• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.12
• /
• pp.291-306
• /
• 2020

Nowadays, Data-Network-AI (DNA)-based intelligent services and applications have become a reality to provide a new dimension of services that improve the quality of life and productivity of businesses. Artificial intelligence (AI) can enhance the value of IoT data (data collected by IoT devices). The internet of things (IoT) promotes the learning and intelligence capability of AI. To extract insights from massive volume IoT data in real-time using deep learning, processing capability needs to happen in the IoT end devices where data is generated. However, deep learning requires a significant number of computational resources that may not be available at the IoT end devices. Such problems have been addressed by transporting bulks of data from the IoT end devices to the cloud datacenters for processing. But transferring IoT big data to the cloud incurs prohibitively high transmission delay and privacy issues which are a major concern. Edge computing, where distributed computing nodes are placed close to the IoT end devices, is a viable solution to meet the high computation and low-latency requirements and to preserve the privacy of users. This paper provides a comprehensive review of the current state of leveraging deep learning within edge computing to unleash the potential of IoT big data generated from IoT end devices. We believe that the revision will have a contribution to the development of DNA-based intelligent services and applications. It describes the different distributed training and inference architectures of deep learning models across multiple nodes of the edge computing platform. It also provides the different privacy-preserving approaches of deep learning on the edge computing environment and the various application domains where deep learning on the network edge can be useful. Finally, it discusses open issues and challenges leveraging deep learning within edge computing.

• The Journal of Internal Korean Medicine
• /
• v.43 no.4
• /
• pp.625-642
• /
• 2022

Objectives: Postoperative ileus (POI) is a common impairment of gastrointestinal motility and causes a delay in postoperative recovery, as well as an increased length of hospital stay, but no single strategy has a significant recuperative effect on POI. Studies of traditional Korean medicine (TKM) have reported improvements in bowel function after surgery. The aim of this systematic review was to assess the effectiveness of TKM on postoperative ileus in gastric cancer patients. Methods: We used six databases to search for studies published from January 1, 2007, until May 11, 2022. The included studies were those reporting gastric cancer patients who received TKM treatment after gastrectomy through indicators related to POI. Results: The search identified 27 RCTs that used herbal medicine (Daegeonjung-tang and Gami-leejoongtang), herbal medicine combined with acupuncture (Sama-tang, Gumiseunggi-tang, Daeseunggi-tang, and Insam-tang), acupuncture, acupuncture and moxibustion, electroacupuncture, warm needling, transcutaneous electroacupuncture (TEA), low-frequency electrical acupoint stimulation (LEAS), moxibustion, auricular acupressure, and ST-36 acupoint injection with neostigmine as treatments for POI. The time to first flatus was shortened by herbal medicine combined with warm needling (Sama-tang, Gumiseunggi-tang, Daeseunggi-tang, and Insam-tang), acupuncture, electroacupuncture, warm needling, TEA, moxibustion, auricular acupressure, and ST-36 acupoint injection with neostigmine (p<0.00001). The time to the first defecation decreased significantly in response to the herbal medicine combined with warm needling (Sama-tang, Gumiseunggi-tang, Daeseunggi-tang, and Insam-tang), acupuncture, electroacupuncture, warm needling, TEA, moxibustion, auricular acupressure, and ST-36 acupoint injection with neostigmine (p<0.00001). No serious adverse events occurred. Conclusions: TKM could be a promising option for preventing and resolving POI in gastric cancer patients after gastrectomy.

• Journal of the Korean Electrochemical Society
• /
• v.25 no.1
• /
• pp.42-49
• /
• 2022

In this study, the surface protection film based on organic-inorganic composite is manufactured for suppressing lithium dendrite growth, and the film is applied on the surface of Li metal negative electrode for lithium metal batteries (LMBs). The film is consist of the polyvinylidene fluoride (PVDF) polymeric binder which has good mechanical strength and high electrochemical stability, and carbon black (Super-P) which has outstanding electrical conductivity as the inorganic compound. First, in order to confirm the suppression of the internal short circuit by the lithium dendrite, the time required for the short circuit is measured while a constant current is continuously applied. As a result, the internal short circuit is delayed in proportion to the carbon black content of the film, and it is significantly delayed than bare Li metal electrode which does not use protection film. The cycle performance of the thick protection film (8 ㎛), is worse than that of the thin film (4 ㎛). However, as the carbon black content of the film increased, the cycle performance is improved. Thus, the surface protection film based on carbon black/PVDF composite can delay the internal short circuit, and has low overvoltage during the cycle. However, more stable cycle performance needs to be built through further improvements.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.5
• /
• pp.213-221
• /
• 2006

