Kaist

Guest Lecture at KOREA–JAPAN S..

Jong-Hwan Kim, the dean of the College of Engineering, gave a guest lecture at the KOREA–JAPAN Symposium on the 4th Industrial Revolution on September 13. Organized by Elsevier Korea and supported by the Embassy of Japan in Korea, the symposium set the stage for Korean and Japanese researchers to engage in active discussions on the fourth industrial revolution, focusing on key research fields such as artificial intelligence (AI) and big data. Dean Jong-Hwan Kim gave a presentation titled “Agent-embedded Robot with Machine Intelligence” in Session 5.

[Munhwa Ilbo] World’s First AI..

With the hosting of the AI World Cup 2018, a first since the robot soccer competition was held in the 1990s, Dean Jong-Hwan Kim of the College of Engineering expressed a desire to “expand the AI World Cup to ball games beyond soccer.” On August 21 to 22, the AI World Cup was held at the Academic Cultural Complex of KAIST’s main campus in Daejeon. The event, which involves both students and researchers, was described by the dean as “an opportunity for anyone to easily learn and experience AI technology through source codes shared by AI leaders.” ―What are some advantages of applying AI technology to soccer? “The organizing committee runs programs submitted by each team from start to finish. At this stage, programming relies on pixel information. However, more realistic simulations may be possible with access to the necessary resources, such as FIFA data on World Cup players. For instance, we can create an AI player modeled after Heung-min Son. The technology can be further developed to analyze the style of play of different players based on video recordings. Strategies employed by high-ranking teams can also be analyzed.” ―The competition seems to lack sponsors despite its significance. Are there any plans to promote the AI World Cup? “The operations behind the AI World Cup have yet to stabilize. By investing in game companies, we will develop game characters and conduct related research. We are also exploring directions in education and entertainment. While some people have a negative view of games, we’d like to think of them as a useful tool in the understanding of AI. The website of the AI World Cup has already attracted 2.1 million visitors, and recorded 15,000 downloads. Once we are prepared, we will focus on securing more investment and support.” ―What was the inspiration behind combining AI and soccer? “Soccer is widely popular in Korea. We started with the hope of Korea becoming a leading country in robot soccer. The AI World Cup was first organized two years ago, after the shock of AlphaGo versus Lee Sedol. We wanted to restore human confidence, which was affected from being defeated by AlphaGo.” Daejeon = By Seong-yeol No nosr＠munhwa.com

[The Cutting Edge] Korean team..

[The Cutting Edge] Korean team takes the win at AI World Cup Aug 27,2018 Artificial intelligence-powered players play football at the Korea Advanced Institute of Science and Technology in Daejeon during the AI World Cup on Wednesday. [YONHAP] Two years after the historic match between AlphaGo and Lee Se-dol, Korea played host to another contest designed to test the potential and limitations of artificial intelligence. This time around, the area of interest is football. The Korea Advanced Institute of Science and Technology (Kaist) held the AI World Cup in Daejeon last week, proving that artificial intelligence is ready to take on the world of football. The three-day competition saw 24 teams from 12 countries pitted against each other. Kaist ultimately took the win, beating competitors from as far afield as Google, the Massachusetts Institute of Technology and the National Cheng Kung University. The AI World Cup was a variant of five-a-side football. Each virtual team consisted of five cube-shaped players of different colors that were pre-programmed with AI algorithms. The teams face off on the virtual football field, playing shorter games with five-minute halves. Kim Woo-joon, a PhD student at Kaist and the leader of the winning AFC-WISRL team, said that the result proves the powerful capacity of reinforcement learning used in the football match. Reinforcement learning is a central type of machine learning that lets software agents learn from the environment and receive rewards by performing certain actions. “The players are trained with an algorithm based on Q-learning, a kind of reinforcement learning,” Kim said. “The algorithm is designed to maximize rewards if an agent follows through with optimal action selections.” Kim explained the key to beating out all the advanced algorithms lies in a myriad of simulation practices that his team has done. “A robot can make five steps in one second in the algorithm,” Kim said, referring to certain actions that a robot can take. “We simulated 3 million different steps to see the optimal result depending on different circumstances.” As much as the AI technique is sophisticated, the head of the organizing committee said that the main goal of the event is to raise awareness of the “entertaining” aspects of AI. “People would like to see the progress made by the technology,” said Professor Kim Jong-hwan, chair of the organizing committee. “But a more essential question is how the technology can contribute to the happiness and well-being of humankind. The AI World Cup was held to show artificial intelligence can entertain people.” As noted by Kim, spectators seemed to have a good time while watching the games. The audience was especially amused when things went wrong - at one point, an AI player couldn’t handle the pressure and stood completely still when they had a chance to score. The World Cup also featured AI-powered journalists and commentators for the games and lectures from AI specialists. “We hope that spectators become familiar with different aspects of AI technology and become more open-minded,” Kim said. BY PARK EUN-JEE [park.eunjee＠joongang.co.kr]

Outstanding Papers of Military..

On December 27, the College of Engineering evaluated the research outcomes of commissioned military officers under the Military-Academy Cooperation Program and held the Night of Military-Academy Cooperation. Candidates expecting to graduate in February 2018 received outstanding paper awards and feedback on their research. Among commissioned military officers, those who graduated with outstanding papers were also presented with outstanding paper awards to boost their morale and inspire others. The event was attended by KAIST’s commissioned military officers, Republic of Korea Army Headquarters, and the head of ROTC at Chungnam National University. The awards were handed out by Jong-Hwan Kim, the dean of the College of Engineering.

International Robot Olympiad H..

The International Robot Olympiad 2017 (https://www.iroc.org/about) was held in Qinhuangdao, Hebei, China from December 16 to 19, 2017. The Olympiad, celebrating its 19th year, was comprised of the following challenges: Robot Gathering, Mission Challenge, Emergency Rescue, Transporter, Battle Cube, Traverse, Energy Savings, Creative Category, and Robot in a Movie. The opening speech was given by Jong-Hwan Kim, the dean of the College of Engineering.

Joint New Year’s Meeting Held ..

On January 3, the College of Engineering held a joint New Year’s meeting for the Planning Committee, PR Committee, and administrative staff. The Planning Committee and PR Committee are new committees established in 2017. Through the New Year’s meeting, the two committees engaged in active discussions on future directions and shared their vision. The meeting began with an opening speech by the dean of the College of Engineering, followed by an introduction of the committees by the respective chairs. The PR Committee, which was expanded from the committee in charge of newsletter editing and publishing, distributed plaques of appreciation to its members for playing a key role in sustaining the College of Engineering Newsletter. The event was attended by the college dean, vice dean, members of the Planning Committee, members of the PR Committee, and administrative staff.

[The Hankook-Ilbo] KAIST Dean ..

KAIST Dean Holds World’s First AI Soccer Competition KAIST School of Engineering Dean Jong-Hwan Kim, who created the world’s very first robot soccer tournament, will hold the world’s first “AI World Cup 2017” tournament in November, integrating artificial intelligence (AI) technologies. On the 11th, KAIST announced that there are three categories to the competition: five players that carried out learning through AI technology comprise one team to play AI soccer followed by AI match commentators and AI reporters that writes news articles based on the match outcome. The soccer match is conducted by executing the AI technology execution program submitted by each team and uploaded to a computer server. Each player should recognize the situation taking place online and perform learning on its own to proceed with the match. The AI commentator provides commentary of the match situation shown online in an easy to understand manner for regular people as well as explaining predictions of the players and the match outcome. The AI reporter writes a news article based on the match results. Undergraduate and graduate school students all over the country are eligible to participate. After an online practice period of one month during October, preliminaries are conducted from November 1~24 where the 16 teams with the highest cumulative match scores will participate in the finals on December 1. Dean Kim explained the motive behind holding the competition. “People know about AI research overseas such as AlphaGo but are not familiar with the AI research carried out domestically,” he said. “One aspect of holding the competition was to let the people know more about AI related R&D and technology application and stimulate their interest in the field.” To prepare for the tournament, KAIST established the “AI World Cup Organizing Committee” last month with Dean Kim as the committee chair. The committee is composed of six subcommittees including the steering and competition committees. KAIST professors in the School of Electrical Engineering, School of Computing, Department of Aerospace Engineering, Department of the Cho Chun Shik Graduate School for Green Transportation, Department of Civil and Environmental Engineering, Graduate School of Culture Technology, and Graduate School of Science and Technology Policy are participating and taking roles in the committee. Dean Kim said that he plans to expand the number of soccer players to 11 and open up applications to international students so that the tournament can establish itself as an international competition and he will introduce the AI soccer tournament during his lecture at the general meeting on a visit to France to participate in an International Federation of Automatic Control (IFAC) meeting. Applications will be taken until September 30 through the website (http://mir.kaist.ac.kr) after inputting the participating competition category. On the 28th, a presentation is planned to explain how to use the system and the participation method for interested students. Also, an open demonstration game is prepared for the OPEN-KAIST event on November 2~3. >> Link to the original news article http://www.hankookilbo.com/v/a99fe47db3414cc1a2809c1a113721b5

[SEDaily] Nvidia seeks KAIST i..

