메뉴 바로가기 컨텐츠 바로가기

PROGRAM

9:00 - 10:00

Registration

Tower A, 2F

10:00 - 12:00

Keynote

Tower A Creative Hall

12:30 - 13:30

Lunch

Tower A, 1B

13:30 - 14:15

Pick-it, AI-powered 3D vision software that gives eyes to robots.

Tower A Creative Hall

Smart factory and collaborative robots have led to innovations in the European manufacturing industry. Artificial Intelligence and 3D vision are key technologies that allow robots to see and think to work flexibly and in collaboration with people. This presentation introduces a Belgian startup company Pick-it N.V. and its software product, Pick-it that gives eyes to robots and makes robot vision easy. We will see how open-source and agile software development cultures can make major changes to the traditional and conservative manufacturing industry in Europe, and together think how it can have positive effects, not only on the industrial but also cultural revolution in Korea.

Robot

Tower A Creative Hall

Seongyong Koo

Pick-it N.V.

Pick-it is 3D vision software for collaborative robots. It was founded in 2016 as a product spinout from Intermodalics, a Belgian startup that develops software for robots. Seongyong Koo joined the Pick-it development team as a senior computer vision engineer in September 2017 and is now a country director for the Korean market. Prior to moving to Belgium, he had developed a deep learning based robot vision system as a senior scientist at the Institute of computer science VI at University of Bonn and a postdoc researcher at Technical University of Munich, Germany.

Open Source Based 5G SDN/NFV platform with a focus on ONOS

Tower D Frontier Hall

We would like to introduce the open-source-based SDN software-defined networking) and NFV (network function virtualization) developed by the Network Business Department for 5G systems, with an emphasis on ONOS (Open Network Operating System) controllers. We will discuss things to consider in the development of open-source-based commercial solutions. We also wish to explore the differences between development projects based on Telco Open Source and on conventional open source, and develop a common ground for the culture and the strategy of open source contribution. Finally, we hope that this session will provide a forum to introduce the SDN/NFV platform for 5G developed by the department, and to share and discuss the direction of future development.

Cloud

Tower D Frontier Hall

Kibeom Park

Samsung Electronics

At the Network Business Department, Kibeom Park manages the development of software-defined network (SDN) and network slicing based on the open network operating system (ONOS).

- Software architect at Samsung Electronics (2013~)

- A speaker at the Samsung Electronics Software Conference (2013)

- A speaker at the third SOSCON (2016)

- Presented exhibits at Mobile World Congress 2017

- A host and a speaker at ONOS Open Source Conference: ONOS Build 2017

- Presented exhibits at the fourth SOSCON – joint booth exhibition with ONF (SDN open source group) (2017)

- A speaker at KRNet 2018

Attack and Defense on Linux kernel

Tower D Conference room 1F

Linux kernel is one of most widely-used kernel over the world. So that Linux has been being an attractive target for attacker. In this presentation, explain which attacks has been on Linux kernel, and which defenses are proposed for stopping the attacks.

Security

Tower D Conference room 1F

Jinbum Park

Samsung Electronics

I'm working at SAMSUNG Research, Security Team. My major job is developing a security solution for product, and writing a paper. My interested area is Linux kernel security, and Side channel attack.

14:30 - 15:15

Deep learning use cases from the sun to the cells

Tower A Creative Hall

There are many reasons why deep learning technique is spreading across science and industry, but it's probably because performance is good with a data-driven learning model. That is, if you just train a model that is performing well in a particular field, you can use it in other areas. Also, new models are updating the accurate records of existing models, revealing the source code along with the paper. So if you have data, can you use magic models right away? However, there are many things to consider when applying the deep learning model in fields that are familiar with traditional methods. Let's look at what challenges we're solving, from replacing existing algorithms to defining new problems. Second, do you really need deep learning on the problem you want to solve? Deep Learning models require a lot of data and a lot of computing resources. Is it necessary to invest in these costs and apply the Deep Learning model? Let's look at the effectiveness of the various problems. Third, what are the things to prepare to apply Deep Learning to real problems? Training a deep learning model requires a variety of environments and tools, such as building datasets, learning, and verification environments. What's the promise before you set out on a long journey that takes considerable time and effort from building large datasets to tuning hyperparameters? I hope you get insight by exploring various attempts and considerations to apply artificial intelligence to fields from the sun to the cells.

AI

Tower A Creative Hall

Taeyoung Kim

InSpace Co., Ltd.

I wrote 'Python deep learning Keras with block' and run 'Kim Taeyoung Keras Blog', 'Keras Korea' and 'Kaggle Korea' in order to help beginners study deep learning using Keras, an intuitive and concise deep learning library. In addition, as CTO at InSpace, I am active in applying deep learning model and deep reinforcement learning agent to medical, drone, satellite and space field.


[Seminar]

- 18.6 '1st deep learning conference all together'

- 18.3 Yangjae R&CD innovation hub, 제4회 AI신기술뎐 “Deep learning Keras all together'

- 18.2 Microsoft, 7th Korea Community Day “Deep learning tools in everyone's hands...'

- 17.12 Naver D2, 'Keras with block and deep learning application from the sun to the cells'

- 17.11 Google, GDG DevFest 2017 Seoul 'Python deep learning Keras with block'

- 17.11 NASA, NASA Ames Conference Center, Radiation Workshop Lighting Talk

Use of Korean language search function in Elasticsearch

Tower D Frontier Hall

This session examines the text analysis feature of the Elasticsearch search engine and the process it uses to search Korean words. It explores Nori, a Korean morphological analyzer developed by Elastic, and discusses the methods used to reduce the file size and enhance the functions of the previous community morphological analyzers.

Big Data

Tower D Frontier Hall

Jongmin Kim

Elastic

Jongmin Kim is a community engineer at Elastic. He runs an Elasticsearch technology community, and communicates with developers through various publications, seminars, and presentations.

Edge Computing and Its future transformation

Tower D Conference room 1F

In this session, Edge Computing and its transformation in near future will be explained. In current scenario, with lot of data and security concerns, Edge became important driving factor for industry to depend on. Edge not only solve the latency for near real-time applications but also avoid unnecessary data transmission to cloud as well as protecting the information from 3rd party access. While utilizing the need for Edge, Edge Computing can be utilized in intelligent way to do on premises analytics. By leveraging current open sources of cloud technologies, edge can become intelligent gateway , data storage and self-evolving machine for confidential data for smart home, smart factory and smart city solutions. Edge not only reduces the cost of deployment in cloud but also provide near real-time values for mission critical operations..

Edge Computing

Tower D Conference room 1F

Ashok Babu Channa

Samsung Electronics

Ashok Babu Channa is a Software Architect at Samsung R&D, Bangalore and works for IoT and Edge platform. His team involved in Edge Computing Activates for Edge X Foundry contributions for Smart Factory solution. He involved in Real-Time messaging framework which provide Event driven architecture between micro services. He also currently holding the maintainer position under Discovery and Connectivity working group in IoTivity Open Source. He has been part of multiple interesting projects like IoTivity, Things Platform, UPnP, DLNA, Web services and streaming solutions . He received a Masters from BITS Pilani, India in Computer Engineering. He has few publications and patents in convergence domain.

15:30 - 16:15

How to contribute Apache MXNet, Emerging Deep Learning Framework

Tower A Creative Hall

Apache MXNet is a fast-growing, deep running open source framework that has gained momentum in the cloud, edge devices, and mobile industries. More than 500 contributors from Amazon, NVIDIA, Intel, and Samsung are working in the MXNet community. This session introduces features of Apache MXNet, experiences and methods of code contribution for university students, data scientists and developers interested in artificial intelligence and machine learning.

AI

Tower A Creative Hall

Channy Yun

Amazon Web Services

Channy Yun is a Principal Technical Evangelist at Amazon Web Services and works with Korean developers to enable them to use AWS cloud services. He has over 20 years of experience in various information systems such as geospatial, content management, e-commerce platform, and building their large-scale API systems such Daum Search and Maps APIs. Now he has special interests in cloud native architectures such as serverless, container, mobile/IoT, and machine learning.

Kyusam Oh

Samsung SDS

Kyusam Oh, works for Samsung SDS Research Center, is a Data Sceintist and Deep Learning Engineer. He has wide variety of experiences with analytics and deep learning projects such as forecast, security, anomaly detection, genomics and biomedical signal analytics. He recently has shown interest in solving real world problem using recurrent neural network which is known as a powerful technique for sequential data.

