• Title/Summary/Keyword: 레이어

Search Result 743, Processing Time 0.031 seconds

A Configuration of the c/tensorflow framework for mixed precision on ARM-based embedded systems (ARM 기반 임베디드 시스템에서 mixed precision 을 위한 c/tensorflow 프레임워크 구성)

  • Lee, Jong-Eun;Lim, Seung-Ho
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2022.11a
    • /
    • pp.21-23
    • /
    • 2022
  • ARM 아키텍처를 사용하는 임베디드 시스템에서 int8, fp16, fp32 데이터를 조합하여 c/c++로 작성된 mixed precision CNN 을 실행시키기 위한 프레임워크 구성으로, 네트워크의 레이어마다 다른 정밀도를 사용하여 네트워크 경량화 및 추론 정확도 향상을 위한 최적의 설정을 탐색하는 실험 및 분석이 가능토록 하는 것을 목적으로 한다. 주요 구성은 network forwarding 중 레이어의 입력이 레이어에 설정된 정밀도와 다를 경우 실행되는 양자화/반양자화를 c/c++로 바인딩된 tensorflow 의 quantization 모듈을 사용하여 진행하고 ARM 시스템에서 c/c++의 fp16 을 사용하기 위해 fp16 를 컴파일이 가능한 ARM compiler 를 사용하는 프레임워크를 제안한다.

The composition and structure of Archival Information Packages(AIP) for a long-term preservation of electronic records (전자기록의 장기보존을 위한 보존정보패키지(AIP) 구성과 구조)

  • YIM, Jin-Hee
    • The Korean Journal of Archival Studies
    • /
    • no.13
    • /
    • pp.41-90
    • /
    • 2006
  • It's needed for the archivists to design a flexible and stable ERMS(Electronic Records Management System) which can ingest and store records through a consistent way and let users search and use records easily what they want. The basis of the design for ERMS are the conceptual composition and the logical and physical structure of the records when they are stored and managed in the ERMS. This paper explains the process of defining components and designing structure of electronic records using 3-layered approaches which consist of conceptual, logical and physical layer and shows advantages of this approaches. After benchmarking the information models of OAIS which is a reference model for the long-term preservation of digital information objects, this paper applies the model of AIP to a record as a 'Record AIP' and discusses the composition and structure of it. It's a critical task to identify mandatory or optional metadata groups which consists of the 'Record AIP's in the conceptual layer. This paper emphases that the metadata group related to services for the record information to users is required as a result of benchmarking OAIS information models. Various issues about the structure of 'Record AIP's are discussed according to the kind of preservation strategy such as migration or emulation and whether the encapsulation of records is required or not in the logical layer.

Illumination Compensation Algorithm based on Segmentation with Depth Information for Multi-view Image (깊이 정보를 이용한 영역분할 기반의 다시점 영상 조명보상 기법)

  • Kang, Keunho;Ko, Min Soo;Yoo, Jisang
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.17 no.4
    • /
    • pp.935-944
    • /
    • 2013
  • In this paper, a new illumination compensation algorithm by segmentation with depth information is proposed to improve the coding efficiency of multi-view images. In the proposed algorithm, a reference image is first segmented into several layers where each layer is composed of objects with a similar depth value. Then we separate objects from each other even in the same layer by labeling each separate region in the layered image. Then, the labeled reference depth image is converted to the position of the distortion image view by using 3D warping algorithm. Finally, we apply an illumination compensation algorithm to each of matched regions in the converted reference view and distorted view. The occlusion regions that occur by 3D warping are also compensated by a global compensation method. Through experimental results, we are able to confirm that the proposed algorithm has better performance to improve coding efficiency.

Log processing using messaging system in SSD Storage Tester (SSD Storage Tester에서 메시징 시스템을 이용한 로그 처리)

  • Nam, Ki-ahn;Kwon, Oh-young
    • Journal of the Korea Institute of Information and Communication Engineering
    • /
    • v.21 no.8
    • /
    • pp.1531-1539
    • /
    • 2017
  • The existing SSD storage tester processed logs in a 1-N structure between server and client using TCP and network file system. This method causes some problems for example, an increase in CPU usage and difficulty in exception handling, etc. In this paper, we implement a log processing message layer that can deal with asynchronous distributed processing using open source messaging system such as kafka, RabbitMQ and compare this layer with existing log transmission method. A log simulator was implemented to compare the transmission bandwidth and CPU usage. Test results show that the transmission using the message layer has higher performance than the transmission using the message layer, and the CPU usage does not show any significant difference The message layer can be implemented more easily than the conventional method and the efficiency is higher than that of the conventional method.

Realization and Test of the Transport Layer Protocol (전달 계층 프로토콜 구현 및 시험에 관한 연구)

  • 전동근;차균현
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.18 no.5
    • /
    • pp.650-662
    • /
    • 1993
  • The thesis describes the realization and test of the transport layer operating on the CLNS (Connection Less Network Service) which is specified by ISO 8073 and ADDENDUM 2 protocol. For realization of the transport layer, five modules and interface primitives were definde. This protocol was formally described by SDL (Specification and Description Language) and was programmed in C-language. For verification of the software, protocol tester was implemented and test scenarios were defined. The verification was accomplished based on the test scenarios. Transport Layer software presented in this thesis may be utilized in a real situation with some modifications. Protocol tester presented in this thesis can also be used for verification of other protocol softwares.

