• Korean journal of food and cookery science
• /
• v.23 no.5
• /
• pp.644-649
• /
• 2007

In this study a purified protein was prepared from pork meat. The product consisted of 0.5% moisture, 3.0% ash, 5.5% ether extract and 88.7% crude protein. Also, a meat oligopeptide mixture was prepared from a pepsin digest of the protein preparation. The two preparations were colorless and odorless powders with low fat contents. The nutritive values of the pork meat protein and oligopeptide mixture were estimated by two methods, one using biological value(BV) and the other employing net protein utilization(NPU) by the nitrogen balance method. The meat oligopeptide mixture showed an excellent nutritive value by both methods. The true digestibilites of both the pork meat protein and the oligopeptide mixture were more than 98%. The above results indicate that the oligopeptide mixture is an excellent material as a dietary nitrogen source for many purposes.

• Korean Journal of Food Science and Technology
• /
• v.8 no.2
• /
• pp.63-69
• /
• 1976

The changes in lysine availability during soybean fermentation were determined by the chemical analysis method as well as the biological methods with rat. The FDNB-reactive lysine determined by the difference (TLMI) method indicated that cooking and Meju fermentation reduced the lysine availability of soybean, but the subsequent ripening restored the availability to the same level of the raw soybean. On the other hand, the Biological Value, NPU, NER and the Relative lysine availability of the rat experiments showed a general decrease in the lysine availability of soybean during the ripening process as well as Meju fermentation.

• Journal of IKEEE
• /
• v.25 no.1
• /
• pp.229-233
• /
• 2021

NPU in an embedded environment performs deep learning algorithms with few hardware resources. By using a technique that reuses data, deep learning algorithms can be efficiently computed with fewer resources. In previous studies, data is reused using a shifter in ScratchPad for data reuse. However, as the ScratchPad's bandwidth increases, the shifter also consumes a lot of resources. Therefore, we present a data reuse technique using the Buffer Round Robin method. By using the Buffer Round Robin method presented in this paper, the chip area could be reduced by about 4.7% compared to the conventional method.

• Journal of Korean Society of Occupational and Environmental Hygiene
• /
• v.23 no.3
• /
• pp.212-221
• /
• 2013

Objectives: Asbestos fibers are accumulated in negative pressure unit (NPU) or vacuum cleaner in asbestos removal. Failure of operation or poor use performance of the NPU or vacuum cleaner causes asbestos fibers to spread in the air. Asbestos contractors should have an NPU fitted with a HEPA , vacuum cleaner, decontamination area, respirators, wetting equipment, and differential pressure manometer to register with the Ministry of Employment and Labor. There should be performance criteria of equipment used in asbestos removal. But there are none such criteria used in asbestos removal. This study was performed to investigate international or national performance criteria of equipment used in asbestos removal. Methods: Regulations, approved code of practice, guidance and national standards of the UK, USA and Korea were reviewed. A survey was done to investigate the requirements for equipment used in asbestos removal in Korea. Results: Air flow of NPUs used in Korea usually covers from$500m^3/h$ to over $3000m^3/h$. Some requirements for NPUs used in Korea were missing compared to the requirements of British standards for NPUs. All NPUs have different missing requirements. The UK also has separate British standard for leak tests for NPUs. Highly hazardous class vacuum cleaners should be used in asbestos removal in the UK. It has national standards on the operation of vacuum cleaners used in asbestos removal. There is only a certification system for less than 2.5kw rated voltage vacuum cleaners for home use in Korea. Powered (-assisted) respirators with mask are recommended in asbestos removal. Type 5 coveralls should be used in asbestos removal in the UK. There are international standards for requirements and leak tests of type 5 coveralls. A manometer for measuring differential pressure is needed in asbestos removal. The manometer's measuring range should cover less than ${\pm}125$ Pa and the gradation of manometer should be less than 2.5 Pa. The definition of decontamination area should be corrected. Installation of airlock and minimum area should be considered in decontamination area. Conclusions: Equipment should be used in asbestos removal proper performance and no leaks. There should be a certification system for equipment used in asbestos removal. This study can help to a certification system for equipment used in asbestos removal in Korea.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2003.10a
• /
• pp.661-664
• /
• 2003

Trade-off of sorority and speed always exists in the latest network environment. Recently, developed security processors is improved very performance, and sorority connection algorithms of a lot of part were embodied by hardware. This high speed security processor is essential ingredient in string network security solution equipment development that require very big band width. In this paper, we wish to describe about design and implementation of 10 Giga VPN equipments. In this system, embodied 10 Giga to use Cavium company's Nitrox-II processor, and supports two SP14-2 interface and PCI interface. All of the password algorithm that password algorithm that support is used in common use VPN equipment for compatibility with common use VPN equipment are supported and support SEED algorithm developed in domestic. Designed to support IPsec and SSL protocol, and supports all of In-Line structure that is profitable in high speed transaction and the Look-Aside structure that is profitable in practical use degree of NPU(Network Processor Unit).

