Start: 2
Optimization techniques are used in all kinds of machine learning problems because in general we would like to minimize the testing error. This course will contain two parts. The first part focuses on convex optimization techniques. We discuss methods for least-squares, linear and quadratic programs, semidefinite programming, and others. We also touch theory behind these methods (e.g., optimality conditions and duality theory). In the second part of this course we will investigate how optimization techniques are applied to various machine learning problems (e.g., SVM, maximum entropy, conditional random fields, sparse reconstruction for signal processing applications). We further discuss that for different machine learning applications how to choose right optimization methods. learn how to use optimization techniques for solving machine learning problems. College of Electrical Engineering & Computer Science Main Campus Chih-Jen Lin 80 Tuesday 2,3,4 CSIE7435 (922EU3940) 3 (College of Electrical Engineering and Computer Science) Graduate Institute of Networking and Multimedia,
(College of Electrical Engineering and Computer Science) