치를 제거함과 동시에 분류 정확도 하락을 최소로 하는 동적 필터 프루닝 방법을 제시한다. 원샷 프루닝 기법, 정적 필터 프루닝 기법과 다르게 제거된 가중치에 대해서 소생 기회를 제공함으로써 더 좋은 성능을 보인다. 또한, 재학습이 필요하지 않기 때문에 빠른 계산 속도와 적은 컴퓨팅 파워를 보장한다. ResNet20 에서 CIFAR10 데이터셋에 대하여 실험한 결과 약 50%의 압축률에도 88.74%의 분류 정확도를 보였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 한국방송∙미디어공학회 2020년도 하계학술대회
• /
• pp.662-663
• /
• 2020

최근 MPEG 에서는 인공신경망 모델을 다양한 딥러닝 프레임워크에서 상호운용 가능한 포맷으로 압축 표현할 수 있는 NNR(Compression of Neural Network for Multimedia Content Description and Analysis) 표준화를 진행하고 있다. 본 논문에서는 MPEG-NNR 에서 CNN 모델을 압축하기 위한 지역 비선형 양자화(Local Non-linear Quantization: LNQ) 기법을 제시한다. 제안하는 LNQ 는 균일 양자화된 CNN 모델의 각 계층의 가중치 행렬 블록 단위로 추가적인 비선형 양자화를 적용한다. 또한, 제안된 LNQ 는 가지치기(pruning)된 모델의 경우 블록내의 영(zero) 값의 가중치들은 그대로 전송하고 영이 아닌 가중치만을 이진 군집화를 적용한다. 제안 기법은 음성 분류를 위한 CNN 모델(DCASE Task)의 압축 실험에서 기존 균일 양자화를 대비 동일한 분류 성능에서 약 1.78 배 압축 성능 향상이 있음을 확인하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 한국해양정보통신학회 2008년도 추계종합학술대회 B
• /
• pp.147-150
• /
• 2008

The real road network is regarded as a grid, and the grid is divided by fixed-sized cells. The path-finding is composed of two step searching. First searching travels on the abstract graph which is composed of a set of psuedo vertexes and a set of psuedo edges that are created by real road network and fixed-sized cells. The result of the first searching is a psuedo path which is composed of a set of selected psuedo edges. The cells intersected with the psuedo path are called as valid cells. The second searching travels with $A^*$ algorithm on valid cells. As pruning search space by removing the invalid cells, it would be possible to reduce the cost of exploring on real road network. In this paper, we present the method of creating the abstract graph and propose a path-finding algorithm on the abstract graph for extracting search space before traveling on real road network.

• Journal of Korea Spatial Information System Society
• /
• 제11권4호
• /
• pp.39-46
• /
• 2009

Generally, Path-finding algorithms which use heuristic function may occur a problem of the increase of exploring cost in case of that there is no way determined by heuristic function or there are 2 way more which have almost same cost. In this paper, we propose an abstract graph for path-finding with dynamic information. The abstract graph is a simple graph as real road network is abstracted. The abstract graph is created by fixed-size cells and real road network. Path-finding with the abstract graph is composed of two step searching, path-finding on the abstract graph and on the real road network. We performed path-finding algorithm with the abstract graph against A* algorithm based on fixed-size cells on road network that consists of 106,254 edges. In result of evaluation of performance, cost of exploring in path-finding with the abstract graph is about 3~30% less than A* algorithm based on fixed-size cells. Quality of path in path-finding with the abstract graph is, However, about 1.5~6.6% more than A* algorithm based on fixed-size cells because edges eliminated are not candidates for path-finding.

• The KIPS Transactions:PartD
• /
• 제12D권2호
• /
• pp.259-266
• /
• 2005

Recently data centric routing or application dependent routing protocols are emerged in mobile ad hoc networks. In this paper, we propose a routing method for query processing in MANET(Mobile Ad hoc NETwork) environment, called road-based query routing, with consideration on real time traffic information of large number of vehicles. In particular, we focus on the method that process arrival time dependent shortest path query in MANET without a central server on the road networks. The main idea of our approach lies in a routing message that includes query predicates based on the road connectivity and on data gathering method in real time from vehicles on the road by ad-hoc network. We unify route discovery phase and data delivery(query processing) phase in our mechanism and reduce unnecessary flooding messages by pruning mobile nodes which are not on the same or neighboring road segments. In order to evaluate the performances of the proposed method, we established a model of road networks and mobile nodes which travel along the roads. The measurement factor is the number of nodes to whom route request is propagated according to each pruning strategy. Simulation result shows that road information is a dominant factor to reduce the number of messages.

• Journal of Broadcast Engineering
• /
• 제26권6호
• /
• pp.778-789
• /
• 2021

Recently, various lightweight methods for using Convolutional Neural Network(CNN) models in mobile devices have emerged. Weight quantization, which lowers bit precision of weights, is a lightweight method that enables a model to be used through integer calculation in a mobile environment where GPU acceleration is unable. Weight quantization has already been used in various models as a lightweight method to reduce computational complexity and model size with a small loss of accuracy. Considering the size of memory and computing speed as well as the storage size of the device and the limited network environment, this paper proposes a method of compressing integer weights after quantization using a video codec as a method. To verify the performance of the proposed method, experiments were conducted on VGG16, Resnet50, and Resnet18 models trained with ImageNet and Places365 datasets. As a result, loss of accuracy less than 2% and high compression efficiency were achieved in various models. In addition, as a result of comparison with similar compression methods, it was verified that the compression efficiency was more than doubled.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권5호
• /
• pp.1433-1449
• /
• 2023

In recent years, the rapid development of social networks has led to a rapid increase in the amount of information available on the Internet, which contains a large amount of sensitive information related to pornography, politics, and terrorism. In the aspect of sensitive image detection, the existing machine learning algorithms are confronted with problems such as large model size, long training time, and slow detection speed when auditing and supervising. In order to detect sensitive images more accurately and quickly, this paper proposes a multiclassification sensitive image detection method based on lightweight Convolutional Neural Network. On the basis of the EfficientNet model, this method combines the Ghost Module idea of the GhostNet model and adds the SE channel attention mechanism in the Ghost Module for feature extraction training. The experimental results on the sensitive image data set constructed in this paper show that the accuracy of the proposed method in sensitive information detection is 94.46% higher than that of the similar methods. Then, the model is pruned through an ablation experiment, and the activation function is replaced by Hard-Swish, which reduces the parameters of the original model by 54.67%. Under the condition of ensuring accuracy, the detection time of a single image is reduced from 8.88ms to 6.37ms. The results of the experiment demonstrate that the method put forward has successfully enhanced the precision of identifying multi-class sensitive images, significantly decreased the number of parameters in the model, and achieved higher accuracy than comparable algorithms while using a more lightweight model design.