2-Ways building HA Kubernetes clusters: Kubespray and kubeadm

Tower D Frontier Hall

How can I build a self-hosted Kubernetes cluster in a production environment? In this session we will compare Kubespray with kubeadm as a way to build a 3-Masters Kubernetes cluster. For Kubespray, I explain the source directory structure, important variables, and how to install modules selectively, etc. For kubeadm, I explain pre-requisites, how to build 3 masters with self-hosting, and how to add nodes, etc.

Cloud

Tower D Frontier Hall

Seungkyu Ahn

SK Telecom

I'm working at SKT and developed Java-based Enterprise Application before 2011. Since then, I have been building a private cloud using OpenStack and developing Kubernetes based service. I've been registered in CNCF Community Speaker's bureau and I'm a leader of the Kubernetes Korea Group.

Have a fun with 'business-card-sized' computers!

Tower D Conference room 1F

There are several name-card-sized computer lately. Raspberry Pi is the most famous, ODROID series is the famous in Korea, Banana Pi and Tinkerboard are made by x64 PC vendors, 96 boards series are high-end board supported by Linaro, PYNQ and Parallela make FPGA easier.it is not only for hobby but also for education and industrial use, lately I oT/Edge computing and education use at elementary/junior high school and college are quite increasing. In this session I will talk the name-card-sized computer histories - where it comes from, how the it grows with use cases and what it is now and discuss several use cases and tips at hobby, education, industrial business, IoT and etc.

IoT

Tower D Conference room 1F

Masafumi

Japanese Raspberry Pi Users Group

Masafumi used to be Core Contributor for OpenSolaris lead porting Laptop project and join OpenSolaris installer (Caiman) project.After he resigned the Contributor, he has founded Japanese Raspberry Pi Users Group, with a few Raspberry Pi enthusiasts and become one of volunteer, who help moderating the forum on Raspberry Pi official website in 2012.Now he also helps Raspberry Pi Foundation for their business in Japan collaborating with Eben Upton, the Founder of Raspberry Pi Foundation.

16:30 - 17:15

Unity and simulation environment for machine learning training with ML Agent

Tower A Creative Hall

Artificial intelligence is able to be learnt effectively in the realistic circumstance when we utilize Machine Learing Agent (ML Agent) provided by Unity, real-time 3D content creation engine. Use-cases can be seen in a variety of area such as autonomous vehicle, safety design of city. Demo will show application cases and simulation environment.

AI

Tower A Creative Hall

John Oh

Unity Technologies Korea

John Oh, Unity Lead Evangelist, joined Unity in 2014. He started his career in the game industry 10 years ago, and has experienced in various fields including PC online game development, services, software development and lectures on game development in university. He introduces the latest tech news and knowledge, solves their difficult issues together and shares the idea with Korea developer community as a lead evangelist.

Kyushik Min

Machine Monitoring and Control Lab.
Department of Automotive Engineering, Hanyang University

I am PhD candidate in Machine Monitoring and Control Lab. (MMC Lab.) of the Department of Automotive engineering, Hanyang University. I am one of the operators of Facebook page Reinforcement learning Korea and lecturer of Fast campus and Programmers. My research topics are ‘self driving car’, ‘driver assistance system’, ‘deep learning’ and ‘deep reinforcement learning’. I like to develop the environments for testing reinforcement learning algorithms using Unity ML-agents. Since the Unity ML-agents was released, I have developed various environments using it. In this presentation, I will talk about the development process using Unity ML-agents and thoughts about Unity ML-agents as an user.

Tracing the C/C++ open-sources for profit and fun

Tower D Frontier Hall

Tracing refers to the flow and performance measurement between function calls, and Uftrace is a tool for tracing C/C ++ programs. Compilers such as GCC provide an interface for tracing, which can be useful for open source analysis. In this presentation, we will introduce Uftrace and share a way to have fun with open source by using the many benefits of tracing

Community

Tower D Frontier Hall

Hanbum Park

KossLab

Currently active contributor to Uftrace as a open-frontier of kosslab.

Open Source Development: Behind the Scenes

Tower D Conference room 1F

What’re the meanings to participate on open source devs? To get some portfolio? To improve some inconvenience? To be looked as cool? The dev/ops of OSS, it doesn’t looks like as your expectation. What’re the condition to be a ‘successful’ OSS? Can we survive as OSS developer? Based on the experience of the chart library ‘billboard.js’, want to share the behind the scenes of OSS maintenance.

