For more details on the courses, please refer to the Course Catalog
| Code | Course Title | Credit | Learning Time | Division | Degree | Grade | Note | Language | Availability |
|---|---|---|---|---|---|---|---|---|---|
| SUP5001 | Deep Neural Networks: Theory and Applications | 3 | 6 | Major | Master/Doctor | 1-4 | - | No | |
| This course covers deep learning based on artificial neural network which is advanced on various industrial. This course, especially, give students basic understanding of modern neural networks and their applications in computer vision and natural language understanding. Students learn about Convuloutional networks, RNNs, LSTM, Dropout and more. This course introduceds the major technology trens driving Deep Learning. | |||||||||
| SUP5002 | Computer Architecture and Its Applications to Artificial Intelligence | 3 | 6 | Major | Master/Doctor | 1-4 | - | No | |
| This course focuses on principles and mechanisms related to the modern computer architecture including numerical representation, arithmetic operations, datapath and pipelining, cache hierarchies, memory systems, storage and I/O systems. As an application of computer architecture, this course also covers recent HW architectures and techniques for efficient training and inference of AI (especially deep learning) models. | |||||||||
| SUP5003 | Machine Learning Algorithms and Applications | 3 | 6 | Major | Master/Doctor | 1-4 | English | Yes | |
| This course introduces various machine learning models for supervised & unsupervised learning, in order to motivate students who are new to machine learning. In addition, it provides assignments for hands-on practice using scikit-learn (machine learning library operating on python), so that students have chance to apply machine learning models to particular problems with real-world datasets. Regarding supervised learning, we begin with linear regression and introduce ridge regression where L2 regularization method is applied to alleviate over-fitting. Logistic regression is introduced, which is a popular tool for binary classification. k-nearest classifier and Naive Bayes classifier with its application to text classification are also introduced. Tree-based models such as CART is introduced, ensemble of trees, including random forest and boosting are covered. Finally support vector machines, which are large-margin classifiers, are introduced to complete the supervised learning topic. Regarding unsupervised learning, we introduce two clustering methods, including k-means clustering and mixture of Gaussians. For dimensionality reduction, we provide principal component analysis (PCA), nonnegative matrix factorization (NMF), and stochastic neighborhood embedding (SNE). | |||||||||
| SUP5004 | Neural Interface and Application | 3 | 6 | Major | Master/Doctor | 1-4 | Korean | Yes | |
| The course covers the material, geometry, and other necessary property of neural interface that are used in diagnosis and clinical applications. For example, one topic is to discuss and compare various way of stimulation such as electrical, optical, pharmacological, at both central and peripheral nerve system. The one goal of the course is to develop the ability of solving issues via various engineering aspects. | |||||||||
| SUP5005 | Cloud Systems For Machine Learning | 3 | 6 | Major | Master/Doctor | 1-4 | - | No | |
| Cloud systems are being used as a core infrastructure for training and inferencing neural networks, which require a large amount of computing resources. This course deals with the core elements of distributed processing and cloud system technology for artificial intelligence and their recent technology trends. Specifically, students will learn the cloud system interface, orchestration technology, and virtualization technology, and learn the principles of distributed learning, the design of distributed learning frameworks, and the design and implementation of a distributed learning cluster in a cloud system. | |||||||||
| SUP5006 | Mathematics for Machine Learning | 3 | 6 | Major | Master/Doctor | 1-4 | English | Yes | |
| This course introduces a few mathematics which is essential for machine learning, to help students to better understand various machine learning models. Regarding linear algebra, we begin with vectors and matrices and cover linear algebraic equations, inner product, vector norms, orthogonal projection. We also introduce a popular matrix decomposition, like SVD and explain how such method is used in machine learning. Regarding probability and distributions, we introduce random variables, expected values, and two exemplar distributions, including Gaussian for continuous random variables and Bernoulli for bianry random variables. Regarding parameter estimation, we introduce maxinum likelihood and MAP estimation methods. Regarding information theory, we introduce entropy, mutual information, KL-divergence, and explain how these are used in machine learning. Regarding continuous optimization, we begin with vector calculus, and introduce two iterative methods such as gradient descdent/ascent, Newton’s method. Finally we conclude this course, having a look at a few machine learning models to understand how relevant mathematics are utilized in machine learning. | |||||||||
| SUP5007 | Intelligent biomedical sensing system design | 3 | 6 | Major | Master/Doctor | - | No | ||
