• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.33 no.8B
• /
• pp.707-718
• /
• 2008

The methodology of classifying traffics is changing from payload based or port based to machine learning based in order to overcome the dynamic changes of application's characteristics. However, current state of traffic classification using machine learning (ML) algorithms is ongoing under the offline environment. Specifically, most of the current works provide results of traffic classification using cross validation as a test method. Also, they show classification results based on traffic flows. However, these traffic classification results are not useful for practical environments of the network traffic monitoring. This paper compares the classification results using cross validation with those of using split validation as the test method. Also, this paper compares the classification results based on flow to those based on bytes. We classify network traffics by using various feature sets and machine learning algorithms such as J48, REPTree, RBFNetwork, Multilayer perceptron, BayesNet, and NaiveBayes. In this paper, we find the best feature sets and the best ML algorithm for classifying traffics using the split validation.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.1
• /
• pp.81-92
• /
• 2023

Traffic state classification is crucial for time-of-day (TOD) traffic signal control. This paper proposed a traffic state classification technique applying Deep-Embedded Clustering (DEC) method that uses a high dimensional traffic data observed at all signalized intersections in a traffic signal control sub area (SA). So far, signal timing plan has been determined based on the traffic data observed at the critical intersection in SA. The current method has a limitation that it cannot consider a comprehensive traffic situation in SA. The proposed method alleviates the curse of dimensionality and turns out to overcome the shortcomings of the current signal timing plan.

• Korean Journal of Remote Sensing
• /
• v.30 no.4
• /
• pp.493-503
• /
• 2014

A hierarchical classification scheme, which can reduce the spectral ambiguity and also reflect crop cultivation patterns from past land-cover maps, is presented for the purpose of the early production of crop classification maps in large-scale crop areas. Specifically, the effects of mixed pixels are minimized not only by applying a hierarchical classification approach based on different spectral characteristics from crop growth cycles, but also by considering temporal contextual information derived from past crop cultivation patterns. The applicability of the presented classification scheme was evaluated by a case study of Iowa State in USA with time-series MODIS 250 m Normalized Difference Vegetation Index(NDVI) data sets and past Cropland Data Layers(CDLs). Corn and soybean, which are major crop types in the study area and also display spectral similarity, could be properly classified by applying different classification stages and accounting for past crop cultivation patterns. The classification result by the presented scheme showed increases of minimum 7.68%p and maximum 20.96%p in overall accuracy, compared with one based on purely spectral information. In addition, the combination of temporal contextual information during classification was less affected by the number of NDVI data sets and the best overall accuracy of 86.63% was achieved. Thus, it is expected that this classification scheme can be effectively used for the early production of large-area crop classification maps in major feed-grain importing countries.

• Journal of Korea Multimedia Society
• /
• v.25 no.1
• /
• pp.39-51
• /
• 2022

Image-based gender classification and age estimation of human are classic problems in computer vision. Most of researches in this field focus just only one task of either gender classification or age estimation and most of the reported methods for each task focus on accuracy performance and are not computationally light. Thus, running both tasks together simultaneously on low cost mobile or embedded systems with limited cpu processing speed and memory capacity are practically prohibited. In this paper, we propose a novel light-weight gender classification and age estimation method based on ensemble multitasking deep learning with light-weight processing neural network architecture, which processes both gender classification and age estimation simultaneously and in real-time even for embedded systems. Through experiments over various well-known datasets, it is shown that the proposed method performs comparably to the state-of-the-art gender classification and/or age estimation methods with respect to accuracy and runs fast enough (average 14fps) on a Jestson Nano embedded board.

• International Journal of Computer Science & Network Security
• /
• v.22 no.10
• /
• pp.406-412
• /
• 2022

Scene classification of very high-resolution (VHR) imagery can attribute semantics to land cover in a variety of domains. Real-world application requirements have not been addressed by conventional techniques for remote sensing image classification. Recent research has demonstrated that deep convolutional neural networks (CNNs) are effective at extracting features due to their strong feature extraction capabilities. In order to improve classification performance, these approaches rely primarily on semantic information. Since the abstract and global semantic information makes it difficult for the network to correctly classify scene images with similar structures and high interclass similarity, it achieves a low classification accuracy. We propose a VHR remote sensing image classification model that uses extracts the global feature from the original VHR image using an EfficientNet-V2L CNN pre-trained to detect similar classes. The image is then classified using a multilayer perceptron (MLP). This method was evaluated using two benchmark remote sensing datasets: the 21-class UC Merced, and the 38-class PatternNet. As compared to other state-of-the-art models, the proposed model significantly improves performance.

