• Journal of the Korea Society of Computer and Information
• /
• v.19 no.6
• /
• pp.19-26
• /
• 2014

An ensemble classifier system is a widely-used multi-classifier system, which combines the results from each classifier and, as a result, achieves better classification result than any single classifier used. Several methods have been used to build an ensemble classifier including boosting, which is a cascade method where misclassified examples in previous stage are used to boost the performance in current stage. Boosting is, however, a serial method which does not form a complete feedback loop. In this paper, proposed is context sensitive SVM ensemble (CASE) which adopts SVM, one of the best classifiers in term of classification rate, as a basic classifier and clustering method to divide feature space into contexts. As CASE divides feature space and trains SVMs simultaneously, the result from one component can be applied to the other and CASE achieves better result than boosting. Experimental results prove the usefulness of the proposed method.

• Journal of Institute of Control, Robotics and Systems
• /
• v.15 no.5
• /
• pp.457-464
• /
• 2009

The weak classifier of AdaBoost algorithm is a central classification element that uses a single criterion separating positive and negative learning candidates. Finding the best criterion to separate two feature distributions influences learning capacity of the algorithm. A common way to classify the distributions is to use the mean value of the features. However, positive and negative distributions of Haar-like feature as an image descriptor are hard to classify by a single threshold. The poor classification ability of the single threshold also increases the number of boosting operations, and finally results in a poor classifier. This paper proposes a weak classifier that uses multiple criterions by adding a probabilistic criterion of the positive candidate distribution with the conventional mean classifier: the positive distribution has low variation and the values are closer to the mean while the negative distribution has large variation and values are widely spread. The difference in the variance for the positive and negative distributions is used as an additional criterion. In the learning procedure, we use a new classifier that provides a better classifier between them by selective switching between the mean and standard deviation. We call this new type of combined classifier the "Mixed Weak Classifier". The proposed weak classifier is more robust than the mean classifier alone and decreases the number of boosting operations to be converged.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.28 no.3
• /
• pp.349-358
• /
• 2004

A low pressure impactor is an impaction device to separate airborne particles into aerodynamic size classes at low pressure condition. We designed a two-stage low-pressure impactor to classify submicron sized environmental aerosols. Performance evaluation was carried out for stages 1 and 2 by using an electrical method. Monodisperse liquid dioctyl sebacate (DOS) particles were generated using evaporation-condensation process followed by electrostatic classification using a DMA (differential mobility analyzer). The test particles were in the range of 0.08∼0.8$\mu\textrm{m}$. For the evaluation of the impactor we used two electrometers; one was connected to the impaction plate of the impactor and the other was to the Faraday cage used as a backup filter. The effect of polydispersity of test aerosols on the performance was investigated. The results showed that the experimental 50-% cutoff diameters at each impactor's operation pressure were 0.53 and 0.187$\mu\textrm{m}$ for stages 1 and stage 2, respectively. The effects of operation pressure on the cutoff diameter and the steepness of collection efficiency curves were also investigated.

• Environmental Analysis Health and Toxicology
• /
• v.13 no.1_2
• /
• pp.43-53
• /
• 1998

Ambient air levels of polycyclic aromatic hydrocarbons(PAHs) are of concern because of their potential for adverse health effects including transformation of some of these substances to mutagens and carcinogens by mammalian microsomal enzyme system. Airbone particulate samples were collected by a conventional high-volume sampler and by an Anderson cascade impactor on 2 to 3 days in each month over a period of 1 year at a representative site of the heavy traffic area of Seoul from beptember 1994 to August 1995. Ten individual PAHs in sizable air particulates of each stage of two months were separated and analyzed by gas chromatography/mass spectrometry. As a results of analysis, the gross concentrations of PAHs in the fine and coarse particles were higher in the winter month than in the spring, followed in descending order by in the fall and summer. In a study of dependency of 10 PAHs compounds on size distribution of particles at heavy traffic area found that about 85% of the total PAHs content was associated with particles less than 2.0um (fine particles) in diameter of winter sampling period. while 79% were associated with this size fraction during summer period. In according to the mean concentrations of the 10 PAHs in 7 size classification from < 0.38 to> 10.1, the 'size was the smaller, PAHs concentration was the higher. Thus it was found that PAHs concentration was greatly affected by air particle size. Annual mean benzo(a)pyrene equivalents was 5.88ng/m$^3$ and obtained by applying, toxic equivalency factor developed by Nisbet and Lagoy.

• Journal of Environmental Health Sciences
• /
• v.28 no.1
• /
• pp.85-92
• /
• 2002

A measurement of indoor microorganism using Reuter Centrifugal Air Sampler(RCS) was undertaken during October 1991 - February 1999 and 6-Stage Cascade Air Sampler was undertaken during May 2001 - June 2001 in Seoul. Sites including book store, department store, theater, subway station, underground shopping center, hospital, office building, sports facility, and eduationa institutio were chosen to measure indoor microorganism. The results were as follows: 1. The average of total microorganism collected on the agar strip GK-A media were, in the order, subway station, hospital, underground shopping center, department store, book store, theater, sports facility, educational institution, office building in sites. The highest concentration of 711cfu/m$^3$ was found in the subway station and the lowest concentration of 44cfu/m$^3$ was found in office building. 2. The average of staphylococci collected on the agar strip GK-S media, in the order, were subway station, underground shopping center, hospital and department store, department store, theater, office building, sports facility and educational institution in sites. The highest concentration of 502cfu/m$^3$ was found in the subway station and the lowest concentration of 14cfu/m$^3$ was found in sports facility and educational institution. 3. The average of fungus collected on the agar strip GK-HS media, in the order, were underground hospital, shopping center, theater, subway station, department store, book store, sports facility, educational institution, and office building in sites. The highest concentration of 252cfu/m$^3$ was found in the hospital and the lowest concentration of 32cfu/m$^3$ was found in office building. 4. Ratio of Indoor/Outdoor, determined by site was 1.12-2.38 in total count, 1.00-2.35 in staphylococci, and 0.99-1.34 in fungus. 5. The positive results of test were 12-24% in indoor and 9-43% in outdoor. 6. By gram staining gram positive cocci were 59.9%, gram positive bacill 24.4%, gram negative bacilli 10.4%, and gram negative cocci 0.5%.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.2
• /
• pp.112-118
• /
• 2009

This paper proposes a rapid detection algorithm of a salient region for music video browsing system, which can be applied to mobile device and digital video recorder (DVR). The input music video is decomposed into the music and video tracks. For the music track, the music highlight including musical chorus is detected based on structure analysis using energy-based peak position detection. Using the emotional models generated by SVM-AdaBoost learning algorithm, the music signal of the music videos is classified into one of the predefined emotional classes of the music automatically. For the video track, the face scene including the singer or actor/actress is detected based on a boosted cascade of simple features. Finally, the salient region is generated based on the alignment of boundaries of the music highlight and the visual face scene. First, the users select their favorite music videos from various music videos in the mobile devices or DVR with the information of a music video's emotion and thereafter they can browse the salient region with a length of 30-seconds using the proposed algorithm quickly. A mean opinion score (MOS) test with a database of 200 music videos is conducted to compare the detected salient region with the predefined manual part. The MOS test results show that the detected salient region using the proposed method performed much better than the predefined manual part without audiovisual processing.