• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.2
• /
• pp.483-503
• /
• 2014

In this paper, human action recognition using pyramid histograms of oriented gradients and collaborative multi-task learning is proposed. First, we accumulate global activities and construct motion history image (MHI) for both RGB and depth channels respectively to encode the dynamics of one action in different modalities, and then different action descriptors are extracted from depth and RGB MHI to represent global textual and structural characteristics of these actions. Specially, average value in hierarchical block, GIST and pyramid histograms of oriented gradients descriptors are employed to represent human motion. To demonstrate the superiority of the proposed method, we evaluate them by KNN, SVM with linear and RBF kernels, SRC and CRC models on DHA dataset, the well-known dataset for human action recognition. Large scale experimental results show our descriptors are robust, stable and efficient, and outperform the state-of-the-art methods. In addition, we investigate the performance of our descriptors further by combining these descriptors on DHA dataset, and observe that the performances of combined descriptors are much better than just using only sole descriptor. With multimodal features, we also propose a collaborative multi-task learning method for model learning and inference based on transfer learning theory. The main contributions lie in four aspects: 1) the proposed encoding the scheme can filter the stationary part of human body and reduce noise interference; 2) different kind of features and models are assessed, and the neighbor gradients information and pyramid layers are very helpful for representing these actions; 3) The proposed model can fuse the features from different modalities regardless of the sensor types, the ranges of the value, and the dimensions of different features; 4) The latent common knowledge among different modalities can be discovered by transfer learning to boost the performance.

• Journal of KIISE
• /
• v.42 no.6
• /
• pp.801-806
• /
• 2015

The Short Message Service (SMS) is one of the most popular communication tools in the world. As the cost of SMS decreases, SMS spam has been growing largely. Even though there are many existing studies on SMS spam detection, researchers commonly have limitation collecting users' private SMS contents. They need to gather the information related to social network as well as personal SMS due to the intelligent spammers being aware of the social networks. Therefore, this paper proposes the Social network Building Scheme for SMS spam detection (SBSS) algorithm that builds synthetic social network dataset realistically, without the collection of private information. Also, we analyze and categorize the attack types of SMS spam to build more complete and realistic social network dataset including SMS spam.

• Journal of the Korea Convergence Society
• /
• v.9 no.7
• /
• pp.1-7
• /
• 2018

Feature selection, one of data preprocessing techniques, is one of major research areas in many applications dealing with large dataset. It has been used in pattern recognition, machine learning and data mining, and is now widely applied in a variety of fields such as text classification, image retrieval, intrusion detection and genome analysis. The proposed method is based on a genetic algorithm which is one of meta-heuristic algorithms. There are two methods of finding feature subsets: a filter method and a wrapper method. In this study, we use a wrapper method, which evaluates feature subsets using a real classifier, to find an optimal feature subset. The training dataset used in the experiment has a severe class imbalance and it is difficult to improve classification performance for rare classes. After preprocessing the training dataset with SMOTE, we select features and evaluate them with various machine learning algorithms.

• The KIPS Transactions:PartD
• /
• v.18D no.5
• /
• pp.385-392
• /
• 2011

Recently, the concern of IT-in-Agriculture convergence technology that combines information technology and agriculture is increasing rapidly. Especially, the crop cultivation related prediction services by spatial data mining (SDM) can play an important role in reducing the damage of natural disaster and enhancing crop productivity. However, the data conversion and integration procedure to acquire the learning dataset of SDM for the prediction service need a lot of effort and time, because of their heterogeneity between distributed data. In addition, calculating spatial neighborhood relationships between spatial and non-spatial data necessitates requires the complicated calculation procedure for large dataset. In this paper, we suggest a SOA-based data integration framework that can effectively integrate distributed heterogeneous data by treating each data source as a service unit and support to find the optimal prediction service by improving productivity of learning dataset for SDM. In our experiment, we confirmed that our framework can be effectively applied to find the optimal prediction service for the frost damage area, by considering the case of peach crop cultivation in Icheon in Korea.

• Nutrition Research and Practice
• /
• v.13 no.6
• /
• pp.521-528
• /
• 2019

BACKGROUND/OBJECTIVES: The aim of this study was to develop Korean food image detection and recognition model for use in mobile devices for accurate estimation of dietary intake. MATERIALS/METHODS: We collected food images by taking pictures or by searching web images and built an image dataset for use in training a complex recognition model for Korean food. Augmentation techniques were performed in order to increase the dataset size. The dataset for training contained more than 92,000 images categorized into 23 groups of Korean food. All images were down-sampled to a fixed resolution of $150{\times}150$ and then randomly divided into training and testing groups at a ratio of 3:1, resulting in 69,000 training images and 23,000 test images. We used a Deep Convolutional Neural Network (DCNN) for the complex recognition model and compared the results with those of other networks: AlexNet, GoogLeNet, Very Deep Convolutional Neural Network, VGG and ResNet, for large-scale image recognition. RESULTS: Our complex food recognition model, K-foodNet, had higher test accuracy (91.3%) and faster recognition time (0.4 ms) than those of the other networks. CONCLUSION: The results showed that K-foodNet achieved better performance in detecting and recognizing Korean food compared to other state-of-the-art models.

