• Journal of Digital Contents Society
• /
• v.16 no.5
• /
• pp.671-680
• /
• 2015

The Joint Bayesian[1] method was published in 2012. Since then, it has been used for binary classification in almost all state-of-the-art face recognition methods. However, no improved methods have been published so far except 2D-JB[2]. In this paper we propose an improved version of the JB method that considers the features of both the given face image and its mirror image. In pattern classification, it is very likely to make a mistake when the value of the decision function is close to the decision boundary or the threshold. By making the value of the decision function far from the decision boundary, the proposed method reduces the errors. The experimental results show that the proposed method outperforms the JB and 2D-JB methods by more than 1% in the challenging LFW DB. Many state-of-the-art methods required tons of training data to improve 1% in the LFW DB, but the proposed method can make it in an easy way.

• Atmosphere
• /
• v.26 no.1
• /
• pp.73-84
• /
• 2016

A method to make ice spike using home refrigerator with ice tray was found. Many experiments have carried out with this method and many natural phenomena occurring on the formation of ice spike are found. A new concept of the Latter Freezing Water (LFW) was imported to explain the ice spike formation. At LFW position on water surface, the Sprout of Super cooled Water (SSW) grows by the Volume Expansion Effect (VEE) caused by the phase change of water in water. And air bubbles that are expelled from ice during freezing process, gather, rise, and detonate at the upper most part of SSW that make SSW freeze and grow upward with the water pipe in it. Together with VEE the capillarity in the water pipe makes the column grow more, that makes the ice spike. Many other findings were succeeded; 1) Ice spike process is completed before the whole water freezes. 2) If water is corrupted or shocked, even though it is very slight, ice spike is not generated. 3) Rain water contains the most LFW among all kind of waters used in experiments. 4) LFW is changed into normal water after passing the ice spike. 5) A new concept of the ice bullet is introduced. 6) The reason of frequent occurrences of the ice spike at Mt. Mai is investigated also.

• Journal of Information Processing Systems
• /
• v.12 no.3
• /
• pp.381-391
• /
• 2016

The Joint Bayesian (JB) method has been used in most state-of-the-art methods for face verification. However, since the publication of the original JB method in 2012, no improved verification method has been proposed. A lot of studies on face verification have been focused on extracting good features to improve the performance in the challenging Labeled Faces in the Wild (LFW) database. In this paper, we propose an improved version of the JB method, called the two-dimensional Joint Bayesian (2D-JB) method. It is very simple but effective in both the training and test phases. We separated two symmetric terms from the three terms of the JB log likelihood ratio function. Using the two terms as a two-dimensional vector, we learned a decision line to classify same and not-same cases. Our experimental results show that the proposed 2D-JB method significantly outperforms the original JB method by more than 1% in the LFW database.

• Journal of the Korean Society of Tobacco Science
• /
• v.11 no.1
• /
• pp.49-64
• /
• 1989

The study was carried out to investigate the effect of control leaf number on growth and chemical characteristics, and on yield and quality In (rue-cured tobacco The rest upper leaves of 14, 16, and 18 leaves per plant were removed at button stage, and the plants were harvested :0, 12, and 16 leaves with discard 4 leaves of the bottom, respectively. 1. There were showed a tendency to increase in length (LL), width (LW), area (LA), fresh (LFW) and dry weight (LDW) per leaf as leaf numbers were decreased, while those per plant and the ratios of increase of those per plant were decreased. 2. In compensation effect (CE), there were increase in LL, LFW, LDW, LL/LW and LDW/LFW with each decreased in leaf numbers while LW was decreased, LA unchanged. 3. There were significant Increases in LL, LW, and LA at 5th~8th stalk positions form bottom (SP), while those .at 13th~16th SF were decreased. Also, LDW at 5th~l0th, and LDW/LFW and weight Per unit leaf area at 11th 16th were Increased as leaf numbers decreased. 4. Mean per plant and CE of nicotine were increased, but those of sugar were decreased with each decrease In leaf numbers. In addition, nicotine contents at 5th~14th SP except for 9th~10th were increased, otherwise sugar contents at 7th~14th were decreased. 5. There were showed a tendency to increase in grade per Plant and in tirade at 5th~8th SP, but decrease in CE of grade as leaf numbers were decreased. There wee significant decrease in grade at 11th~14th SP. Also, ratio of 5 grade was decreased, while ratios of 3 and 4 grades were increased. 6. When the 2 leaves at 3rd~4th SP were harvested, price was decrease 2.6% and yield was increased 5.8 %. 7. When the 6 leaves from convention were removed at but ton stage, effects of apparent, compensational, and net per leaf on yield were -3.9%, +2.6%, and -6.5%, respectively, and those on price were +1.1%, -1.6% and +2.7%, respectively.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.22 no.1
• /
• pp.81-89
• /
• 2018

