machine+learning
-
Streamlining the Process of Materials Discovery
The materials platform M3I3 reduces the time for materials discovery by reverse engineering future materials using multiscale/multimodal imaging and machine learning of the processing-structure-properties relationship
Developing new materials and novel processes has continued to change the world. The M3I3 Initiative at KAIST has led to new insights into advancing materials development by implementing breakthroughs in materials imaging that have created a paradigm shift in the discovery of materials. The Initiative features the multiscale modeling and imaging of structure and property relationships and materials hierarchies combined with the latest material-processing data.
The research team led by Professor Seungbum Hong analyzed the materials research projects reported by leading global institutes and research groups, and derived a quantitative model using machine learning with a scientific interpretation. This process embodies the research goal of the M3I3: Materials and Molecular Modeling, Imaging, Informatics and Integration.
The researchers discussed the role of multiscale materials and molecular imaging combined with machine learning and also presented a future outlook for developments and the major challenges of M3I3. By building this model, the research team envisions creating desired sets of properties for materials and obtaining the optimum processing recipes to synthesize them.
“The development of various microscopy and diffraction tools with the ability to map the structure, property, and performance of materials at multiscale levels and in real time enabled us to think that materials imaging could radically accelerate materials discovery and development,” says Professor Hong.
“We plan to build an M3I3 repository of searchable structural and property maps using FAIR (Findable, Accessible, Interoperable, and Reusable) principles to standardize best practices as well as streamline the training of early career researchers.”
One of the examples that shows the power of structure-property imaging at the nanoscale is the development of future materials for emerging nonvolatile memory devices. Specifically, the research team focused on microscopy using photons, electrons, and physical probes on the multiscale structural hierarchy, as well as structure-property relationships to enhance the performance of memory devices.
“M3I3 is an algorithm for performing the reverse engineering of future materials. Reverse engineering starts by analyzing the structure and composition of cutting-edge materials or products. Once the research team determines the performance of our targeted future materials, we need to know the candidate structures and compositions for producing the future materials.”
The research team has built a data-driven experimental design based on traditional NCM (nickel, cobalt, and manganese) cathode materials. With this, the research team expanded their future direction for achieving even higher discharge capacity, which can be realized via Li-rich cathodes.
However, one of the major challenges was the limitation of available data that describes the Li-rich cathode properties. To mitigate this problem, the researchers proposed two solutions: First, they should build a machine-learning-guided data generator for data augmentation. Second, they would use a machine-learning method based on ‘transfer learning.’ Since the NCM cathode database shares a common feature with a Li-rich cathode, one could consider repurposing the NCM trained model for assisting the Li-rich prediction. With the pretrained model and transfer learning, the team expects to achieve outstanding predictions for Li-rich cathodes even with the small data set.
With advances in experimental imaging and the availability of well-resolved information and big data, along with significant advances in high-performance computing and a worldwide thrust toward a general, collaborative, integrative, and on-demand research platform, there is a clear confluence in the required capabilities of advancing the M3I3 Initiative.
Professor Hong said, “Once we succeed in using the inverse “property−structure−processing” solver to develop cathode, anode, electrolyte, and membrane materials for high energy density Li-ion batteries, we will expand our scope of materials to battery/fuel cells, aerospace, automobiles, food, medicine, and cosmetic materials.”
The review was published in ACS Nano in March. This study was conducted through collaborations with Dr. Chi Hao Liow, Professor Jong Min Yuk, Professor Hye Ryung Byon, Professor Yongsoo Yang, Professor EunAe Cho, Professor Pyuck-Pa Choi, and Professor Hyuck Mo Lee at KAIST, Professor Joshua C. Agar at Lehigh University, Dr. Sergei V. Kalinin at Oak Ridge National Laboratory, Professor Peter W. Voorhees at Northwestern University, and Professor Peter Littlewood at the University of Chicago (Article title: Reducing Time to Discovery: Materials and Molecular Modeling, Imaging, Informatics, and Integration).This work was supported by the KAIST Global Singularity Research Program for 2019 and 2020.