| 1. Acquisition of bio-signals using physical sensors The physical sensor will utilize a resistive strain sensor and can be manufactured using a flexible composite material. As a skin-attachable sensor, students should come up with application target of using the sensor on the skin. The learning contents are as follows. 1) Transfer of general knowledge of skin-attached sensors 2) Deduction of resistance sensor manufacturing and utilization plan 2. Biosignal acquisition and analysis using OpenBCI OpenBCI is an open source hardware device that acquires biosignals based on TI's ADS1299 chip. This course aims to learn the basics of biosignal processing and wirelessly transmitting them to a PC using OpenBCI's Cyton board. By connecting the manufactured bio-sensor to the OpenBCI system, it enables analysis using machine learning. The learning contents are as follows. 1) Practice of configuring OpenBCI hardware and interworking with biosensors 2) Transmission of bio-signals through wireless communication 3) Transmission of biosignals to Python using open source libraries such as Brainflow 4) Basic digital signal processing technology such as filtering for biosignal conditioning 3. By convergence of bio-fusion sensor and artificial intelligence technology, we utilizes an artificial intelligence model suitable for the target application (collect/process the necessary dataset, learn the artificial intelligence model, and then apply to verify function and performance.). T | |||||||||
| SUP5008 | System Intelligence | 3 | 6 | Major | Master/Doctor | Korean | Yes | ||
| With the rapid development of AI and broad adoption of AI applications, it is required to have the system structure that can optimally support model learning and inference at scale. This application and system technology is also evolving in the direction of automation, efficiency, and optimization, leveraging data-driven machine learning technology. Based on interdisciplinary works on systems, networks, and machine learning, this course discusses (1) various AI-based methods to address system problems and (2) network and system configuration techniques required to apply AI to real-world problems. The schedule for the course is as follows. ● Weeks 1-5: Machine learning for system problem solving (e.g., reinforcement learning, meta learning, imitation learning, self-supervised learning.) ● Weeks 6-10: Machine learning-based approaches to system problems (e.g., automation, resource management, scheduling.) ● Week 11-15: System structures for real-world AI systems (e.g., distributed learning platform, model compression, acceleration, etc.) | |||||||||
| SUP5009 | Bioelectronic Devices and Intelligent Information Processing | 3 | 6 | Major | Master/Doctor | 1-2 | - | No | |
| The course covers the flexible electronic/bioelectronic materials and functional devices. Furthermore, various fabrication methods and characteristic analyses regarding soft electrodes, sensors, and non-volatile memories will be addressed. In addition, this course covers diverse artificial intelligence applications for effectively achieving high-fidelity physiological signals and biomedical images. For example, one topic is to discuss recent progress and limitation of sensors, memory devices, and artificial intelligence and predict their future prospects. The one goal of the course is to develop the ability of solving issues via various engineering aspects. | |||||||||
| SUP6001 | Directed study 1 | 3 | 6 | Major | Doctor | Korean | Yes | ||
| This course offers a directed and independent study for Ph.D degree students on the first level. Most subjects assigned to students are related to the Ph.D thesis topic for each student, which include literature survey, directed experiments, data analysis, and writing proposal. The aim of class is to provide various conceptual backgrounds for the research, current problem and research needs, experimental design, and analysis for the results, which may not be offered by regular classes in the graduate program. Grade is given by the research advisor of students based on subjective evaluation for discussion and/or written reports. | |||||||||
| SUP6002 | Directed study 2 | 3 | 6 | Major | Doctor | Korean | Yes | ||
| This course offers a directed and independent study for Ph.D degree students on the second level. Most subjects assigned to students are related to the Ph.D thesis topic for each student, which include literature survey, directed experiments, data analysis, and writing proposal. The aim of class is to provide various conceptual backgrounds for the research, current problem and research needs, experimental design, and analysis for the results, which may not be offered by regular classes in the graduate program. Grade is given by the research advisor of students based on subjective evaluation for discussion and/or written reports. | |||||||||
| SUP7001 | AI Capstone Design 1 | 3 | 6 | Major | Bachelor/Master/Doctor | Korean | Yes | ||
| Capstone Design is an educational program that enhances creative problem-solving skills through the entire process of planning, designing, and producing a work based on the theories learned at university in order to solve problems encountered by students in actual industrial settings. AI Capstone Design 1 aims to develop a team of students with various majors to design AI application tasks required by society and industry, and to strengthen creative and comprehensive design capabilities. Students learn various techniques used in design, learn how to apply theory through practice, and cultivate engineering problem-solving skills. | |||||||||
| SUP7002 | AI Capstone Design 2 | 3 | 6 | Major | Bachelor/Master/Doctor | - | No | ||
| AI Capstone Design 2 is a convergence education program that allows students to experience communication and cooperation with the purpose of cultivating creativity, practical skills, teamwork and leadership by solving problems specified or linking companies in the field of AI. Team based students collaborate with mentors and advisory professors to devise ideas for solving engineering problems based on effective teamwork in performing team projects. The purpose of excellent works is to link them to commercialization or start-up through patents, support for entry to domestic and international competitions, and business feasibility review. | |||||||||