Austenitic stainless steels have excellent corrosion resistance, durability and fire resistance. Especially, since STS304L among austenitic types is a low-carbon variation of STS304 and has excellent intergranular corrosion resistance, it can often be used under the welded condition without heat treatment after field welding. This paper investigated ultimate behaviors such as ultimate strength and weld metal fracture mechanism of STS304L fillet-welded connections with TIG(tungsten inert gas) welding through test results. Main variables of specimens are weld length and welding direction against loading. Fracture of specimens are classified into three modes(tensile fracture, shear fracture and block shear fracture). Ultimate strengths were compared according to the welding direction and weld length and TFW series with transverse fillet weld had the highest strength compared with other types(LFW series with longitudinal fillet weld and FW series with all round weld). It is known that current design specifications such as KBC 2016 and AISC2010 underestimated the strength of TFW and LFW specimens and provided unconservative estimates for FW specimens. Finally, strength equations were proposed considering material properties of STS 304L material.

• The Optical Journal
• /
• s.105
• /
• pp.21-23
• /
• 2006

미국 달러의 강세로 인해 유럽 및 아시아는 다시금 경제적인 안정과 성장을 달성하고 있는 가운데 중국은 저가 마킹 장비의 주도로 두자리수의 성장을 지속하고 일본 또한 2005년에 는 2% 성장을 이루는 등 아시아 시장의 성장이 눈에 띈다. 반면 유럽은 공작기계분야에서 2005년 3% 성장을 기록하는데 그친 것으로 나타났다. 본 고는 세계 레이저 시장중 다이오드 분야를 제외한 시장에 대해 언급하고 있다. <출처:LaserFocusWorld- LASER MARKETPLACE 2006: Market's messagrs are mixed(http://lfw.pennnet.com)>

• Journal of Information Processing Systems
• /
• v.14 no.6
• /
• pp.1464-1479
• /
• 2018

In face recognition, sometimes the number of available training samples for single category is insufficient. Therefore, the performances of models trained by convolutional neural network are not ideal. The small sample face recognition algorithm based on novel Siamese network is proposed in this paper, which doesn't need rich samples for training. The algorithm designs and realizes a new Siamese network model, SiameseFacel, which uses pairs of face images as inputs and maps them to target space so that the $L_2$ norm distance in target space can represent the semantic distance in input space. The mapping is represented by the neural network in supervised learning. Moreover, a more lightweight Siamese network model, SiameseFace2, is designed to reduce the network parameters without losing accuracy. We also present a new method to generate training data and expand the number of training samples for single category in AR and labeled faces in the wild (LFW) datasets, which improves the recognition accuracy of the models. Four loss functions are adopted to carry out experiments on AR and LFW datasets. The results show that the contrastive loss function combined with new Siamese network model in this paper can effectively improve the accuracy of face recognition.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.1
• /
• pp.50-56
• /
• 2020

With the rapid development of deep neural network theory and application research, the effect of face detection has been improved. However, due to the complexity of deep neural network calculation and the high complexity of the detection environment, how to detect face quickly and accurately becomes the main problem. This paper is based on the relatively simple model of the MTCNN model, using FDDB (Face Detection Dataset and Benchmark Homepage), LFW (Field Label Face) and FaceScrub public datasets as training samples. At the same time of sorting out and introducing MTCNN(Multi-Task Cascaded Convolutional Neural Network) model, it explores how to improve training speed and Increase performance at the same time. In this paper, the dynamic image pyramid technology is used to replace the traditional image pyramid technology to segment samples, and OHEM (the online hard example mine) function in MTCNN model is deleted in training, so as to improve the training speed.

• Tribology and Lubricants
• /
• v.14 no.4
• /
• pp.72-78
• /
• 1998

In this work, the effects of binder materials on the tribological behavior were experimentally evaluated for air curable MoS$_2$bonded film lubricants. The binders tested were basically alkyd and acryl based resins, and also were modified by nitro-cellulose in various weight percentages. Effects on the binder ratio to solid lubricants and the molecular weight of binders were also investigated in all aspects of the tribological behavior. For the tests, LFW-1 and Falex tester were used to measure mainly the endurance life and load carrying capacity of bonded lubricants. Results showed that lubricants of methacrylic resin has the better performance than those of other resins, and also the properties of lubricants both of alkyd and acrylic resins could be improved by modification with nitro-cellulose. It was also shown that the optimum ratio of nitro-cellulose related closely to the binder material and the testing machine.

• Proceedings of the Korean Society of Tribologists and Lubrication Engineers Conference
• /
• 1997.04a
• /
• pp.232-241
• /
• 1997

The tribological behavior for MoS$_2$ bonded films is evaluated according to the international standard testing methods, such as ASTM D 2625(Falex tester) and ASTM D 2714(LFW-1 tester). It has been well known that the tribological behavior for MoS$_2$ bonded films is affected by several factors. However, in this work, the wear life for MoS$_2$ bonded films is mainly experimentally measured with different contact geometry, and evaluated in terms of the frictional heating according to the contact geometry of tribotester. The test results show that the wear life of MoS$_2$ bonded films is significantly affected not only by the frictional heating, but also by the contact pressure, test running-in conditions, and the contact conformity.