Publication:
“Reducing Time to Discovery: Materials and Molecular Modeling, Imaging, Informatics and Integration,” S. Hong, C. H. Liow, J. M. Yuk, H. R. Byon, Y. Yang, E. Cho, J. Yeom, G. Park, H. Kang, S. Kim, Y. Shim, M. Na, C. Jeong, G. Hwang, H. Kim, H. Kim, S. Eom, S. Cho, H. Jun, Y. Lee, A. Baucour, K. Bang, M. Kim, S. Yun, J. Ryu, Y. Han, A. Jetybayeva, P.-P. Choi, J. C. Agar, S. V. Kalinin, P. W. Voorhees, P. Littlewood, and H. M. Lee, ACS Nano 15, 3, 3971–3995 (2021) https://doi.org/10.1021/acsnano.1c00211
Profile:
Seungbum Hong, PhD
Associate Professor
seungbum@kaist.ac.kr
http://mii.kaist.ac.kr
Department of Materials Science and Engineering
KAIST
(END)
2021.04.05 View 16023 -
Taesik Gong Named Google PhD Fellow
PhD candidate Taesik Gong from the School of Computing was named a 2020 Google PhD Fellow in the field of machine learning.
The Google PhD Fellowship Program has recognized and supported outstanding graduate students in computer science and related fields since 2009. Gong is one of two Korean students chosen as the recipients of Google Fellowships this year. A total of 53 students across the world in 12 fields were awarded this fellowship.
Gong’s research on condition-independent mobile sensing powered by machine learning earned him this year’s fellowship. He has published and presented his work through many conferences including ACM SenSys and ACM UbiComp, and has worked at Microsoft Research Asia and Nokia Bell Labs as a research intern. Gong was also the winner of the NAVER PhD Fellowship Award in 2018.
(END)
2020.10.15 View 14110 -
Quantum Classifiers with Tailored Quantum Kernel
Quantum information scientists have introduced a new method for machine learning classifications in quantum computing. The non-linear quantum kernels in a quantum binary classifier provide new insights for improving the accuracy of quantum machine learning, deemed able to outperform the current AI technology.
The research team led by Professor June-Koo Kevin Rhee from the School of Electrical Engineering, proposed a quantum classifier based on quantum state fidelity by using a different initial state and replacing the Hadamard classification with a swap test. Unlike the conventional approach, this method is expected to significantly enhance the classification tasks when the training dataset is small, by exploiting the quantum advantage in finding non-linear features in a large feature space.
Quantum machine learning holds promise as one of the imperative applications for quantum computing. In machine learning, one fundamental problem for a wide range of applications is classification, a task needed for recognizing patterns in labeled training data in order to assign a label to new, previously unseen data; and the kernel method has been an invaluable classification tool for identifying non-linear relationships in complex data.
More recently, the kernel method has been introduced in quantum machine learning with great success. The ability of quantum computers to efficiently access and manipulate data in the quantum feature space can open opportunities for quantum techniques to enhance various existing machine learning methods.
The idea of the classification algorithm with a nonlinear kernel is that given a quantum test state, the protocol calculates the weighted power sum of the fidelities of quantum data in quantum parallel via a swap-test circuit followed by two single-qubit measurements (see Figure 1). This requires only a small number of quantum data operations regardless of the size of data. The novelty of this approach lies in the fact that labeled training data can be densely packed into a quantum state and then compared to the test data.
The KAIST team, in collaboration with researchers from the University of KwaZulu-Natal (UKZN) in South Africa and Data Cybernetics in Germany, has further advanced the rapidly evolving field of quantum machine learning by introducing quantum classifiers with tailored quantum kernels.This study was reported at npj Quantum Information in May.
The input data is either represented by classical data via a quantum feature map or intrinsic quantum data, and the classification is based on the kernel function that measures the closeness of the test data to training data.
Dr. Daniel Park at KAIST, one of the lead authors of this research, said that the quantum kernel can be tailored systematically to an arbitrary power sum, which makes it an excellent candidate for real-world applications.
Professor Rhee said that quantum forking, a technique that was invented by the team previously, makes it possible to start the protocol from scratch, even when all the labeled training data and the test data are independently encoded in separate qubits.
Professor Francesco Petruccione from UKZN explained, “The state fidelity of two quantum states includes the imaginary parts of the probability amplitudes, which enables use of the full quantum feature space.”
To demonstrate the usefulness of the classification protocol, Carsten Blank from Data Cybernetics implemented the classifier and compared classical simulations using the five-qubit IBM quantum computer that is freely available to public users via cloud service. “This is a promising sign that the field is progressing,” Blank noted.