Web

Tower D Conference room 1F

Jae Sung Park

NAVER PaaS

Doing Front-end tech research at NAVER and leading the dev of OSS chart library billboard.js. Also, awarded as MS MVP, trying to exploring new techs and its sharing to the community.

9:00 - 10:00

Registration

Tower A, 2F

10:00 - 10:45

NNStreamer: Neural Networks as Filters of Stream Pipelines

Tower A Creative Hall

Recently, we have witnessed widespread of deep neural networks and their applications. Simultaneously, the range of applicable devices for such deep neural networks is expanding to personal, mobile, or even wearable/edge devices. A new challenge of such systems is to configure and execute data streams between inputs, pre/post-processors, neural networks, and applications. We propose to implement general neural network supporting filters for Gstreamer, which is actively developed at https://github.com/nnsuite/nnstreamer. With NNStreamer, neural network developers can easily configure data stream pipelines and execute them with high efficiency. Media stream developers can now use arbitrary deep neural networks as yet another media filters with much less efforts.

AI

Tower A Creative Hall

MyungJoo Ham

Samsung Electronics

MyungJoo Ham, Ph.D., has been working in Samsung Electronics as a software developer after receiving Ph.D. degree from UIUC in 2009. Recently, he has been developing common development environment and software platform for on-device AI projects varying from autonomous driving systems to consumer electronics. Before joining AI Center of Samsung, he had been mostly working on Tizen as an architect and developer. He has been a maintainer of a couple of Linux kernel subsystems and contributor of .NET core, Open Build Service, Tensorflow, and a few more open source projects.

User review of Elasticsearch Fulltext search service

Tower D Frontier Hall

Search technology in the past was technology which was not easy to access, but the emergence of Elasticsearch enabled full-text searches and other search services of natural language with a straightforward setup. This session shares the story of trials and errors experienced during the use of Elasticsearch while adding a search function to Afreeca. The story is divided into a technical part and service (planning) part. The session examines the technical knowledge whose use was considered for the launch of the initial service when I first used Elasticsearch, and shares the methods of dealing with areas that cannot be addressed by technical means.

Big Data

Tower D Frontier Hall

Kris Jeoung

AfreecaTV

Kris Jeoung is a backend developer who lives by the motto, “trials and errors today for a more convenient tomorrow.” He has developed and is providing services which process thousands to hundreds of millions of requests per day. Mr. Jeoung developed and operates Afreeca TV’s search service and cache cluster. Mr. Jeoung is the author of “This is Redis” and “Java Network Girl Neti”, and is currently working on a book for backend developers, which he plans to publish in the second half of 2018.

Introduction of private blockchain and share implementing case of customizable and lightweight blockchain engine (it-chain)

Tower D Conference room 1F

In this session, we will introduce an introduction to the private blockchain and describe the architecture and implementation of the customizable and lightweight blockchain engine (it-chain) being developed as open source and share the experience of developing blockchain as a open source committer.

Blockchain

Tower D Conference room 1F

Junbeom Lee

KossLab

As a developer of Kosslab, I'm developing open-source it-chain (Customizable and lightweight private blockchain). I am interested in software architecture, machine learning/deep learning, and blockchain. I like to study aiming at not only theory but also implementation.

11:00 - 11:40

DevSecOps meets Open Source

Tower A Creative Hall

Extending DevOps to your software, How do you make your application DevOps-Ready? You can meet DevSecOps to apply security with open source.

Security

Tower A Creative Hall

Johnny Cho

Woowa Brothers

I'm working for Woowa Brothers as a Security Professional. Open Source Software and Security are two major research areas throughout my whole life.

Virtualizing Telco Infra using SDN based Virtual Network Solution SONA

Tower D Frontier Hall

We have been developing SDN based network virtualization solution (SONA) for OpenStack. SONA uses VxLAN to provide fully independent virtual networks to 5G services and supports multiple gateways to provide multiple paths to gateways for its performance. Also, SONA supports containers created by either Docker Swarm or Kubernetes and allow communication among containers and VMs, by controlling their flows using an SDN controller. In this talk, I will present the overview of ONOS, which is the open source SDN controller and the architecture and use cases of SONA in SK Telecom, including network acceleration feature using Smart NIC.

