COMMUNITY

Content is available only at each community time.
After the event, you can freely available the entire video.

  • AI Robotics KR
  • 13:00 ~ 14:00 (KST)
Introduction to the AI Robotics KR community

Introduction to the AI Robotics KR community and activities

  • Bill Park
  • I am a dream 'maker' developer. Im currently developing servers through 'Arduino' at 'Embedded'. Ive been running towards whats in front of me for so long that this is where ive come to be. My dream is to make people's life more comfortable.
Takeaways from the Neural Acceleration Study experience

Neural network acceleration was studied for 6 months. Normally the concept of neural networks and deep learning studies involve methods of designing models. I have made presentations on papers on wider themes involving acceleration methodologies and my work on hardware designs for neural network acceleration studies as well. Id like to talk about the things that i had to deal with while managing those studies as well as the things i have experienced while being involved in them as well.

  • Sangsoo Park
  • Hello. My name is Sangsu Park and im currently pursuing my PhD at Hanyang University. My field of study is focused on the different methodologies in accelerating the neural network. More specifically, low energy consumption, highly efficeint neural networks acceleration hardware architecture research and research on efficient neural network acceleration methodologies via current processors (CPU/GPU/NPU). In addition to my own research, im also involved in various projects as well. The different environments where home appliances like robot cleaning equipment have to operate vary greatly. As such, im working on hardware accelerators for learning transfer framework (LG electronics) to overcome these differing environments as well as voice recognition technology. Another is the NDP project (Samsung Electronics) which is accelerated by having the recommended systems calculate near the memory. Through all these, im researching the different methodologies on neural network acceleration and leading those studies.
Sharing the experience on sensor fusion studies and projects

[Project introduction]
Project 'Meerkat' was initially started to create a robot that would organize materials in a small store environment (convenience stores etc) where students who love robots gather. Since february, the project team comprises of 3 members who have each been assigned responsibilities for the main hardware, main software and manipulators. Among those areas, ive been involved in developing the motion software, which finds their path through reinforcement learning. While executing learning through simulations and trying to reduce the gap between reality and simulations, i became interested in Pose estimation and eventually got involved with the sensor fusion study headed by AI Robotics Korea. Instead of estimating correct positioning through input from only computer vision, what ive learned from this study was the fusion of data from other sensors to estimate proper positioning. I applied what ilearned here to the current projects that im involved with.

[Real to Sim]
The Kalman filter was used to measure the parameters that went into the simulator. In order to enable the simulator to mimic the dynamics, linear speed and acceleration of a real car, depth cameras and IMU were utilized to measure the speed and calculate the fusion values. In addition, after establishing the car's locational values and dispersion, these values were applied into the simulator so that the car can learn by reading noise that would normally be picked up in a real life environment of the car's current location.

[Sim to Real]
State values recevied from the simulator and the Kalman filter fused status values to simulate reality, were then used as input to the help drive the vehicle.

  • Jungeun Park
  • Jungeun Park
  • My name is Jungeun Park. Im a 23 year old undergraduate student that been assigned to manage AI Robotics Korea. I major in mechanical engineering at Korea University and is studying various areas focused on robotic engineering. My main areas of interest are robotic reinforcement learning and positioning estimation. I also have a lot of interest in artificial intelligence and actively in pursuit of more experiences and knowledge.
Android on-device learning application - centered around Tensorflow lite

This presentation is on the deep learning based open source project: experiences on the development of a real time expression reading service by a mercy android developer will be presented. We'll be covering a little on the on-device application process after converting a deep learning model to a tensorflow lite type but much more on the developed device that was the result of a collaboration between the developers and designers. More precisely, the presentation will cover the theory behind the process for a recommended expression based deep learning model and how the model is converted to a tensorflow lite library and eventually creating a service. In addition, the presentation will also cover how th performance of this device can be maximized through this process as well as issues and solutions in how the deep learning model was optimized.

  • Changyeon Seo
  • Im an android developer who started from the bottom and grew with the industry. I believe that I was able to grow to this point in my career due to all the knowledge that my developer predessors were able to disclose to me. With this foundation, I am passing it forward by sharing my experiences with new technology.
  • DATAYANOLJA
  • 14:00 ~ 15:00 (KST)
Taking a look at DATAYANOLJA 2020

Ill be conveying what ive heard directly from the DATYANOLJA director and provide a teaser of the interesting points of DATAYANOLJA

  • Mingu Kang
  • I am a data analyst that likes analyzing data and the communities invovled in data analysis. I am constantly trying to figure out how to get people to be closer to data.
  • Jinyeong Kim
  • Jinyeong Kim
  • I am a data visualization designer and studied in the Netherlands in a Master's program focusing on 'Generative Visualization'. Im currently lecturing at SungKyun Kwan University, SADI (Samsung Art and Design Institute) etc and also concurrently managing "flecto" which is a visual design studio.
  • Ilsup Lee
  • As the head of the DATAYANOLJA community, I convey the stories of the people who handle data and provide a space for those people to communicate.
The Concept of Geometric Brownian Motion and Simulation of Yahoo Stock Price

This presentation will be covering the process of deducing GBM based on mathematical theories and applying this to simulations that contain the movements of yahoo stock price data.