In August 2014, master’s and doctorate program candidates take a group photo together in front of Apple headquarters during their participation at a seminar held at Silicon Valley, US, hosted by the research laboratory under KAIST Professor Jeong-Ho Kim (third from the left). Soo-Min Choi (right) who participated at the event is currently doing an internship at Nvidia and plans to officially join Nvidia after obtaining his doctorate degree next year. /photo credit=Professor Jeong-Ho Kim Recently, KAIST College of Engineering Dean Jong-Hwan Kim (School of Electrical Engineering professor) received an unexpected email. It was from Jensen Huang CEO of Nvidia, a leader in the artificial intelligence (AI) computing field, revealing his intent on sponsoring the AI World Cup hosted by KAIST. Huang even proposed to “jointly establish an AI robotics research center.” On the 31st, KAIST Director of Office of Research Affairs Jung-Ho Kim revealed that “the establishment of an AI robotics research center for the continuous R&D and cultivation of professional human resources in the AI field is currently being discussed with Nvidia”. Kim also said that “(Nvidia) hopes to be one step ahead technologically through this research center and their intent to secure AI intellects appear to be strong.” While there was the match between professional Go player Sedol Lee of 9 dan rank and AlphaGo, this is the first time that an AI competition was held over sports. This competition to be held in November invites undergraduate and graduate students and will involve 5 mini robots that learned on their own based on AI technology to make one team where the team with the most goals through the opponent team’s goal posts wins. In addition to the AI soccer players, there are two categories of AI commentator that analyzes videos of the match and provides commentary and AI reporter that writes up the match results into a news article. Each match is carried out by executing the AI algorithm program developed by each team and uploaded to a computer server. The soccer match between the mini robots is shown on-screen followed by the match commentary and article composition. As the competition to entice AI talent grows fiercer worldwide, efforts to take the lead have taken various forms including CEOs of global companies directly contacting Korean universities. Nvidia, Google, Apple, Facebook, and Intel among others are actively seeking intellectuals in Korea. The scope of headhunting has expanded from Korean students studying abroad in laboratories in prominent US engineering universities to Korean university students with research achievements or original ideas. The technology firms have begun their search for skilled talent beyond national borders. KAIST Professor Jeong-Ho Kim is explaining the recent semiconductor design technology to Intel at the Intel headquarters located in Silicon Valley, US, in August 2014. Dr. Ga-Won Kim (second from left in the right line) and Yoo-Jung Shim (third) sitting next to Professor Kim (first) had received offers from Intel afterwards and are not currently working in Silicon Valley. /photo credit=Professor Jeong-Ho Kim Such offers are concentrated towards students and specialists in fields related to the 4th Industrial Revolution such as AI, big data, and robotics. In the last three years, around 45 KAIST graduates from the Schools of Electrical Engineering and Computing have been scouted by Nvidia, Apple, and Intel. Around 10 Seoul National University School of Engineering graduates have also been scouted by Silicon Valley. Among the 70 or so alumni of Professor Kim’s laboratory, researching 3-dimensional semiconductor design, around 23 students are working in Silicon Valley. Dae-Hyun Jeong, who joined Nvidia 10 years ago and participated in the semiconductor design part, is currently working as an AI server module designer and manager. With the recommendation of alumni senior Jeong, Soo-Min Choi (KAIST School of Electrical Engineering) began her internship at Nvidia early this year and is expected to join Nvidia after receiving her doctorate degree in August next year. Professor Kim said that “Choi would frequently meet CEO Huang in the hallway and greet him comfortably ? not only that, a while back Huang invited Choi to his home for a meal together”. Professor Kim continued, “Global IT companies are focusing their efforts in acquiring skilled specialists for the 4th Industrial Revolution, especially AI, regardless of nationality and age.” He also added that “the scarcity in AI talent has led to Silicon Valley firms taking an interest in undergraduate students as well.” Companies are quickly finding students with research achievements and potential through internship programs. Like this, global IT firms can maintain their dominance in the search for talent due to their high annual income level, stock options, and company name value as industry leaders. A researcher that has just joined a large Silicon Valley company is guaranteed an average annual income of 130,000 US dollars (approximately 150 million Korean won) and significant stock options. Those with outstanding research accomplishments are also provided visas offered for scouted top intellects. Dr. Ga-Won Kim and Dr. Yoo-Jung Shim, both KAIST graduates and currently working at Intel and Apple, decided to work in the US after being offered scout visas. /Reporter Min-Jeong Jeong jminj＠sedaily.com >> Link to the original news article http://www.sedaily.com/NewsView/1OJYSFXHXO/

[iRobot News] What is Robot Te..

