• Journal of Internet Computing and Services
• /
• v.20 no.1
• /
• pp.1-10
• /
• 2019

Since big-data text mining extracts many features and data, clustering and classification can result in high computational complexity and low reliability of the analysis results. In particular, a term document matrix obtained through text mining represents term-document features, but produces a sparse matrix. We designed an advanced genetic algorithm (GA) to extract features in text mining for detection model. Term frequency inverse document frequency (TF-IDF) is used to reflect the document-term relationships in feature extraction. Through a repetitive process, a predetermined number of features are selected. And, we used the sparsity score to improve the performance of detection model. If a spam mail data set has the high sparsity, detection model have low performance and is difficult to search the optimization detection model. In addition, we find a low sparsity model that have also high TF-IDF score by using s(F) where the numerator in fitness function. We also verified its performance by applying the proposed algorithm to text classification. As a result, we have found that our algorithm shows higher performance (speed and accuracy) in attack mail classification.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.23 no.4
• /
• pp.365-373
• /
• 2019

An Image Infra-Red(IIR) seeker is widely used for precision guided munitions to provide intelligent and precise target detection in terms of high kill probability. However, there have been issues in determining the performance versus cost trade-offs due to high cost of seeker comparing to other units of the munitions. In this paper, performance/cost evaluations have been carried out to find the most cost-effective solution for developing the IIR seekers. The relationships between the critical parameters and cost are investigated to determine the optimal point which represents the low cost with high performance. It is expected that the presented approach will be able to be used for guidelines to select the appropriate IIR seeker for the given operating conditions and can be useful to estimate the cost effectiveness of the precision guided munitions at early design stage.

• Convergence Security Journal
• /
• v.21 no.2
• /
• pp.67-79
• /
• 2021

Lessons Learned(LL) is a military term defined as all activities that promote future development by finding problems and need improvement in education and reality in the field of warfare development. In this paper, we focus on presenting actual examples and applying AI analysis inference techniques to solve revealed problems in promoting LL activities, such as long-term analysis, budget problems, and necessary expertise. AI legal advice services using cognitive computing-related technologies that have already been practical and in use, were judged to be the best examples to solve the problems of LL. This paper presents intelligent LL inference techniques, which utilize AI. To this end, we want to explore theoretical backgrounds such as LL analysis definitions and examples, evolution of AI into Machine Learning, cognitive computing, and apply it to new technologies in the defense sector using the newly proposed L²-OODA ensemble algorithm to contribute to implementing existing power improvement and optimization.

• Journal of Digital Convergence
• /
• v.19 no.8
• /
• pp.219-224
• /
• 2021

The convergence of various industries has removed the boundaries of software application fields and reduced the restrictions on convergence fields. Software requirements are diversified and they want to reconfigure software requirements in a fast cycle. Since various changes in requirements have to be accepted technically, research on methodologies and standards to increase the efficiency of software productivity and methods for standardizing and producing software are needed. In this study, we studied how the reusability and complexity of the software asset reconfiguration system appeared according to the developer's characteristics and environment to utilize the assets optimized in previous studies. At this time, we measured how the change in complexity according to the usability and asset composition method that appears according to the developer's characteristics appears, but there is a limit to the collected data, so it is necessary to secure the quality of the measured value through continuous data collection. In addition, an intelligent system application plan is needed to supplement the problem of context classification in the use stage of complex assets.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.12
• /
• pp.702-708
• /
• 2020

Recently, as LED display systems have become larger, research on effective power control methods for the systems has been in progress. This paper proposes a power control method to minimize power loss due to the difference in LED characteristics for each channel of a backlight unit (BLU) system. The proposed LED driver IC has a power optimization function and detects the minimum headroom voltage for constant current operation of all channels and linearly controls the DC-DC converter output. Thus, it minimizes power consumption due to unnecessary additional voltage. In addition, it does not require a voltage sensing comparator or a voltage generation circuit for each channel. This has a great advantage in reducing the chip size and for stabilization when implementing an integrated circuit. In order to verify the proposed function, an IC was designed using Cadence and Synopsys' design tools, and it was fabricated with a Magnachip 0.35um 5V/40V CMOS process. The experiments confirmed that the proposed power control method controls the minimum required voltage of the BLU system.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.4
• /
• pp.625-632
• /
• 2021

This paper studies the effect of the number of filters in the CNN (Convolutional Neural Network) layer on the performance of a noise attenuator. Speech is estimated from a noised speech signal using a 64-neuron, 16-kernel CNN filter and an error back-propagation algorithm. In this study, in order to verify the performance of the noise attenuator with respect to the number of filters, a program using Keras library was written and simulation was performed. As a result of simulation, it can be seen that this system has the smallest MSE (Mean Squared Error) and MAE (Mean Absolute Error) values when the number of filters is 16, and the performance is the lowest when there are 4 filters. And when there are more than 8 filters, it was shown that the MSE and MAE values do not differ significantly depending on the number of filters. From these results, it can be seen that about 8 or more filters must be used to express the characteristics of the speech signal.