Cloud

Tower D Frontier Hall

Sangho Shin

SK Telecom

He received B.S from Korea University and received his master’s degree and Ph.D in Computer Science from Columbia University. He is leading SDN projects at NIC Research Lab in SK Telecom, and he is a borad member of ONOS and CORD open source projects. Prior to join SK Telecom, he worked at LG for three years and Software R&D Center in Samsung for six years as a senior researcher. He also worked at ON.Lab as a visiting scholar and contributed to developing Segment Routing use case. His interests are VoIP, Wireless Networking, SDN, and Home Networking.

BPF infrastructure in kernel

Tower D Conference room 1F

How to safely inject code into running kernel ? BPF is a kernel infrastructure for code injection named "Safe dynamic programs and tools". Recently BPF is heavily used for XDP and Tracing. This presentation talks about internals of the core BPF infrastructure, what does it consist of ? and how does 'safely' inject kernel code on it.

Platform

Tower D Conference room 1F

Taeung Song

KossLab

Taeung is a Software Engineer in KOSSLAB(Korea Opensource Software Developers Lab) and have been contributing to opensource projects related with Kernel and Tracing technology.

11:50 - 12:30

Amazon FreeRTOS - IoT Operating System for Microcontrollers

Tower A Creative Hall

The internet of things (IoT) of the Edge cloud computing is evolving to facilitate implementation of AI technology. AWS Greengrass is software which allows safe execution of local computing, messaging, data caching and synchronization in a connected device. Particularly with Edge, a user may use AWS Greengrass to execute AWS Lambda functions, keep device data in a synchronized state, and communicate with other devices safely.

IoT

Tower A Creative Hall

Richard Barry

AWS

Richard Barry is a founder of FreeRTOS and principal engineer in Amazon Web Services. He started the FreeRTOS project in 2003, developed and promoted FreeRTOS for more than a decade through his Real Time Engineers Ltd. FreeRTOS is the cross platform standard solution for microcontrollers, which is ported to 27 microcontroller cores, 17 tool chains, and is downloaded more than 80,000 times a year - making it the leading kernel in its class - a claim that is backed up by the results of the 2011 EETimes embedded market survey.

Building a CI/CD environment using Kubernetes, Docker, and Jenkins

Tower D Frontier Hall

This session introduces “Immutable Infrastructure” and “Infrastructure as Codes” in an environment of DevOps. We will examine issues in the current environment and the problems solved by Docker, and explore what Kubernetes is and why it was necessary. We also consider what continuous integration/continuous delivery (CI/CD) is. We build Pipeline as Code using Jenkins, and examine examples of CI/CD pipelines.

Cloud

Tower D Frontier Hall

Hokwang Lee

Samsung Electronics

Hokwang Lee is in charge of software development infrastructure, and has recently developed a keen interest in DevOps application.

Open source based Network automation platform

Tower D Conference room 1F

We hope this occasion will be an opportunity to understand the ETSI standards, which helped start the virtualization of network equipment a few years ago. We wish to understand the development of the ETSI standards into ONAP, which will complete the automation based on the E2E solutions, and their differences. We will also share the solutions at Network Business Department for the effective NFV management and the directions of the tech development.

Cloud

Tower D Conference room 1F

Iikook Yun

Samsung Electronics

At Network Business Department of Samsung Electronics, Iikook Yun oversees NFV management and orchestration (MANO) based on the ETSI standards. He is also in charge of developing virtualization resources and service management platform based on the Open Network Automation Platform (ONAP).

12:30 - 13:30

Lunch

Tower A, 1B

13:30 - 14:15

An Introduction to DevOps Practice for Effective and Reliable Service Operation on Cloud

Tower A Creative Hall

I will share the DevOps practice to improve the efficiency and relibility of server system for IoT and AI service. We tried to utilize the open-source solutions to achieveour goal. We have been doring best effort to reduce the downtime and incident of service for customer satification.

Cloud

Tower A Creative Hall

Young-ook Kim

Samsung Electronics

DevOps Engineer of Mobile Ad. Service

Immersive Web (Web XR)

Tower D Frontier Hall

This session explains what Immersive Web is, why the Web XR Device API came up, and how the API is designed and used. Especially, the following topics are covered. We'll exclude graphic-related contents if possible for your understanding.


