• ETRI Journal
• /
• v.44 no.2
• /
• pp.286-299
• /
• 2022

Human activity recognition in real time is a challenging task. Recently, a plethora of studies has been proposed using deep learning architectures. The implementation of these architectures requires the high computing power of the machine and a massive database. However, handcrafted features-based machine learning models need less computing power and very accurate where features are effectively extracted. In this study, we propose a handcrafted model based on three-dimensional sequential skeleton data. The human body skeleton movement over a frame is computed through joint positions in a frame. The joints of these skeletal frames are projected into two-dimensional space, forming a "movement polygon." These polygons are further transformed into a one-dimensional space by computing amplitudes at different angles from the centroid of polygons. The feature vector is formed by the sampling of these amplitudes at different angles. The performance of the algorithm is evaluated using a support vector machine on four public datasets: MSR Action3D, Berkeley MHAD, TST Fall Detection, and NTU-RGB+D, and the highest accuracies achieved on these datasets are 94.13%, 93.34%, 95.7%, and 86.8%, respectively. These accuracies are compared with similar state-of-the-art and show superior performance.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.101-101
• /
• 2020

Traditional stochastic simulation of hydroclimatological variables often underestimates the variability and correlation structure of larger timescale due to the difficulty in preserving long-term memory. However, the Long Short-Term Memory (LSTM) model illustrates a remarkable long-term memory from the recursive hidden and cell states. The current study, therefore, employed the LSTM model in stochastic generation of hydrologic and climate variables to examine how much the LSTM model can preserve the long-term memory and overcome the drawbacks of conventional time series models such as autoregressive (AR). A trigonometric function and the Rössler system as well as real case studies for hydrological and climatological variables were tested. Results presented that the LSTM model reproduced the variability and correlation structure of the larger timescale as well as the key statistics of the original time domain better than the AR and other traditional models. The hidden and cell states of the LSTM containing the long-memory and oscillation structure following the observations allows better performance compared to the other tested conventional models. This good representation of the long-term variability can be important in water manager since future water resources planning and management is highly related with this long-term variability.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.17 no.11
• /
• pp.3121-3144
• /
• 2023

As interest in the metaverse has grown, there has been a demand for avatars that can represent individual users. Consequently, research has been conducted to reduce the time and cost required for the current 3D human modeling process. However, the recent automatic generation of 3D humans has been focused on creating avatars with a realistic human form. Furthermore, the existing methods have limitations in generating avatars with imbalanced or unrealistic body shapes, and their utilization is limited due to the absence of datasets. Therefore, this paper proposes a new framework for automatically transforming and creating stylized 3D avatars. Our research presents a definitional approach and methodology for creating non-realistic character avatars, in contrast to previous studies that focused on creating realistic humans. We define a new shape representation parameter and use a deep learning-based method to extract character body information and perform automatic template mesh transformation, thereby obtaining non-realistic or unbalanced human meshes. We present the resulting outputs visually, conducting user evaluations to demonstrate the effectiveness of our proposed method. Our approach provides an automatic mesh transformation method tailored to the growing demand for avatars of various body types and extends the existing method to the 3D cartoon stylized avatar domain.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.29-35
• /
• 2020

In this paper, we propose a new structured entity recognition DeNERT model. Recently, the field of natural language processing has been actively researched using pre-trained language representation models with a large amount of corpus. In particular, the named entity recognition, which is one of the fields of natural language processing, uses a supervised learning method, which requires a large amount of training dataset and computation. Reinforcement learning is a method that learns through trial and error experience without initial data and is closer to the process of human learning than other machine learning methodologies and is not much applied to the field of natural language processing yet. It is often used in simulation environments such as Atari games and AlphaGo. BERT is a general-purpose language model developed by Google that is pre-trained on large corpus and computational quantities. Recently, it is a language model that shows high performance in the field of natural language processing research and shows high accuracy in many downstream tasks of natural language processing. In this paper, we propose a new named entity recognition DeNERT model using two deep learning models, DQN and BERT. The proposed model is trained by creating a learning environment of reinforcement learning model based on language expression which is the advantage of the general language model. The DeNERT model trained in this way is a faster inference time and higher performance model with a small amount of training dataset. Also, we validate the performance of our model's named entity recognition performance through experiments.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.8
• /
• pp.355-364
• /
• 2023