• Korean Journal of Food Science and Technology
• /
• v.8 no.3
• /
• pp.121-128
• /
• 1976

Cooked soybean, soybean curd, soymilk residue and fermented soybean Meju products were prepared in the laboratory and the protein supplementary effect of these foods to the rice diet was determined by the rat feeding experiments. The soybean products providing 20% of the dietary protein were added to steamed rice and the protein digestibility (TD), Biological Value(BV), NPU, PER and Partial Carcass Nitrogen Value of the diets were measured. The protein supplementary effect of soybean products indicated that the non-fermented products generally improved the protein quality of the rice diet, whereas the fermented products did not but reduced it in some cases. There was observed a possible antinutritional effect for the diet supplemented with Home-made Meju. This growth retarding effect of Home-made Meju disappeared during the subsequent ripening in the brine for 6 months.

• Electronics and Telecommunications Trends
• /
• v.33 no.4
• /
• pp.14-22
• /
• 2018

Deep neural networks (DNNs) are widely used in various domains such as speech and image recognition. DNN software frameworks such as Tensorflow and Caffe contributed to the popularity of DNN because of their easy programming environment. In addition, many companies are developing neuromorphic processing units (NPU) such as Tensor Processing Units (TPUs) and Graphical Processing Units (GPUs) to improve the performance of DNN processing. However, there is a large gap between NPUs and DNN software frameworks due to the lack of framework support for various NPUs. A bridge for the gap is a DNN software platform including DNN optimized compilers and DNN libraries. In this paper, we review the technical trends of DNN software platforms.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.41 no.3
• /
• pp.281-291
• /
• 1992

This paper deals with the digital autopilot implementation for a launch vehicle. We propose a hardware and software system for digital autopilot implemented by microprocessor. The hardware system designed in this paper consists of CPU and memory board with 80286 MPU and 80287 NPU and I/O interface with A/D and D/A converters. The software system developed is composed of power-on self-test program, initializing program, interrupt service program, and control program. The performance of the overall system controlled by the digital autopilot implemented in this paper is evaluated via real-time simulations, which show that the control performances are satisfactory.

• Electronics and Telecommunications Trends
• /
• v.37 no.1
• /
• pp.53-62
• /
• 2022

With increasing size of transformer-based neural networks, a light-weight algorithm and efficient AI accelerator has been developed to train these huge networks in practical design time. In this article, we present a survey of state-of-the-art research on the low-precision computational algorithms especially for floating-point formats and their hardware accelerator. We describe the trends by focusing on the work of two leading research groups-IBM and Seoul National University-which have deep knowledge in both AI algorithm and hardware architecture. For the low-precision algorithm, we summarize two efficient floating-point formats (hybrid FP8 and radix-4 FP4) with accuracy-preserving algorithms for training on the main research stream. Moreover, we describe the AI processor architecture supporting the low-bit mixed precision computing unit including the integer engine.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.226-227
• /
• 2021

딥러닝을 활용한 영상 분석 기술은 GPU 하드웨어의 발전으로 인하여 소프트웨어 기반 처리 기술이 급격히 발전하였고 기존 패턴 분석 기술 대비 높은 정확도를 보여주고 있다. PC나 특정 하드웨어에서 동작하는 소프트웨어 기반 영상분석기술은 적용분야의 한계가 발생하였다. 신경망 기술을 하드웨어로 구현한 NPU(Network processing unit)의 개발로 고가의 플랫폼이 아닌 임베디드 플랫폼에서의 딥러닝 구현이 가능해졌다. 반면에 하드웨어에서 활용 가능한 네트워크가 제한적임으로 인하여 구현 가능한 딥러닝 모델의 크기, 메모리 등의 한계가 있으며 시시각각 변하는 딥러닝 기술에 기반한 최신모델 또는 고성능 모델을 구동하기에는 한계가 발생하였다. 이를 해결하기 위하여 본 연구에서는 Distillation 기법을 적용한 임베디드 시스템을 개발하고 이에 기반한 딥러닝 모델의 구현 및 상황에 따른 가변적 딥러닝 모델의 적용이 가능한 시스템을 구현하였다.