- What is Immersive Web?

- WebXR API History

- WebXR API Internals

- How to implement WebXR app

Web

Tower D Frontier Hall

Jinho Bang

Samsung Electronics

I'm working at Samsung and develop a web browser as know as Samsung Internet. I'm also active contributor in many open sources such as Chromium/Blink, W3C and Node.js.

Byoungkwon Ko

Samsung Electronics

I am a S/W developer developing multimedia applications in Samsung Electronics. I have contributed to Chromium/Blink with personal interest in open source.

Development of distributed web engine technology using multi-device edge computing

Tower D Conference room 1F

In light of ongoing advances in web technology and the ever-increasing complexity and performance requirements of web contents, this session introduces a multi-device distributed processing web engine technology. This technology enables the effective provision of the required services by overcoming the resource restrictions of single devices and making efficient use of idle resources in multiple devices.

Edge Computing

Tower D Conference room 1F

Eun Namgung

Samsung Electronics

Eun Namgung is in charge of various web technology-based components and services at Tizen Development Team at Samsung Electronics. The services provided by the web technology include web engines, web runtimes, and web application frameworks applicable to the Tizen platform.

14:30 - 15:15

AI service infra framework development experience story

Tower A Creative Hall

This talk introduces the open source AI framework and shares the anxieties and solutions in the development process.

AI

Tower A Creative Hall

Yeonhee Lee

Electronics and Telecommunications Research Institute

I am leading I am leading an artificial intelligence framework development project.

Developing Samsung Internet with Open source Chromium

Tower D Frontier Hall

Samsung Internet is the secure, private and convenient mobile web browser optimized for Galaxy devices. In this talk, we will share how Samsung Internet was developed, and the lessons learned while using Chromium - one of the most dynamic open source projects – in productization.

Web

Tower D Frontier Hall

Gangil Han

Samsung Electronics

Samsung Internet developer. Always thinking of how to make Samsung Internet an attractive and exciting web browser to both users and developers.

OpenStack Trove – Open Source Database as a Service solution for the Cloud

Tower D Conference room 1F

Databases are crucial and ubiquities elements of most applications running in the cloud. However, database management in single-instance or clustered deployment scenarios requires expertise and deep understanding of the underlying infrastructure. The main goal of Database as a Service (DBaaS) is to deliver a service responsible for on-demand provisioning and management of various database management systems including relational and non-relational databases. Key features provided by the service include, among others: provisioning of single and multi-node database deployments, scaling (out, up and in), data backup and recovery, configuration management, and failover support. During the session, a DBaaS solution based on OpenStack Trove will be presented along with a comparison with commercial alternatives such as AWS RDS. In addition, the path of our team towards the open source oriented development will be described: from the closed circle of project maintenance, through understanding the potential of open source contribution, ending with full involvement in the project, taking the leadership and adopting upstream-first workflow. The advantages and disadvantages of such approach will be presented, as well as obstacles that the team had to face at an early stage.

Cloud

Tower D Conference room 1F

Bartosz Zurkowski

Samsung Electronics

Passionate about designing large-scale distributed systems and enthusiast of innovative technologies in the cloud. Computer science graduate at the AGH University of Science and Technology in Krakow. During the studies and after graduation, closely associated with the IT industry. He has 4 years of professional experience. At first worked as software engineer designing microservice-based applications for advertising companies. Later responsible for administration of virtual resource infrastructure in AWS cloud and deploying architectural components related to telemetry. More than a year ago started the journey with OpenStack at Samsung R&D, working on a cloud project as a member of the core team building the Database as a Service platform for 5G requirements. He actively participates in the development of OpenStack Trove by enhancing its features in the scope of automated database provisioning, cluster management and data recovery. Broad database experience allows him to elaborate advanced solutions oriented to provide high data capacity and availability. At the same time, he is pursuing a PhD in computer science in the field of NFV and multi-site cloud computing. As part of this venture, he had the opportunity to gain knowledge about a wide range of native cloud technologies, as well as interact with exotic OpenStack initiatives such as: Tacker, Trio2o and Tricircle.