  • Eunsu Kim
  • My name is Eunsu Kim and I have an immense curiosity for new places and applying data from valid sources.
  • Minjeong Seo
  • My name is Minjeong Seo and I dream of becoming a clean code developer and love challenges
Using opencv to develop a mask enforcement application

In an environment like these days where daily usage of masks are important, the application is to find people who are not wearing masks and keep sending out nagging messages.

  • Areum Yang
  • I am the front leader at DATAYANOLJA and am growing my professional career while at DATAYANOLJA
  • Heena Yoon
  • My name is Heena Yoon and I'm a novice in this field. Im currently trying to find ways to play with data
Predicting the success of a movie from 'word of mouth' data

Using data based on the reviews of the moviegoers on the movie's first debut, press exposure information, social network mentions etc, predictions are made on the first week and second week ticket sales numbers and the screening counts.

  • Suyeon Shim
  • My name is Sooyeon Shim. I like talking to different people and dream of becoming a data analyst.
  • KERAS Korea
  • 15:00 ~ 16:00 (KST)
LEGO_MNIST : Transplanting a Keras learned model into Legos without a deep learning library

As robot lego products were introduced, a thought of transplanting AI into legos came to mind. Much like the science fiction movie scenes where AI or a brain is transplanted into robots. Applying MicroPython on Lego spike prime products is possible. But an attempt was made to transplant AI with a basically pure python that does not have a pre-installed separate deep learning library. After having MNIST learn the image differentiation model using the Keras deep learning library from the Google Colab cloud, the experienced data are downloaded and transplanted into Lego spike prime. It is hoped that a wide variety of experiments will be generated from this attempt to put life into lego.

  • Taeyoung Kim
  • Taeyoung Kim
  • Taeyoung Kim is the CEO of AIFactory and the Microsoft Regional Director. In addition to predicting the gaps in the internet, digital, mobile and cloud arena, he is also predicting AI gaps to achieve a vision where everyone can easily afford the use of AI.
    He is constantly expanding his knowledge on AI through the communities and applying them to his work. He has published "Block together with Python deep learning Keras". This book matches the AI library with lego blocks and is an AI workbook that has no matehematical symbols or any type of math in it. There is also a symbolic anecdote where a child thought the book was on lego blocks but after following the book, eventually became capable of building an AI model. In order to overcome the disadvantage of being in regional districts, he is expanding his knowledge together with approximately 16000 members in the Kera and Kaggle communities
    With an educational foundation in the field of aerospace, he is involved in various signal/visual processing and analysis projects in various fields like astonomy/satellite/drone/healthcare/medical/network etc. IN march of 2019, he was listed as a primary cowriter for a research in analyzing the various visual data of the Sun's surface through deep learning in the Nature Astronomy journal. He has presented on his various AI related experiences at Global AI bootcamp, GDG dev fest, NASA FDL, NVidia AI, Samsung Open Source Conference etc.
    AI model research is not easy but thanks to the democratization of AI technology, the ability for AI service technology to overcome resistance to penetration is not considered to be to be high. Similarly, with the advent of Youtube, the difficulty of penetrating the broadcasting industry has long disappeared. The only element of being competitive being the ability to have a unique individualistic content. It is predicted that we will be getting to a generation where content is sought after beyond the technology. It is a dream to grow AIFactory to a story network platform so that anyone can flaunt their AI model and communicate with one another.
Points to consider when intending to use deep learning models as commercial solutions

Disclosing the points to consider when intending to use deep learning models as commercial solutions

  • Taeyoung Lee
  • As a chief research engineer, he has been involved in improving the efficiency of data pipelines and models as a service by finding solutions to efficiently connect the limited Legacy environments to deep learning models through work in the Posco people finder chatbot, Tensor MSA (deep learning framework), Shinhan bank deep learning framework, Soulmate chatbot, Soul reach roboadvisor, and AI OCR vision platform project.
  • OROCA
  • 16:00 ~ 17:00 (KST)
Introduction to the OROCA Open source software and hardware community

Introduction to OROCA's history and activities in Open source software and hardware community

  • Yoonseok Pyo
  • I have been coordinating public lectures and public projects for the 'Open Source Robot technology community (www.oroca.org) oriented around open robotics and great interest in 'robot engineering public forums' where individuals who dream of robots can enjoy sharing stories and dialog.
Control of robots utilizing chatbots

As a project that is conducted online, there are many projects being introducted overseas on controlling robots using telegram. But this is a workshop to introduce a project controlling robots using the kakaotalk bot which is predominantly used in Korea.