Link to download the full-text paper:
https://www.nature.com/articles/s41534-020-0272-6
-Profile
Professor June-Koo Kevin Rhee
rhee.jk@kaist.ac.kr
Professor, School of Electrical Engineering
Director, ITRC of Quantum Computing for AIKAIST
Daniel Kyungdeock Parkkpark10@kaist.ac.krResearch Assistant ProfessorSchool of Electrical EngineeringKAIST
2020.07.07 View 14214 -
Professor Jong Chul Ye Appointed as Distinguished Lecturer of IEEE EMBS
Professor Jong Chul Ye from the Department of Bio and Brain Engineering was appointed as a distinguished lecturer by the International Association of Electrical and Electronic Engineers (IEEE) Engineering in Medicine and Biology Society (EMBS). Professor Ye was invited to deliver a lecture on his leading research on artificial intelligence (AI) technology in medical video restoration. He will serve a term of two years beginning in 2020.
IEEE EMBS's distinguished lecturer program is designed to educate researchers around the world on the latest trends and technology in biomedical engineering. Sponsored by IEEE, its members can attend lectures on the distinguished professor's research subject.
Professor Ye said, "We are at a time where the importance of AI in medical imaging is increasing.” He added, “I am proud to be appointed as a distinguished lecturer of the IEEE EMBS in recognition of my contributions to this field.”
(END)
2020.02.27 View 13630 -
Professor Minsoo Rhu Recognized as Facebook Research Scholar
Professor Minsoo Rhu from the School of Electrical Engineering was selected as the recipient of the Systems for Machine Learning Research Awards presented by Facebook.
Facebook launched the award last year with the goal of funding impactful solutions in the areas of developer tookits, compilers and code generation, system architecture, memory technologies, and machine learning accelerator support.
A total of 167 scholars from 100 universities representing 26 countries submitted research proposals, and Facebook selected final 10 scholars. Professor Rhu made the list with his research topic ‘A Near-Memory Processing Architecture for Training Recommendation Systems.’ He will receive 5,000 USD in research funds at the award ceremony which will take place during this year’s AI Systems Faculty Summit at the Facebook headquarters in Menlo Park, California.
Professor Rhu’s submission was based on research on ‘Memory-Centric Deep Learning System Architecture’ that he carried out for three years under the auspices of Samsung Science and Technology Foundation from 2017. It was an academic-industrial cooperation research project in which leading domestic companies like Samsung Electronics and SK Hynix collaborated to make a foray into the global memory-centric smart system semiconductor market.
Professor Rhu who joined KAIST in 2018 has led various systems research projects to accelerate the AI computing technology while working at NVIDIA headquarters from 2014.
(END)
2020.02.21 View 12005 -
Object Identification and Interaction with a Smartphone Knock
(Professor Lee (far right) demonstrate 'Knocker' with his students.)
A KAIST team has featured a new technology, “Knocker”, which identifies objects and executes actions just by knocking on it with the smartphone. Software powered by machine learning of sounds, vibrations, and other reactions will perform the users’ directions.
What separates Knocker from existing technology is the sensor fusion of sound and motion. Previously, object identification used either computer vision technology with cameras or hardware such as RFID (Radio Frequency Identification) tags. These solutions all have their limitations. For computer vision technology, users need to take pictures of every item. Even worse, the technology will not work well in poor lighting situations. Using hardware leads to additional costs and labor burdens.
Knocker, on the other hand, can identify objects even in dark environments only with a smartphone, without requiring any specialized hardware or using a camera. Knocker utilizes the smartphone’s built-in sensors such as a microphone, an accelerometer, and a gyroscope to capture a unique set of responses generated when a smartphone is knocked against an object. Machine learning is used to analyze these responses and classify and identify objects.
The research team under Professor Sung-Ju Lee from the School of Computing confirmed the applicability of Knocker technology using 23 everyday objects such as books, laptop computers, water bottles, and bicycles. In noisy environments such as a busy café or on the side of a road, it achieved 83% identification accuracy. In a quiet indoor environment, the accuracy rose to 98%.
The team believes Knocker will open a new paradigm of object interaction. For instance, by knocking on an empty water bottle, a smartphone can automatically order new water bottles from a merchant app. When integrated with IoT devices, knocking on a bed’s headboard before going to sleep could turn off the lights and set an alarm. The team suggested and implemented 15 application cases in the paper, presented during the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp 2019) held in London last month.
Professor Sung-Ju Lee said, “This new technology does not require any specialized sensor or hardware. It simply uses the built-in sensors on smartphones and takes advantage of the power of machine learning. It’s a software solution that everyday smartphone users could immediately benefit from.” He continued, “This technology enables users to conveniently interact with their favorite objects.”