• 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.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.7
• /
• pp.1749-1773
• /
• 2024

Recent advances in deep neural networks (DNNs) have greatly improved the accuracy and universality of various intelligent applications, at the expense of increasing model size and computational demand. Since the resources of end devices are often too limited to deploy a complete DNN model, offloading DNN inference tasks to cloud servers is a common approach to meet this gap. However, due to the limited bandwidth of WAN and the long distance between end devices and cloud servers, this approach may lead to significant data transmission latency. Therefore, device-edge collaborative inference has emerged as a promising paradigm to accelerate the execution of DNN inference tasks where DNN models are partitioned to be sequentially executed in both end devices and edge servers. Nevertheless, collaborative inference in heterogeneous edge environments with multiple edge servers, end devices and DNN tasks has been overlooked in previous research. To fill this gap, we investigate the optimization problem of collaborative inference in a heterogeneous system and propose a scheme CIS, i.e., collaborative inference scheme, which jointly combines DNN partition, task offloading and scheduling to reduce the average weighted inference latency. CIS decomposes the problem into three parts to achieve the optimal average weighted inference latency. In addition, we build a prototype that implements CIS and conducts extensive experiments to demonstrate the scheme's effectiveness and efficiency. Experiments show that CIS reduces 29% to 71% on the average weighted inference latency compared to the other four existing schemes.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.18 no.7
• /
• pp.1726-1748
• /
• 2024

Trajectory planning is vital for autonomous systems like robotics and UAVs, as it determines optimal, safe paths considering physical limitations, environmental factors, and agent interactions. Recent advancements in trajectory planning and future location prediction stem from rapid progress in machine learning and optimization algorithms. In this paper, we proposed a novel framework for Spatial-temporal transformer-based feed-forward neural networks (STTFFNs). From the traffic flow local area point of view, skip-gram model is trained on trajectory data to generate embeddings that capture the high-level features of different trajectories. These embeddings can then be used as input to a transformer-based trajectory planning model, which can generate trajectories for new objects based on the embeddings of similar trajectories in the training data. In the next step, distant regions, we embedded feedforward network is responsible for generating the distant trajectories by taking as input a set of features that represent the object's current state and historical data. One advantage of using feedforward networks for distant trajectory planning is their ability to capture long-term dependencies in the data. In the final step of forecasting for future locations, the encoder and decoder are crucial parts of the proposed technique. Spatial destinations are encoded utilizing location-based social networks(LBSN) based on visiting semantic locations. The model has been specially trained to forecast future locations using precise longitude and latitude values. Following rigorous testing on two real-world datasets, Porto and Manhattan, it was discovered that the model outperformed a prediction accuracy of 8.7% previous state-of-the-art methods.

• Smart Structures and Systems
• /
• v.33 no.4
• /
• pp.291-300
• /
• 2024

Uncertainty of the model, system delay and drive dynamics can be considered as normal uncertainties, and the main source of uncertainty in the seismic control system is related to the nature of the simulated seismic error. In this case, optimizing the management strategy for one particular seismic record will not yield the best results for another. In this article, we propose a framework for online management of active structural management systems with seismic uncertainty. For this purpose, the concept of reinforcement learning is used for online optimization of active crowd management software. The controller consists of a differential controller, an unplanned gain ratio, the gain of which is enhanced using an online reinforcement learning algorithm. In addition, the proposed controller includes a dynamic status forecaster to solve the delay problem. To evaluate the performance of the proposed controllers, thousands of ground motion data sets were processed and grouped according to their spectrum using fuzzy clustering techniques with spatial hazard estimation. Finally, the controller is implemented in a laboratory scale configuration and its operation is simulated on a vibration table using cluster location and some actual seismic data. The test results show that the proposed controller effectively withstands strong seismic interference with delay. The goals of this paper are towards access to adequate, safe and affordable housing and basic services, promotion of inclusive and sustainable urbanization and participation, implementation of sustainable and disaster-resilient buildings, sustainable human settlement planning and manage. Simulation results is believed to achieved in the near future by the ongoing development of AI and control theory.