15:30 - 16:15

Automatically Generating Search Heuristics of Software Testing with Machine Learning

Tower A Creative Hall

4In the Fourth Industrial Revolution, software (SW) is the brain of a smart world where everything is connected to IT. Recently, the demand for "safe software" explodes as all industries depend on SW. For example, Microsoft has successfully detected a number of security vulnerabilities in the Windows 7 operating system through the SW testing tool 'SAGE'. The core technique used is concolic testing that combines concrete and symbolic executions, which aims to increase code coverage and to detect vulnerabilities. The limitation of concolic testing is that the effectiveness heavily depends on the so-called search heuristic. Hence, many experts have manually designed a search heuristic for each SW. However, we found that the existing manually-crafted heuristics did not consistently perform well on arbitrary SWs. To overcome the problem, Korea university research team presents a technique for automatically generating a search heuristic optimized for each target SW. And this work is accepted in International Conference on Software Engineering (ICSE), the world's most premier software engineering conference. In particular, using the technique improves the detection rate of SW vulnerabilities by up to 20 times. We expect that all SW companies will have high demands for our technique in the future. The tool, named ParaDySE, is implemented in IoT cube, vulnerability automatic analysis platform which can be used by anyone, and is also available as open source in github. In this talk, I aim to explain the background for concolic testing and the core technique of ParaDySE.

Security

Tower A Creative Hall

Sooyoung Cha

Center for Software Security & Assurance, Korea University

I am Sooyoung Cha, currently a PhD student at Korea University's Software Analysis Lab. My research area is to improve the performance of software testing via machine-learning, thereby detecting security vulnerabilities in software. Recently, the works are accepted as papers in ‘International Conference on Software Engineering (ICSE)’ and ‘IEEE / ACM International Conference on Automated Software Engineering (ASE)’. I also have implemented the technique in IoTcube system of Center for Software Security and Assurance (CSSA) and make our tool publicly available for anyone in the world to build safe software. (Module name: CTEST)

Modern C++ Open Source Development

Tower D Frontier Hall

It requires many steps to build an open source project written in C++. Due to the characteristics of C++ language, different settings are needed for each platform, and the lack of an integrated tool to manage packages means that the method of library installation is different for each project. Moreover, modern C++ (C++11 or higher) requires a new installation of the compiler for some operating system versions which provide continuous integration (CI). There are also features which are not supported by certain platforms, creating many difficulties. This session will use real-life examples to examine the difficulties faced when a developer builds an open source project using modern C++, and the solutions to address them.

Community

Tower D Frontier Hall

Dongha Park

Alchera

He is currently in charge of developing and distributing cross-platform libraries at Alchera. As a management member of C++ Korea Facebook group, he is very interested in C++ usage in various conditions. Recently, he has been working hard to invigorate the localization project of the core guideline into Korean.

Piece of cake - testing remote embedded devices made easy (using open-hardware MuxPi)

Tower D Conference room 1F

Designing custom hardware, though might seem intimidating at first, can simplify many activities. Paweł will present a fresh open hardware approach at providing remote access (including device flashing, debugging and power management) to embedded devices - MuxPi, successor of SD MUX board (showcased during FOSDEM 2016 and ELC 2016). This talk will introduce novices to the topic of testing on remote embedded devices and previous attempts to do so. Presentation will also introduce components of SLAV - the new testing laboratory: Boruta (board farm management system), Weles (LAVA-compatible testing framework), Perun (binary image testing system). It will hopefully encourage attendees to try this open hardware design out in their own setups and share feedback on what can be further improved.

Platform

Tower D Conference room 1F

Pawel Wieczorek

Samsung Electronics

Paweł Wieczorek works at Samsung R&D Institute Poland since 2014. Starting as an access control developer, Paweł contributed to the security framework of operating system for embedded devices. At that time, he introduced testing automation practices in software development and still actively develops automated test system. Currently he's a platform release engineer. Speaker on ELC and ELCE, FOSDEM, FrOSCon and a few others.

16:30 - 17:15

From Backend.AI to AIN: The Future of Cheaper, Faster, and Safer AI Development

Tower A Creative Hall

The increasing diversity of machine learning frameworks and cloud services has motivated AI development for everyone. Still, it needs relentless efforts and time-consuming tasks to manage computing infrastructures and optimize the computing costs. To resolve such pain points, Lablup has been developing Backend.AI. Based upon it, we are now going to start a new chapter, the AI Network (AIN) project. Like what SETI@Home did, AIN offers instant exchange of computing resources with AI modelling computations. The AIN's blockchain enables such trades between untrusted peers without complicated contracts and configurations, while Backend.AI takes care of security problems in remote code execution and automation of infrastructure management. This talk covers technical challeges we have faced and identified to realize our goals. In particular, I will go deeper in the design of the Callosum RPC Transport Library for inter-cluster remote code execution and our approach to the computation verification problem for AI modelling in the AIN blockchain.