As advanced cyber threats continue to increase in recent years, it is difficult to detect new types of cyber attacks with existing pattern or signature-based intrusion detection method. Therefore, research on anomaly detection methods using data learning-based artificial intelligence technology is increasing. In addition, supervised learning-based anomaly detection methods are difficult to use in real environments because they require sufficient labeled data for learning. Research on an unsupervised learning-based method that learns from normal data and detects an anomaly by finding a pattern in the data itself has been actively conducted. Therefore, this study aims to extract a latent vector that preserves useful sequence information from sequence log data and develop an anomaly detection learning model using the extracted latent vector. Word2Vec was used to create a dense vector representation corresponding to the characteristics of each sequence, and an unsupervised autoencoder was developed to extract latent vectors from sequence data expressed as dense vectors. The developed autoencoder model is a recurrent neural network GRU (Gated Recurrent Unit) based denoising autoencoder suitable for sequence data, a one-dimensional convolutional neural network-based autoencoder to solve the limited short-term memory problem that GRU can have, and an autoencoder combining GRU and one-dimensional convolution was used. The data used in the experiment is time-series-based NGIDS (Next Generation IDS Dataset) data, and as a result of the experiment, an autoencoder that combines GRU and one-dimensional convolution is better than a model using a GRU-based autoencoder or a one-dimensional convolution-based autoencoder. It was efficient in terms of learning time for extracting useful latent patterns from training data, and showed stable performance with smaller fluctuations in anomaly detection performance.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.25 no.1
• /
• pp.63-69
• /
• 2015

There are various researches which proposed an automatic instance generation from freetext on the web. Existing researches that focused on English, adopts pattern representation which is generated by simple rules and regular expression. These simple patterns achieves high performance, but it is not suitable in Korean due to differences of characteristics between Korean and English. Thus, this paper proposes a novel method for generating patterns and instances which focuses on Korean. A proposed method generates high quality patterns by taking advantages of dependency relations in a target sentences. In addition, a proposed method overcome restrictions from high degree of freedom of word order in Korean by utilizing postposition and it identifies a subject and an object more reliably. In experiment results, a proposed method shows higher precision than baseline and it is implies that proposed approache is suitable for self-knowledge learning system.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.41 no.5
• /
• pp.61-70
• /
• 2004

The operation mode of neural network is divided into learning and recognition process. Learning is updating process of weight until neural network archives target result from input pattern. Recognition is arithmetic process of input pattern and weight. Traditional inner product process is focused to improve processing speed and hardware complexity. There is no hardware architecture to distinguish between loaming and recognition mode of neural network. In this paper we propose the new architecture of low power inner product processor for reconfigurable neural network. The proposed architecture is similar with bit-serial inner product processor on learning mode. It have several advantages which are fast processing base on bit-level, suitability of hardware implementation and pipeline architecture to compute data. And proposed architecture minimizes active units and reduces consumption power on recognition mode. Result of simulation shows that active units is depend on bit representation of weight, but we can reduce active units about 50 precent.

• The KIPS Transactions:PartB
• /
• v.11B no.4
• /
• pp.485-490
• /
• 2004

In data representation, the clustering performs a grouping process which combines given data into some similar clusters. The various similarity measures have been used in many researches. But, the validity of clustering results is subjective and ambiguous, because of difficulty and shortage about objective criterion of clustering. The fuzzy clustering provides a good method for subjective clustering problems. It performs clustering through the similarity matrix which has fuzzy membership value for assigning each object. In this paper, for objective fuzzy clustering, the clustering algorithm which joins principal components analysis as a dimension reduction model with bayesian learning as a statistical learning theory. For performance evaluation of proposed algorithm, Iris and Glass identification data from UCI Machine Learning repository are used. The experimental results shows a happy outcome of proposed model.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.8
• /
• pp.81-88
• /
• 2017

This paper proposes a method of detecting rain regions from a single image. More specifically, a way of training the deep convolutional neural network based on the collected rain and non-rain patches is presented in a supervised manner. It is also shown that the proposed rain detection method based on deep convolutional neural network can provide better performance than the conventional rain detection method based on dictionary learning. Moreover, it is confirmed that the application of the proposed rain detection for rain removal can lead to some improvement in detail representation on the low-frequency regions of the rain-removed images. Additionally, this paper introduces the rain transfer method that inserts rain patterns into original images, thereby producing rain effects on the resulting images. The proposed rain transfer method could be used to augment rain patterns while constructing rain database.

• The Journal of Korean Association of Computer Education
• /
• v.11 no.3
• /
• pp.43-55
• /
• 2008

This study investigates how learners extract their implicit knowledge into explicit form of the knowledge. The process of implicit-explicit transfer is known to help learners to reconstruct and refine their knowledge which was constructed before in some ways. Also we investigate which types of explicit form are more effective when it is delivered to other learners. In a classroom-based learning environment, students take educational content that is delivered by instructor and go through the process in which they try to fit the content into their cognitive structure by reconstructing the knowledge into their cognitive model. When they try to deliver their own cognitive model for the knowledge to other learners, they have to transform it into explicit form, and through the process, they reconstruct and refine the cognitive model of the knowledge, and find effective and appropriate way to express it. In this research, we experimented the process on a group of 77 college students and analyzed the results. We also did peer evaluated experiments to see which types of explicit format and factors are more effective than others. The results indicate that the types of explicit form of implicit knowledge play an important role in effectiveness of learning.