The research was supported in part by the Next-Generation Information Computing Development Program through the National Research Foundation of Korea funded by the Ministry of Science and ICT and an Institute for Information & Communications Technology Promotion (IITP) grant funded by the Ministry of Science and ICT.
Figure: An example knock on a bottle. Knocker identifies the object by analyzing a unique set of responses from the knock, and automatically launches a proper application or service.
2019.10.02 View 30910 -
Policy Debate Series for Industry 4.0
(Photo caption: President Shin takes the podium as the first speaker of a year-long monthly policy dabate series on Industry 4.0 on May 11.)
KAIST will kick off a monthly policy debate series on Industry 4.0 every Thursday from May 11 at the Startup KAIST building. The year-long series, featuring professors from key technology fields associated with Industry 4.0, is designed to help policy makers from government, industry, and research institutes respond better to the ramifications that Industry 4.0 brings about in each sector.
The series will help them establish the vision and strategy that will work for the new industrial environment to take the lead in the new industrial era.
Twelve professors, including President Sung-Chul Shin, from departments that are researching emerging technologies will speak on the megatrend of new technology, while facilitating debates and Q& A sessions with participants.
The participants will include officials from the government complexes in Sejong and Daejeon cities, government-funded research institutes in Daejeon, and businessmen, among others. For registration, please go to https://startup.kaist.ac.kr/register.
Schedule
Speaker
Theme
May 11
President Sung-Chul Shin
Challenges and Innovations of KAIST in the Era of Industry 4.0
June 8
Professor Jonghwan Kim
Machine Intelligence and Deep Learning
July 6
Professor Jun Ho Oh
Robot Technology and the Future
Aug. 3
Professor Hyunchul Shim
Unmanned Vehicle Technology and Industry 4.0
Sept. 7
Professor Hawoong Jeong
Complex Systems and Data Science
Oct. 12
Professor Yongdae Kim
Technology, Policy, and the Fostering of Talents: Industry 4.0 and Information Protection
Nov. 9
Professor Sang Yup Lee
The Role of Biotechnology in Industry 4.0
Dec. 7
Professor Meeyoung Cha
AI-Based Research for Fake News Detection
2018 Jan. 4
Professor Joungho Kim
Innovation for the Korean Semiconductor Industry: Kim’s Law
Feb. 8
Professor Jaekyun Moon
Education for Industry 4.0
March 8
Professor Sang Kil Cha
Artificial Intelligence Cyber Warfare: Its Present and Future
April 5
Professor Jaeseung Jeong
The Future of Brain Engineering and Artificial Intelligence
2017.05.08 View 12481 -
Professor Jinwoo Shin of the Electrical Engineering Department Receives the 2015 ACM SIGMETRICS Rising Star Research Award
Professor Jinwoo Shin of the Electrical Engineering Department at KAIST was selected as the recipient of the 2015 ACM SIGMETRICS Rising Star Research Award.
As a computer systems performance evaluation community, SIGMETRICS annually awards a junior researcher. He was selected as the 8th annual recipient, being the first from an Asian university.
Professor Shin was recognized for his work in theoretical analysis of stochastic queueing networks and machine learning. He said, “I would like to contribute to the expansion of computing and network theory in Korea wherein those fields are unrecognized.”
He has received numerous awards including Kennneth C. Sevcik (Best Student Paper) Award at SIGMETRICS 2009, George M. Sprowls (Best MIT CS PhD Thesis) Award 2010, Best Paper Award at MOBIHOC 2013, Best Publication Award from INFORMS Applied Probability Society 2013, and Bloomberg Scientific Research Award 2015.
2015.06.21 View 11431 -
Professor Jinwoo Shin Receives the Bloomberg Scientific Research Award
Professor Jinwoo Shin (https://sites.google.com/site/mijirim/) of the Electrical Engineering Department at KAIST has been selected as one of the three winners to receive the first Bloomberg Scientific Research Award this month. The newly created award is presented to researchers in computer science who conduct high-quality research in such areas as machine learning, natural language processing, machine translation, statistics, and theory.
Professor Shin submitted his research proposal entitled “Scalable Probabilistic Deep Leaning,” and the award will support funding his research for one year. For details, please click on the link below for an article released by Bloomberg News, announcing the winners of the award:
Bloomberg News, April 28, 2015
“Announcing the Winners of the Bloomberg’s First Scientific Research Program”
https://3blmedia.com/News/Announcing-Winners-Bloombergs-First-Scientific-Research-Program
2015.04.30 View 9636