• Journal of Convergence for Information Technology
• /
• v.9 no.9
• /
• pp.1-12
• /
• 2019

A dialog system becomes a new way of communication between human and computer. The dialog system takes human voice as an input, and gives a proper response in voice or perform an action. Although there are several well-known products of dialog system (e.g., Amazon Echo, Naver Wave), they commonly suffer from a problem of out-of-domain utterances. If it poorly detects out-of-domain utterances, then it will significantly harm the user satisfactory. There have been some studies aimed at solving this problem, but it is still necessary to study about this intensively. In this paper, we give an overview of the previous studies of out-of-domain detection in terms of three point of view: dataset, feature, and method. As there were relatively smaller studies of this topic due to the lack of datasets, we believe that the most important next research step is to construct and share a large dataset for dialog system, and thereafter try state-of-the-art techniques upon the dataset.

• International Journal of Computer Science & Network Security
• /
• v.21 no.9
• /
• pp.19-30
• /
• 2021

Diabetes Mellitus (DM) is one of common chronic diseases leading to severe health complications that may cause death. The disease influences individuals, community, and the government due to the continuous monitoring, lifelong commitment, and the cost of treatment. The World Health Organization (WHO) considers Saudi Arabia as one of the top 10 countries in diabetes prevalence across the world. Since most of the medical services are provided by the government, the cost of the treatment in terms of hospitals and clinical visits and lab tests represents a real burden due to the large scale of the disease. The ability to predict the diabetic status of a patient without the laboratory tests by performing screening based on some personal features can lessen the health and economic burden caused by diabetes alone. The goal of this paper is to investigate the prediction of diabetic and prediabetic patients by considering factors other than the laboratory tests, as required by physicians in general. With the data obtained from local hospitals, medical records were processed to obtain a dataset that classified patients into three classes: diabetic, prediabetic, and non-diabetic. After applying three machine learning algorithms, we established good performance for accuracy, precision, and recall of the models on the dataset. Further analysis was performed on the data to identify important non-laboratory variables related to the patients for diabetes classification. The importance of five variables (gender, physical activity level, hypertension, BMI, and age) from the person's basic health data were investigated to find their contribution to the state of a patient being diabetic, prediabetic or normal. Our analysis presented great agreement with the risk factors of diabetes and prediabetes stated by the American Diabetes Association (ADA) and other health institutions worldwide. We conclude that by performing class-specific analysis of the disease, important factors specific to Saudi population can be identified, whose management can result in controlling the disease. We also provide some recommendations learnt from this research.

• Journal of the Semiconductor & Display Technology
• /
• v.22 no.1
• /
• pp.11-17
• /
• 2023

This paper proposes a lightweight deep learning network for detecting an image splicing forgery. The research on image forgery detection using CNN, a deep learning network, and research on detecting and localizing forgery in pixel units are in progress. Among them, CAT-Net, which learns the discrete cosine transform coefficients of images together with images, was released in 2022. The DCT coefficients presented by CAT-Net are combined with the JPEG artifact learning module and the backbone model as pre-learning, and the weights are fixed. The dataset used for pre-training is not included in the public dataset, and the backbone model has a relatively large number of network parameters, which causes overfitting in a small dataset, hindering generalization performance. In this paper, this learning module is designed to learn the characterization depending on the DCT domain in real-time during network training without pre-training. The DCT RRU-Net proposed in this paper is a network that combines RRU-Net which detects forgery by learning only images and JPEG artifact learning module. It is confirmed that the network parameters are less than those of CAT-Net, the detection performance of forgery is better than that of RRU-Net, and the generalization performance for various datasets improves through the network architecture and training method of DCT RRU-Net.

• Journal of the Society of Disaster Information
• /
• v.19 no.4
• /
• pp.968-975
• /
• 2023

Purpose: We would like to confirm that the false positive rate of flames/smoke is high when detecting fires. Propose a method and dataset to recognize and classify fire situations to reduce the false detection rate. Method: Using the video as learning data, the characteristics of the fire situation were extracted and applied to the classification model. For evaluation, the model performance of Yolov8 and Slowfast were compared and analyzed using the fire dataset conducted by the National Information Society Agency (NIA). Result: YOLO's detection performance varies sensitively depending on the influence of the background, and it was unable to properly detect fires even when the fire scale was too large or too small. Since SlowFast learns the time axis of the video, we confirmed that detects fire excellently even in situations where the shape of an atypical object cannot be clearly inferred because the surrounding area is blurry or bright. Conclusion: It was confirmed that the fire detection rate was more appropriate when using a video-based artificial intelligence detection model rather than using image data.

• Journal of Internet Computing and Services
• /
• v.25 no.4
• /
• pp.57-64
• /
• 2024

This paper aims to enhance the accuracy of music genre classification for music tracks where genre information is not provided, by utilizing machine learning to classify a large amount of music data. The paper proposes collecting and preprocessing data instead of using the commonly employed GTZAN dataset in previous research for genre classification in music. To create a dataset with superior classification performance compared to the GTZAN dataset, we extract specific segments with the highest energy level of the onset. We utilize 57 features as the main characteristics of the music data used for training, including Mel Frequency Cepstral Coefficients (MFCC). We achieved a training accuracy of 85% and a testing accuracy of 71% using the Support Vector Machine (SVM) model to classify into Classical, Jazz, Country, Disco, Soul, Rock, Metal, and Hiphop genres based on preprocessed data.