Seoul National University

About: Seoul National University is a education organization based out in Seoul, South Korea. It is known for research contribution in the topics: Population & Catalysis. The organization has 65879 authors who have published 138759 publications receiving 3715170 citations. The organization is also known as: SNU & Seoul-dae.

Structured Pruning of Deep Convolutional Neural Networks

Abstract: Real time application of deep learning algorithms is often hindered by high computational complexity and frequent memory accesses. Network pruning is a promising technique to solve this problem. However, pruning usually results in irregular network connections that not only demand extra representation efforts but also do not fit well on parallel computation. We introduce structured sparsity at various scales for convolutional neural networks, which are channel wise, kernel wise and intra kernel strided sparsity. This structured sparsity is very advantageous for direct computational resource savings on embedded computers, parallel computing environments and hardware based systems. To decide the importance of network connections and paths, the proposed method uses a particle filtering approach. The importance weight of each particle is assigned by computing the misclassification rate with corresponding connectivity pattern. The pruned network is re-trained to compensate for the losses due to pruning. While implementing convolutions as matrix products, we particularly show that intra kernel strided sparsity with a simple constraint can significantly reduce the size of kernel and feature map matrices. The pruned network is finally fixed point optimized with reduced word length precision. This results in significant reduction in the total storage size providing advantages for on-chip memory based implementations of deep neural networks.

Identification of heavy-flavour jets with the CMS detector in pp collisions at 13 TeV

Abstract: Many measurements and searches for physics beyond the standard model at the LHC rely on the efficient identification of heavy-flavour jets, i.e. jets originating from bottom or charm quarks. In this paper, the discriminating variables and the algorithms used for heavy-flavour jet identification during the first years of operation of the CMS experiment in proton-proton collisions at a centre-of-mass energy of 13 TeV, are presented. Heavy-flavour jet identification algorithms have been improved compared to those used previously at centre-of-mass energies of 7 and 8 TeV. For jets with transverse momenta in the range expected in simulated events, these new developments result in an efficiency of 68% for the correct identification of a b jet for a probability of 1% of misidentifying a light-flavour jet. The improvement in relative efficiency at this misidentification probability is about 15%, compared to previous CMS algorithms. In addition, for the first time algorithms have been developed to identify jets containing two b hadrons in Lorentz-boosted event topologies, as well as to tag c jets. The large data sample recorded in 2016 at a centre-of-mass energy of 13 TeV has also allowed the development of new methods to measure the efficiency and misidentification probability of heavy-flavour jet identification algorithms. The b jet identification efficiency is measured with a precision of a few per cent at moderate jet transverse momenta (between 30 and 300 GeV) and about 5% at the highest jet transverse momenta (between 500 and 1000 GeV).

Discrepancy between the tuberculin skin test and the whole-blood interferon gamma assay for the diagnosis of latent tuberculosis infection in an intermediate tuberculosis-burden country.

Abstract: ContextA recently developed whole-blood interferon γ (IFN-γ) assay based on stimulation with the Mycobacterium tuberculosis–specific antigens early secreted antigenic target 6 and culture filtrate protein 10 shows promise for the diagnosis of latent tuberculosis (TB) infection.ObjectiveTo compare the tuberculin skin test (TST) and the whole-blood IFN-γ assay in the diagnosis of latent TB infection according to the intensity of exposure.Design and SettingA prospective comparison between the whole-blood IFN-γ assay and the TST using a 2-TU dose of purified protein derivative RT23 in a population with intermediate TB burden was conducted sequentially between February 1, 2004, and February 28, 2005, in a Korean tertiary referral hospital.ParticipantsOf 273 participants, 220 (95.7%) had received BCG vaccine. Participants were grouped according to their risk of infection: group 1, no identifiable risk of M tuberculosis infection (n = 99); group 2, recent casual contacts (n = 72); group 3, recent close contacts (n = 48); group 4, bacteriologically or pathologically confirmed TB patients (n = 54).Main Outcome MeasuresLevels of agreement between the TST and the IFN-γ assay and the likelihood of infection in the various groups.ResultsFor the TST with a 10-mm induration cutoff, the positive response rate in group 1 was 51%; group 2, 60%; group 3, 71%, and group 4, 78%. For the IFN-γ assay, the positive response rate in group 1 was 4%; group 2, 10%; group 3, 44%; and group 4, 81%. The overall agreement between the TST and the IFN-γ assay in healthy volunteers was κ = 0.16. The odds of a positive test result per unit increase in exposure across the 4 groups increased by a factor of 5.31 (95% confidence interval [CI], 3.62-7.79) for the IFN-γ assay and by a factor of 1.52 (95% CI, 1.20-1.91) for the TST (P<.001). Using a 15-mm induration cutoff for the TST did not make a substantial difference to the test results.ConclusionThe IFN-γ assay is a better indicator of the risk of M tuberculosis infection than TST in a BCG-vaccinated population.

Allergic Rhinitis and its Impact on Asthma (ARIA): Achievements in 10 years and future needs

Abstract: Allergic rhinitis (AR) and asthma represent global health problems for all age groups. Asthma and rhinitis frequently coexist in the same subjects. Allergic Rhinitis and its Impact on Asthma (ARIA) was initiated during a World Health Organization workshop in 1999 (published in 2001). ARIA has reclassified AR as mild/moderate-severe and intermittent/persistent. This classification closely reflects patients' needs and underlines the close relationship between rhinitis and asthma. Patients, clinicians, and other health care professionals are confronted with various treatment choices for the management of AR. This contributes to considerable variation in clinical practice, and worldwide, patients, clinicians, and other health care professionals are faced with uncertainty about the relative merits and downsides of the various treatment options. In its 2010 Revision, ARIA developed clinical practice guidelines for the management of AR and asthma comorbidities based on the Grading of Recommendation, Assessment, Development and Evaluation (GRADE) system. ARIA is disseminated and implemented in more than 50 countries of the world. Ten years after the publication of the ARIA World Health Organization workshop report, it is important to make a summary of its achievements and identify the still unmet clinical, research, and implementation needs to strengthen the 2011 European Union Priority on allergy and asthma in children.