• Journal of IKEEE
• /
• v.13 no.2
• /
• pp.159-166
• /
• 2009

So far, many researches have been done to retrieve music information using static classification descriptors such as genre and mood. Since static classification descriptors are based on diverse content-based musical features, they are effective in retrieving similar music in terms of such features. However, human emotion or mood transition triggered by music enables more effective and sophisticated query in music retrieval. So far, few works have been done to evaluate the effect of human mood transition by music. Using formal representation of such mood transitions, we can provide personalized service more effectively in the new applications such as music recommendation. In this paper, we first propose our Emotion State Transition Model (ESTM) for describing human mood transition by music and then describe a music classification and recommendation scheme based on the ESTM. In the experiment, diverse content-based features were extracted from music clips, dimensionally reduced by NMF (Non-negative Matrix Factorization, and classified by SVM (Support Vector Machine). In the performance analysis, we achieved average accuracy 67.54% and maximum accuracy 87.78%.

• Journal of the Korean Geographical Society
• /
• v.38 no.5
• /
• pp.667-685
• /
• 2003

To address the requirements of gap analysis for species protection, as well as the needs of state and federal agencies for detailed digital land cover, a 43-class map at the vegetation alliance level was created for the state of Kansas using multi-temporal Thematic Mapper imagery. The mapping approach included the use of three-date multi-seasonal imagery, a two-stage classification approach that first masked out cropland areas using unsupervised classification and then mapped natural vegetation with supervised classification, visualization techniques utilizing a map of small multiples and field experts, and extensive use of ancillary data in post-hoc processing. Accuracy assessment was conducted at three levels of generalization (Anderson Level I, vegetation formation, and vegetation alliance) and three cross-tabulation approaches. Overall accuracy ranged from 51.7% to 89.4%, depending on level of generalization, while accuracy figures for individual alliance classes varied by area covered and level of sampling.

• Journal of Sensor Science and Technology
• /
• v.22 no.2
• /
• pp.162-173
• /
• 2013

The purpose of this paper is to classify VOC gases by emulating the characteristics found in biological olfaction. For this purpose, we propose new signal processing method based a polymeric chemical sensor array consisting of 4096 sensors which is created by NEUROCHEM project. To remove unstable sensors generated in the manufacturing process of very large scaled chemical sensor array, we used discrete wavelet transformation and cosine similarity. And, to remove the supernumerary redundancy, we proposed the method of selecting candidates of representative sensor representing sensors with similar features by Fuzzy c-means algorithm. In addition, we proposed an improved algorithm for selecting representative sensors among candidates of representative sensors to better enhance classification ability. However, Classification for very large scaled sensor array has a great deal of time in process of learning because many sensors are used for learning though a redundancy is removed. Throughout experimental trials for classification, we confirmed the proposed method have an outstanding classification ability, at transient state as well as steady state.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.12
• /
• pp.3383-3397
• /
• 2023

Scene graphs serve as semantic abstractions of images and play a crucial role in enhancing visual comprehension and reasoning. However, the performance of Scene Graph Generation is often compromised when working with biased data in real-world situations. While many existing systems focus on a single stage of learning for both feature extraction and classification, some employ Class-Balancing strategies, such as Re-weighting, Data Resampling, and Transfer Learning from head to tail. In this paper, we propose a novel approach that decouples the feature extraction and classification phases of the scene graph generation process. For feature extraction, we leverage a transformer-based architecture and design an adaptive calibration function specifically for predicate classification. This function enables us to dynamically adjust the classification scores for each predicate category. Additionally, we introduce a Distribution Alignment technique that effectively balances the class distribution after the feature extraction phase reaches a stable state, thereby facilitating the retraining of the classification head. Importantly, our Distribution Alignment strategy is model-independent and does not require additional supervision, making it applicable to a wide range of SGG models. Using the scene graph diagnostic toolkit on Visual Genome and several popular models, we achieved significant improvements over the previous state-of-the-art methods with our model. Compared to the TDE model, our model improved mR@100 by 70.5% for PredCls, by 84.0% for SGCls, and by 97.6% for SGDet tasks.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.8
• /
• pp.3865-3883
• /
• 2016

Infrastructure as a Service (IaaS) encapsulates computer hardware into a large amount of virtual and manageable instances mainly in the form of virtual machine (VM), and provides rental service for users. Currently, VM anomaly incidents occasionally occur, which leads to performance issues and even downtime. This paper aims at detecting anomalous VMs based on performance metrics data of VMs. Due to the dynamic nature and increasing scale of IaaS, detecting anomalous VMs from voluminous correlated and non-Gaussian monitored performance data is a challenging task. This paper designs an anomaly detection framework to solve this challenge. First, it collects 53 performance metrics to reflect the running state of each VM. The collected performance metrics are testified not to follow the Gaussian distribution. Then, it employs independent components analysis (ICA) instead of principal component analysis (PCA) to extract independent components from collected non-Gaussian performance metric data. For anomaly detection, it employs multi-class Bayesian classification to determine the current state of each VM. To evaluate the performance of the designed detection framework, four types of anomalies are separately or jointly injected into randomly selected VMs in a campus-wide testbed. The experimental results show that ICA-based detection mechanism outperforms PCA-based and LDA-based detection mechanisms in terms of sensitivity and specificity.