On the 23rd, KAIST announced the Top 10 Core Patent Technologies of KAIST for 2017, which are technologies with potential for immediate commercialization amongst the technologies that are expected to lead the 4th Industrial Revolution. The university said that the top 10 technologies were selected from outstanding technologies patented through research and development by taking into consideration the applicability to various fields, market size, and technological innovation with a focus on the technologies that are expected to dominate the 4th Industrial Revolution with significant ripple effects such as ICT (Information and Communications Technology), unmanned transportation, artificial intelligence, robotics, IoT (Internet of Things), nanotechnology, and big data analysis. The selected technologies will receive support for prototype manufacturing through business model development, priority domestic and international marketing, and IP R&D analysis. A large scale presentation event will be held on September 12 at the Seoul COEX, inviting about 200 people including company officials interested in the top 10 core patent technologies. This event is a part of the technology commercialization innovation that KAIST President Sung-Chul Shin presented as one of five innovations for KAIST when he was inaugurated in March this year. The presentation event will provide detailed information on the technologies and on-site consultation regarding technology transfer. Among the announced top 10 technologies, technologies that can be applied to robotics and autonomous driving vehicles are introduced here. ▲Digital lifeform generation system and control method for emotion-based conversations depending on the emotions of the user (Professor Jong-Hwan Kim) Developed by Professor Kim, the digital lifeform technology that provides emotional services according to the emotions of the user analyzes the external, vocal, and inclination data for the user and stores them to a digital DNA. A new digital lifeform is generated based on the digital DNA and an emotional service is provided to the user through the generated lifeform optimized for the user. Conversations with stories are possible between the user and the digital lifeform, and hence it is expected that the technology can be applied to hardware training robots, entertainment, and healthcare fields for treatment of depression and sense of loneliness. ▲Laser integrated precision measurement system for smart factory implementation (Professor Seung-Woo Kim) The laser integrated precision measurement system technology for smart factory implementation applies multi-target measurement technology to production facilities to carry out real-time diagnosis for changes of the facilities in the production line and simultaneous diagnoses of multiple facilities using a single measurement device. Various measurement data are then integrated with artificial intelligence to correct the production facility in real-time. Improvement of the measurement system precision can enhance product quality, and the technology can be applied to thermal deformation measurements of large scale precision facilities, alignment of large structures, and monitoring of machine tools. ▲Location recognition and map-making for autonomous driving of indoor and outdoor moving robots (Professor Hyun Myung) Location recognition and map-making technology for autonomous driving of indoor and outdoor moving robots is a high performance navigation technology employing low cost equipment. Since conventional low cost sensors were integrated, response to various environments is possible and the disadvantages of each sensor can be compensated for, resulting in accurate location data collection within an average of 10cm even for dynamic environments. Professor Myung explained that “the conventional technology had large errors for location recognition in dynamic environments and outdoor environments required expensive sensors so it is expected that the location recognition and map-making technology using low cost sensors and applicable to various environments will be in the spotlight.” ▲Miniature, low power, high linearity 5G beamforming IC optimization using variable gain phase shifter (Professor Sung-Chul Hong) This technology is a miniature, low power, high linearity beamforming IC technology with a novel structure for 5G mobile communications. The miniature IC was realized by removing the damper and a variable gain phase shifter was fabricated to minimize the phase error according to the gain variation, resulting in increased efficiency at low power. This 5G mobile communications technology key to the 4th Industrial Revolution is applicable to IoT (Internet of Things) and 5G vehicle-to-everything (V2X). ▲Beamforming based multi-channel radio resource integration for 5G communication capacity increase (Professor Dong-Ho Cho) The beamforming based multi-channel radio resource clustering technology for 5G communications capacity increase is a core 5G mobile communications technology that enables large capacity transmissions of N-fold by integrating N pattern/polarization antennas. Enhanced communication speed, reduced interference, and reduced transmission power consumption are expected effects of this technology, which is applicable to autonomous driving communication infrastructures. Other technologies announced in the top 10 included the ▲healthcare IoT (Internet of Things) gas sensor for early diagnosis of patients using their outbreathing(Professor Il-Doo Kim), ▲Personalized cancer therapy using big data and body network simulation (Professor Gwang-Hyun Cho), ▲High sensitivity strain sensor for human body motion detection (Professor Oh-Ok Park), ▲Hradware based chip & flash memory data security (Professor Yang-Kyu Choi), ▲Near-infrared neuroimaging bio healthcare device (Professor Hyun-Min Bae). Kil-Soo Jang ksjang＠irobotnews.com >> Link to the original news article http://www.irobotnews.com/news/articleView.html?idxno=11515