  • Kyunam Choi
  • As a staff for the "Open source robot technology community (OROCA)", I am mainly active in the Pankyo region's OROCA Pankyo region group. Through this group, we conduct workshops and gatherings for the exchange of AI, autonomous driving, robot control related technology. I am also currently doing research on related technology at the AI healthcare research center at Gachon University.
Reinforcement learning applied soccer

Referencing chatbot soccer videos using reinforcement learning, an open project is being conducted based on open AI ros. Related material and project content will be introduced.

  • Minwoo Kim
  • Im a high frequency antenna circuitry design engineer. For the last 2 years, ive started the Auturbo movement and am currently involved in arm part. While being involved in various Ros based autonomous mobility projects, ive learned to enjoy coding. And if opportunity permits, i intend to continue working in various types of robot autonomous mobility projects.
Developing the turning drive based omnidirectional degrees of freedom mobile platform

Auturbo mobile robot part enabled mobility as well as developed an actuation module that enabled steering. This development enabled the creation of a mobile robot with omnidirectional degrees of freedom and this presentation covers that development process.

  • Jungam Seo
  • As a group member, I like studying ROS and autonomous driving and exchanging ideas and knowledge. Im also lead the mobile robot part at Auturbo. This year, my part members and I have developed a fun mobile robot so lets take a look at that and enjoy it together.
Project Unicorn

The primary objective of project unicorn is to drive the approximately 85 cm unicorn gundam (PG shape) designed via Non-based by using OpenJigWare, an opensoure robot management program. Primarily printed and assembled but not completed, the unicorn gundam is currently disassembled to be reprinted with more detail and upgraded. The objective is to complete the build and enable mobility before maker faire. This and currently in progress work will be shared once its completed. After that, several more systems will be added.

  • Donggeun Yoo
  • SOSCON 2020 presents the [Open source robot management program Openjigware] and [Project-Unicorn]. These projects are currently extended and the remaining topics of the presentation will soon be introduced at a later time.
Project J-Drive

The walking robot actuators require technology to safely interact with its surroundings with high efficiency. This presentation will introduce an open project developing an affordable robot actuator based on QDD among the various methods that would satisfy this requirement.

  • Ji woo Kim
  • Im a student performing development as a hobby. Im currently actively involved in OROCA's OPenJigWare and am greatly interested in innovative technologies.
  • Women Who Code Seoul
  • 17:00 ~ 18:00 (KST)
If you want to train to be a leader, join the community

In the midst of a career change, I came to know about 'Women who code'. Through launching the Seoul network and in the process of connecting with the organization of the US region IT women professional conference, im training to be a global leader. In a constantly fast paced changing world, together with various talented women and sharing their courage, ambitions and worries, we are planning a successful future.
#why : to excel in technology careers
#applaudher : test your skills and talents and let us show you the method to achieve success without compromise
#connect : A global IT women festival, come join us at connect.

  • Yujin Chung
  • Currently the marketing lead at NOOM, a global healthcare company and also director of 'women who code' Seoul.
Stories from the manpower in the outskirts in maintaining high technology (feat AI)

I consider developing experience in the tech arena as ;riding the wave'. Youre preparing for it and then when the wave hits, you get up and stand on both feets to ride it. Rather than a perfect preparation, establishing a balance is more important. Penetrating the high tech wall and my fearful experience with others in facing a seemingly difficult business. I want to prepare to 'ride the career wave' while sharing my experiences.
*Tech Minority : A career journey towards an AI startup as a woman, a non-professional, a non developer.
*Examining the AI business : various business fields that i previously didnt know about
*taking on an unfamiliar High tech challenge = possible opportunity

  • Hyoeun Lee
  • I am currently an outskirts resource that meet all the tech minority conditions. Started in the tech training field and currently in a role of executing high tech communications with the outside. Currently team lead of the Super BAI content and director of 'Women who code' Seoul.
Work together with Women who code Seoul

Women who code Seoul is currently looking for new management talent who are part of the global IT women community. Activities of the Seoul 'Women who code' management will be introduced and will provide information on joining the management team.

  • Soojung Kim
  • Currently working as a career management service, Join startup PO for growing talent. Engaged in the training industry, Ive learned to understand the importance of the culture of 'growing together'. With that experience, Im currently actively working as a director for the Women who code Seoul community.