Phone segmentation of speech waveform is especially important for concatenative text to speech synthesis which uses segmented corpora for the construction of synthetic units. because the quality of synthesized speech depends critically on the accuracy of the segmentation. In the beginning. the phone segmentation was manually performed. but it brings the huge effort and the large time delay. HMM-based approaches adopted from automatic speech recognition are most widely used for automatic segmentation in speech synthesis, providing a consistent and accurate phone labeling scheme. Even the HMM-based approach has been successful, it may locate a phone boundary at a different position than expected. In this paper. we categorized adjacent phoneme pairs and analyzed the mismatches between hand-labeled transcriptions and HMM-based labels. Then we described the dominant error patterns that must be improved for the speech synthesis. For the experiment. hand labeled standard Korean speech DB from ETRI was used as a reference DB. Time difference larger than 20ms between hand-labeled phoneme boundary and auto-aligned boundary is treated as an automatic segmentation error. Our experimental results from female speaker revealed that plosive-vowel, affricate-vowel and vowel-liquid pairs showed high accuracies, 99%, 99.5% and 99% respectively. But stop-nasal, stop-liquid and nasal-liquid pairs showed very low accuracies, 45%, 50% and 55%. And these from male speaker revealed similar tendency.

• Journal of the Korea Society for Simulation
• /
• v.33 no.2
• /
• pp.1-11
• /
• 2024

A cyber-physical system is a technology that connects the physical systems of a manufacturing environment with a cyber space to enable simulation. One of the major challenges in this technology is the seamless communication between these two environments. In complex manufacturing processes, it is crucial to adapt to various protocols of manufacturing equipment and ensure the transmission and reception of a large volume of data without delays or errors. In this study, we propose a method for constructing agent models for real-time simulation-capable cyberphysical systems. To achieve this, we design data collection units as independent agent models and effectively integrate them with existing simulation tools to develop the overall system architecture. To validate the proposed structure and ensure reliability, we conducted empirical testing by integrating various equipment from a real-world smart microfactory system to assess the data collection capabilities. The experiments involved testing data delay and data gaps related to data collection cycles. As a result, the proposed approach demonstrates flexibility by enabling the application of various internal data collection methods and accommodating different data formats and communication protocols for various equipment with relatively low communication delays. Consequently, it is expected that this approach will promote innovation in the manufacturing industry, enhance production line efficiency, and contribute to cost savings in maintenance.

• Korean Journal of Environmental Biology
• /
• v.21 no.1
• /
• pp.72-76
• /
• 2003

We investigated the effects of molting-hormone insecticide tebufenozide on fourth larvae of the midge C. flaviplumuon in growth development from fourth-instar to adult. Fourth-instar larvae were exposed test concentrations were chosen control, 10 $\mu\textrm{g}$$L^{-1}$, 30 $\mu\textrm{g}$$L^{-1}$, 60 $\mu\textrm{g}$$L^{-1}$ and 100$\mu\textrm{g}$$L^{-1}$. In control and 10 $\mu\textrm{g}$$L^{-1}$, 10~20% individuals succeed adults while low emergence rates were observed surplus 30 $\mu\textrm{g}$$L^{-1}$. Death larvae appeared a various day along the concentration. We observed the converged day: control, 10 $\mu\textrm{g}$$L^{-1}$, 30 $\mu\textrm{g}$$L^{-1}$ and 60~100 $\mu\textrm{g}$$L^{-1}$ were day 4 to day 18, day 13 to day 16, day 9 to day 17 and day 9 to day 15, respectively. Therefore we concerned a tendency that exposure concentration was high, death period was short. Also, due to molting hormone disruption, development of midge was postponed relatively low concentration such as 10 $\mu\textrm{g}$$L^{-1}$. The converged days of pupa stage revealed control, 10 $\mu\textrm{g}$$L^{-1}$, 30 $\mu\textrm{g}$$L^{-1}$ and 60~100$\mu\textrm{g}$$L^{-1}$ were day 11~15, day 16~20, day 9~15 and day 5~9, respectively. About 10% individuals grow up the pupa-stage in 60~100 $\mu\textrm{g}$$L^{-1}$ condition. Sex ratio of the emerged adults was not different in this study. Death pupae frequently appeared in day 13~16 in control, day 6~23 in 10 $\mu\textrm{g}$$L^{-1}$, day 13~16 in 30 $\mu\textrm{g}$$L^{-1}$ and day 6~16 in 60~100 $\mu\textrm{g}$$L^{-1}$.

• Pediatric Infection and Vaccine
• /
• v.17 no.2
• /
• pp.91-100
• /
• 2010

Purpose : Recurrent otitis media (ROM) is usually defined as ${\geq}$3 distinct and well-documented episodes within 6 months or ${\geq}$4 episodes within 12 months. ROM is sufficiently important to warrant consideration of chemoprophylaxis, tympanostomy tube placement. There also is evidence that children with ROM are at risk for both hearing loss and speech delay. However, studies of ROM have been notably insufficient. In addition, even though environmental, racial, and sociocultural differences can affect risk factors, few studies have been conducted with regard to recurrent otitis media in Korea. Methods : This study was conducted from July 2009 to January 2010 with infants and children who were younger than 60 months old, who visited the out-patient clinics at Han-Il General Hospital and Kyunghee University Hospital. Data were collected by interview using a pre-formed sheet. Among a total of 892 infants and children, 457 were excluded, and the remaining 435 were allocated to 104 with ROM and 331 as a control group. Results : Attendance at daycare centers (P<0.001, OR=2.85), allergic rhinitis (P=0.026, OR=2.32), past history of bronchiolitis (P=0.003, OR=2.33), and low socioeconomic status (P=0.005, OR=2.00) were found to have a close significant correlation with ROM. Risk factors such as sex, having a sibling, breast-feeding, use of pacifiers, atopy, pneumococcal vaccination, influenza vaccination, smoking of parents, and indoor smoking are not relevant. Conclusion : Attendance at daycare centers, allergic rhinitis, past history of bronchiolitis, and low socioeconomic status have been identified as risk factors for ROM.