  • PDF

A Study of Schema Mapping Technique for Efficient Establishment of Geospatial DB (지형DB의 효율적 구축을 위한 스키마매핑 기법 연구)

  • Kang, Joon-Mook;Park, Joon-Kyu;Jo, Hyeon-Wook;Lee, Cheol-Hee
    • Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
    • /
    • v.26 no.1
    • /
    • pp.1-7
    • /
    • 2008
  • The government is spending a large amount of budget to establish NGIS due to the importance of geospatial information. This paper deals with the study of schema mapping technique which can efficiently establish military topographic DB forest layer using digital forest map and other existing geospatial DB. In order to accomplish this each schema of these types of DB was analyzed then, mapping table which connects related attributes was created and finally, military topographical DB forest layer of the area of interest was efficiently established. This research was able to display solutions that resolved compatibility issues between existing geospatial DB of these types, established by the schema mapping techniques described in this research.

A Study of Query Processing Model to applied Meta Rule in 4-Level Layer based on Hybrid Databases (하이브리드 데이터베이스 기반의 4단계 레이어 계층구조에서 메타규칙을 적용한 질의어 수행 모델에 관한 연구)

  • Oh, Ryum-Duck
    • Journal of the Korea Society of Computer and Information
    • /
    • v.14 no.6
    • /
    • pp.125-134
    • /
    • 2009
  • A biological data acquisition based on web has emerged as a powerful tool for allowing scientists to interactively view entries form different databases, and to navigate from one database to another molecular-biology database links. In this paper, the biological conceptual model is constructed hybrid biological data model to represent interesting entities in the data sources to applying navigation rule property for each biological data source based on four biological data integrating layers to control biological data. When some user's requests for application service are occurred, we can get the data from database and data source via web service. In this paper, we propose a query processing model and execution structure based on integrating data layers that can search information on biological data sources.

Design and Implementation of a Spatial-Operation-Trigger for Supporting the Integrity of Meet-Spatial-Objects (상접한 공간 객체의 무결성 지원을 위한 공간 연산 트리거의 설계 및 구현)

  • Ahn, Jun-Soon;Cho, Sook-Kyoung;Chung, Bo-Hung;Lee, Jae-Dong;Bae, Hae-Young
    • Journal of KIISE:Computing Practices and Letters
    • /
    • v.8 no.2
    • /
    • pp.127-140
    • /
    • 2002
  • In a spatial database system, the semantic integrity should be supported for maintaining the data consistency. In the real world, spatial objects In boundary layer should always meet neighbor objects, and they cannot hold the same name. This characteristic is an implied concept in real world. So, when this characteristic is disobeyed due to the update operations of spatial objects, it is necessary to maintain the integrity of a layer. In this thesis, we propose a spatial-operation-trigger for supporting the integrity of spatial objects. The proposed method is defined a spatial-operation-trigger based on SQL-3 and executed when the constraint condition is violated. A spatial-operation-trigger have the strategy of execution. Firstly, for one layer, the spatial and aspatial data triggers are executed respectively. Secondly, the aspatial data trigger for the other layers is executed. Spatial-operation-trigger for one layer checks whether the executed operation updates only spatial data, aspatial data, or both of them, and determines the execution strategy of a spatial-operation-trigger. Finally, the aspatial data trigger for the other layers is executed. A spatial-operation-trigger is executed in three steps for the semantic integrity of the meet-property of spatial objects. And, it provides the semantic integrity of spatial objects and the convenience for users using automatic correcting operation.

Development of Bio-Inspired System Software Architecture Based on Multi-Agents Framework (멀티 에이전트 프레임워크 기반의 생태계 모방 시스템 소프트웨어 아키텍처 개발)

  • Lee, Hak-Joo;Song, Ji-Young;Lee, Keun-Jun;Park, Sung-Yong;Jung, Sung-Won;Yang, Ji-Hoon;Nang, Jong-Ho
    • Proceedings of the Korean Society for Bioinformatics Conference
    • /
    • 2003.10a
    • /
    • pp.258-267
    • /
    • 2003
  • 네트워크의 발전으로 인해 현재 존재하는 소프트웨어 구조에 몇 가지 문제점을 내포하게 되었다. 이러한 환경에 대응하기 위해 생태계 구조를 모방한(Bio-inspired) 네트워크 기반적응 생존형 시스템을 제안한다. 여기서는 생태계의 여러 특성 중 적응성(adaptability) 확장성(scalability), 생존성(survivability)을 모델링 하고자한다. 이 시스템은 상기의 특성을 포함하는 몇 개의 계층으로 구성되어 있다. 독립된 개체의 역할을 하는 에코전트 레이어와 에코전트의 활동을 지원하는 에코전트 플랫폼 레이어, 효율적인 네트워크 활용을 위한 플랫폼 콜레버레이션 레이어로 이루어져 있다. 본 논문에서는 이러한 시스템의 구체적인 기능과 구성 그리고 이 시스템의 활용 분야에 대해 살펴본다.

  • PDF

An Efficient Flooding Algorithm with Selection of Retransmission Nodes (재전송 노드의 선택에 의한 효율적인 Flooding 알고리즘)

  • Choi, Seung-Joon;Yoo, Sang-Jo
    • Proceedings of the Korea Information Processing Society Conference
    • /
    • 2007.05a
    • /
    • pp.907-910
    • /
    • 2007
  • 본 논문에서는 Ad-hoc 네트워크에서 패킷 flooding에 의해 발생하게 되는 broadcast storm problem [1]을 해결하기 위한 크로스 레이어 기반의 flooding 기법을 제안한다. 제안된 flooding 기법은 크로스 레이어 프로토콜을 기반으로 하여 물리계층과 MAC (Medium Access Control) 계층에서 이웃 노드들에 대한 정보를 수집하고, 이를 이용하여 효율적인 재전송 노드의 수를 결정한다. 모의 실험을 통한 성능평가에서는 전체 네트워크 노드에 대한 수신 비율이 simple flooding 에 근접하며, 전송 비율 및 평균중복패킷 수에서도 좋은 성능을 보임으로서 broadcast storm problem을 해결할 수 있다.

  • PDF