AI

Tower A Creative Hall

Joongi Kim

Lablup Inc.

Joongi is the main author of Backend.AI and CTO in Lablup. He received Ph.D in Computer Science from KAIST by developing a GPU-accelerated packet processing framework offering world-first 80 Gbps performance. He has diverse experiences on analysis and design of scalable backend systems via a research internship in Microsoft Research Cambridge and an engineering internship in NexR (acquired by KT). He is also an open-source enthusiast. He has contributed to a number of open source projects, such as Textcube, iPuTTY, Python, aiodocker, aiohttp, pyzmq, DPDK, and more. Currently enrolled as a global open-frontier program member in KossLab, sponsored by NIPA.

A Story of ARM Server for Data Center that is Completed as Open Source

Tower D Frontier Hall

As the fourth industrial revolution, represented by machine learning and artificial intelligence, is accelerating, cloud-based infrastructure investment is rapidly increasing, but the demand for cost savings such as electricity taxes is growing proportionately. More than 99% of servers in the data center, which are core to cloud infrastructure, are Intel CPU-based servers, and hardware servers such as motherboards are developed in Taiwan, making it difficult to find optimized servers in the domestic environment and cost savings. As a result, we want to share the low-powered ARM server with open source based on how our company has developed the first low-powered ARM server in Korea, from motherboards to software. In the cloud era, which is a large infrastructure that cannot be managed passively by people, we share examples of how servers can be managed efficiently and build a competitive environment optimized for the cloud era instead of similar products.

Cloud

Tower D Frontier Hall

MyoungHwan Yoo

XSLAB Inc.

He is the first and only ARM server for the data center in Korea and has been working as an operating staff for seven years at OpenStack Korea Community, the largest community of cloud open sources. He served as a member of the Cloud Technical Committee of the Integrated Government Information Center and SW Maestro Mentor, currently serving as Adjunct Professor of ERICA Campus at Hanyang University and technical policy advisor for smart city in Guro-gu, Seoul. He is a community evangelist who created the Korea Community Day (KCD), the first community-oriented event in Korea, that is very interested in the community that needs to grow open source. That's why he loves to chat with developers and share my knowledge and experiences.

Intelligent smart construction monitoring system Using machine learning Based on EdgeX

Tower D Conference room 1F

In a construction site, there are many dangerous elements such as noisy, gas leak and vibration needed by alerts, so that real-time monitoring system to detect the alerts using machine learning techniques (DL, RL) can provide more effective solution and approach to recognize dangerous construction elements. Representatively, to monitor these elements CCTV (Closed Circuit TeleVision) should be locally and continuously broadcasting in a situation of construction site. At that time, it is in-effective and wasteful even if the CCTV is constantly broadcasting unchangeable scenes in high definition. However, when any alert should be detected due to the dangerous elements, the streaming should be converted to high quality streaming data to rapidly show and defect the dangerous situation. To approach technically, DL is one of the solutions to automatically detect these kinds of dangerous situations with prediction in an advance. It can provide the transform data including with the high-rate streaming video and quickly prevent the other risks. RL is additionally important role to efficiently manage and monitor with the given dataset in real time.

Edge Computing

Tower D Conference room 1F

Kwihoon Kim

Electronics and Telecommunications Research Institute

He is a principal research engineer in the intelligent IoE networking research section at the Electronics and Telecommunication Research Institute (ETRI), Daejeon, Republic of Korea. He received his B.S. and M.S. degrees at KAIST in 1998 and 2000, respectively. He worked at LG DACOM from 2000 to 2005 and is a research engineer at ETRI since 2005. He is an editor and rapporteur of ITU-T SG11 since 2006. His fields of interest are fog/edge computing, Internet of Things, 5G/IMT2020, deep learning, machine learning, reinforcement learning, GAN, and knowledge-converged intelligent service.

*Please note that above session schedule is subject to change due to certain circumstances.