• Science of Emotion and Sensibility
• /
• v.13 no.4
• /
• pp.703-710
• /
• 2010

In this study, a tactile stimulator was developed to resolve some problems from the previous version of the system such as system configuration, inappropriate stimulation control and additional problems. The developed tactile stimulator consists of control unit, drive unit and vibrator unit. The control unit was controlled by E-Prime software to generate appropriate vibration pulses. The drive unit supplies enough energy to the vibrator to generate effective stimulation pulses. The vibrator unit consists of small coin type vibrator and velcro, and was made to be attached at the hand easily. The developed tactile stimulator was designed by small-size, light-weight, low-power, simple-fabrication, max 35 channels and little delay time from instruction signal of E-Prime software to vibrator. The duration and magnitude of stimulation was controlled by 10 grades and the problems concerning stimulation control were compensated by wideband frequency ranges. Additionally, the electrical safety was ensured by low voltage operation. Vibrator was made to be attached on finger as well as on any part of the subject. Since this tactile stimulator is developed based on E-Prime software which is widely used in cognitive science, it is believed that this stimulator be suitable for the wide application of cognitive science study.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.9
• /
• pp.63-73
• /
• 2015

This work proposes a 2-channel time-interleaved (T-I) 10b 120MS/s pipeline SAR ADC minimizing offset mismatch between channels without any calibration scheme. The proposed ADC employs a 2-channel SAR and T-I topology based on a 2-step pipeline ADC with 4b and 7b in the first and second stage for high conversion rate and low power consumption. Analog circuits such as comparator and residue amplifier are shared between channels to minimize power consumption, chip area, and offset mismatch which limits the ADC linearity in the conventional T-I architecture, without any calibration scheme. The TSPC D flip-flop with a short propagation delay and a small number of transistors is used in the SAR logic instead of the conventional static D flip-flop to achieve high-speed SAR operation as well as low power consumption and chip area. Three separate reference voltage drivers for 4b SAR, 7b SAR circuits and a single residue amplifier prevent undesirable disturbance among the reference voltages due to each different switching operation and minimize gain mismatch between channels. High-frequency clocks with a controllable duty cycle are generated on chip to eliminate the need of external complicated high-frequency clocks for SAR operation. The prototype ADC in a 45nm CMOS technology demonstrates a measured DNL and INL within 0.69LSB and 0.77LSB, with a maximum SNDR and SFDR of 50.9dB and 59.7dB at 120MS/s, respectively. The proposed ADC occupies an active die area of 0.36mm2 and consumes 8.8mW at a 1.1V supply voltage.

• KOREAN JOURNAL OF CROP SCIENCE
• /
• v.29 no.4
• /
• pp.334-341
• /
• 1984

Yield response of 30 rice varieties widely grown during the past 70 years in Korea was examined at the six transplanting dates from May 10 to June 29 at the 10 day intervals in 1982. Early japonica varieties grown before 1960s showed the highest yield at the June 9 and 19 transplantings and yields decreased with earlier or later transplantings except the earliest transplanting of May 10. However, yield of recent japonica varieties grown after 1960s was similar among the transplanting dates from May 10 to June 19, but yield was lower at the June 29 transplanting. In the indica/japonica varieties grown after mid- 1970s, yield was highest at the May 10, 20, and 30 transplantings and it decreased with further delay of transplantings largely due to the decreased panicle number, percent ripened grains, and 1,000 grain weight. This may indicate that recently developed varieties may be more adaptable toearly transplanting than older varieties. The number of panicles per hill was variable among the transplanting dates in all varieties, but it decreased as transplanting dates delayed after June 9. In japonica varieties the number of spikelets per panicle, percent ripened grains, and 1,000 grain weight increased as transplanting was delayed. However, in indica/japonica varieties the number of spikelets per panicle was similar among the transplanting dates and percent ripened grains and 1,000 grain weight decreased as transplanting date was delayed after June 9. The percent ripened grains of japonica varieties decreased with increase in the number of spikelets per hill but increased as heading date delayed. However, in indica/japonica varieties the percent ripened grains was not affected by both the number of spikelets per hill and heading date except that of June 29 transplanting of which percent ripened grains was extremely low because of delayed heading although the number of spikelets per hill was low. Grain/Straw ratio was not affected by heading date in japonica varieties, but it decreased significantly as heading date was delayed in indica/japonica varieties.