[iRobot News] Successful Machi..

▲Research progress is presented at the 2017 1st Machine Intelligence Learning Workshop held at the KAIST Institute main hall in Daejeon. A workshop was successfully held, overviewing the progress of the machine intelligence learning algorithm research project for the Artificial Intelligence Flagship Project “Digital Companion” through support from the Ministry of Science, ICT and Future Planning, and searching for future directions with regard to the project. The 2017 1st Machine Intelligence Learning Workshop was held at the KAIST Institute main hall in Daejeon on the past 14th where 250 participants were present including research and development project personnel, company officials, and participating researchers. ▲ Participants listen to the lecture on machine intelligence learning. A total of 22 research topics were presented for 10 minutes each. Despite the short research period, in-depth research progress was presented covering issues including adaptability of the current machine learning algorithm for new learning data and autonomy to learn independently without human interference. The presentation hall became host to the sharing and exchange of new artificial intelligence technologies and theories that change rapidly. ▲ Project Head Jong-Hwan Kim KAIST MIR MSREP (Machine Intelligence and Robotics Multi-Sponsored Research and Education Platform) Head Jong-Hwan Kim, who oversees the entire project, said, “We aim to develop practical machine intelligence learning algorithms that can be applied to the digital companion platform to manage our everyday lives - learning models based on long term memory, algorithms that learn human intelligence, automatic differentiation of objects, and technologies that can recognize various situations that can occur in everyday life through interaction between people and digital companions - and plan to transfer them to companies.” ▲ Ministry of Science, ICT and Future Planning (MSIP) Institute for Information and Communications Technology Promotion (IITP) SW Computing CP Hyung-Chul Kim Ministry of Science, ICT and Future Planning (MSIP) Institute for Information and Communications Technology Promotion (IITP) SW Computing CP Hyung-Chul Kim, who planned and oversees the digital companion development project, introduced the vision of the future for the digital companion and the core missions that need to be solved in his address at the workshop as well as expressing high expectations for the development of the learning algorithm currently underway at KAIST. >> Link to the original news article http://www.irobotnews.com/news/articleView.html?idxno=10399

[iRobot News] First Machine In..

Hosted by the Machine Intelligence and Robotics Multi-Sponsored Research and Education Platform (MIR MSREP Head: KAIST Professor Jong-Hwan Kim), the 1st Machine Intelligence Learning Open Workshop will be held on the coming 14th in the 2nd Floor Matrix Hall of the KAIST KI Building in Daejeon. MIR MSREP was founded by the KAIST College of Engineering Information and Electrical Research Center based on the Intelligence Information Flagship Research and Development Project (2016.12 ~ 2020.12) funded by the Ministry of Science, ICT and Future Planning and Institute for Information and Communications Technology Promotion (IITP). This workshop will commemorate the establishment of the research and education platform. The Intelligence Information Flagship Research and Development Project aims to develop adaptive machine learning technology that autonomously expands through learning of collected user and environmental multi-modal data from continuous interaction between the user and a digital companion in everyday life situations. The fundamental goal lies in the development of machine intelligence learning technology, which is an intelligence learning that simulates the thought process of humans by computers or robots. MIR MSREP plans to pursue various endeavors including support of participating companies in the platform, international joint research, and cultivation of KAIST master’s and doctorate program students in the field of AI, bringing together theory and skills through research and development as well as agreements with companies that seek AI human resources. Companies, research institutes, and students following AI technology are invited to participate in this workshop. Participation applications are taken until the 11th and applicants can submit the applications to hywoo＠rit.kaist.ac.kr with the following information: name/affiliation/position/contact number/email address.