• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.21 no.6
• /
• pp.161-169
• /
• 2011

Recent intelligent video surveillance system asks for development of more advanced technology for analysis and recognition of video data. Especially, machine learning algorithm such as Support Vector Machine (SVM) is used in order to recognize objects in video. Because SVM training demands massive amount of computation, parallel processing technique is necessary to reduce the execution time effectively. In this paper, we propose a parallel processing method of SVM training with a multi-core processor. The results of parallel SVM on a 4-core processor show that our proposed method can reduce the execution time of the sequential training by a factor of 2.5.

• Journal of Korea Multimedia Society
• /
• v.21 no.2
• /
• pp.173-180
• /
• 2018

This paper proposes a method to improve the performance of face recognition via super-resolution method using sparse representation and deep learning from low-resolution facial images. Recently, there have been many researches on ultra-high-resolution images using deep learning techniques, but studies are still under way in real-time face recognition. In this paper, we combine the sparse representation and deep learning to generate super-resolution images to improve the performance of face recognition. We have also improved the processing speed by designing in parallel structure when applying sparse representation. Finally, experimental results show that the proposed method is superior to conventional methods on various images.

• Electronics and Telecommunications Trends
• /
• v.33 no.4
• /
• pp.1-13
• /
• 2018

Deep learning (DL) models have been widely applied to AI applications such image recognition and language translation with big data. Recently, DL models have becomes larger and more complicated, and have merged together. For the accelerated training of a large-scale deep learning model, model parallelism that partitions the model parameters for non-shared parallel access and updates across multiple machines was provided by a few distributed deep learning frameworks. Model parallelism as a training acceleration method, however, is not as commonly used as data parallelism owing to the difficulty of efficient model parallelism. This paper provides a comprehensive survey of the state of the art in model parallelism by comparing the implementation technologies in several deep learning frameworks that support model parallelism, and suggests a future research directions for improving model parallelism technology.

• Proceedings of the Korean Society for Cognitive Science Conference
• /
• 2010.05a
• /
• pp.60-65
• /
• 2010

Language learning involves linguistic environments around the learner. So the variation in training input to which the learner is exposed has been linked to their language learning. We explore how linguistic experiences can cause differences in learning linguistic structural features, as investigate in a probabilistic graphical model. We manipulate the amounts of training input, composed of natural linguistic data from animation videos for children, from holistic (one-word expression) to compositional (two- to six-word one) gradually. The recognition and generation of sentences are a "probabilistic" constraint satisfaction process which is based on massively parallel DNA chemistry. Random sentence generation tasks succeed when networks begin with limited sentential lengths and vocabulary sizes and gradually expand with larger ones, like children's cognitive development in learning. This model supports the suggestion that variations in early linguistic environments with developmental steps may be useful for facilitating language acquisition.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.2 no.2
• /
• pp.34-41
• /
• 1992

A robust and reliable learning and reasoning mechanism is addressed based upon fuzzy set theory and fuzzy associative memories. The mechanism stores a priori an initial knowledge base via approximate learning and utilizes this information for decision-making systems via fuzzy inferencing. We called this fuzzy computer architecture a 'fuzzy hypercube' processing all the rules in one clock period in parallel. Fuzzy hypercubes can be applied to control of a class of complex and highly nonlinear systems which suffer from vagueness uncertainty. Moreover, evidential aspects of a fuzzy hypercube are treated to assess the degree of certainty or reliability together with parameter sensitivity.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.45 no.3
• /
• pp.444-450
• /
• 1996

A digital hardware architecture for artificial neural network with learning capability is described in this paper. It is a modified hardware architecture known as HANNIBAL(Hardware Architecture for Neural Networks Implementing Back propagation Algorithm Learning). For implementing an efficient neural network hardware, we analyzed various type of multiplier which is major function block of neuro-processor cell. With this result, we design a efficient digital neural network hardware using serial/parallel multiplier, and test the operation. We also analyze the hardware efficiency with logic level simulation. (author). refs., figs., tabs.

• Journal of the Korean Society for Precision Engineering
• /
• v.16 no.1 s.94
• /
• pp.200-209
• /
• 1999

This paper presents a precise tracking control scheme for the proposed parallel robot using artificial neural network. This control scheme is composed of three feedback controllers and one feedforward controller. Conventional PD controller and artificial neural network are used as feedback and feedforward controller respectively. A backpropagation learning strategy is applied to the training of artificial neural network, and PD controller outputs are used as target outputs. The PD controllers are designed at the robot dynamics based on inter-relationship between active joints and moving platform. Feedback controllers insure the total stability of system, and feedforward controller generates the control signal for trajectory tracking. The precise tracking performance of proposed control scheme is proved by computer simulation.

• Journal of Distribution Science
• /
• v.15 no.8
• /
• pp.65-73
• /
• 2017

Purpose - Although e-learning has this advantage, currently many organizations have failed to recognize the necessity for basic e-learning educational training. It follows that practitioners working in the above organizations face the difficulty of having to find educational training processes of boosting their capabilities by themselves, rather than being able to utilize the educational training processes offered by e-learning. So of their own accord, learners have considered the necessity of information relating to being able to choose between high quality educational training processes. The purpose of this study is to propose opening e-learning content information for enabling an efficient choice of learning processes related to e-learning. Research design, data, and methodology - To pinpoint the items of e-learning content information, the study was initiated according to the following process. First, information relating to e-learning content (offered on e-learning websites) was researched. Second, based on the items of information which emerged from the research, selection and validity verification took place with 5 e-learning specialists as the subjects. Third, the opinions of adult learners at K University were collated relating to the items of information which emerged from the research. Results - The e-learning content information was comprised of 16 items in order to improve the choosing process for learner's e-learning contents. The analysis results showed that when learners were choosing e-learning processes, the most highly considered item was 'mobile support' (4.35). Following this (in order) were 'tuition fees' (4.30), 'certificate issuing' (4.23), and 'awareness of educational institution' (4.18). The least considered items were 'recruiting learners' (3.01) and 'tutor support' (3.18). Conclusions - The 16 items of e-learning content information in this study, were deemed to be helpful to learners in providing them with a choice of desirable e-learning process when this process was offered to them. Following this, there is a need for service institutions offering e-learning processes to make public the information suggested by this study. Research into educational methods additionally points to a necessity for not only e-learning forms, but also offline educational methods and a combination of blended learning to be offered and run parallel to e-learning.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.29B no.8
• /
• pp.1-6
• /
• 1992

This paper proposed a neural computer architecture for the learning of script character pattern recognition categories. Oriented filter with complex cells preprocess about the input script character, abstracts contour from the character. This contour normalized and inputed to the ART. Top-down attentional and matching mechanisms are critical in self-stabilizing of the code learning process. The architecture embodies a parallel search scheme that updates itself adaptively as the learning process unfolds. After learning ART self-stabilizes, recognition time does not grow as a function of code complexity. Vigilance level shows the similarity between learned patterns and new input patterns. This character recognition system is designed to adaptable. The simulation of this system showed satisfied result in the recognition of the hand written characters.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10b
• /
• pp.401-405
• /
• 1993

This paper presents a neuro adaptive control method for nonlinear dynamical systems based on artificial neural network systems. The proposed neuro adaptive controller consists of 3 layers artificial neural network system and parallel PD controller. At the early stage in learning or identification process of the system characteristics the PD controller works mainly in order to compensate for the inadequacy of the learning process and then gradually the neuro contrller begins to work instead of the PD controller after the learning process has proceeded. From the simulation studies the neuro adaptive controller is seen to be robust and works effectively for nonlinear dynamical systems from a practical applicational points of view.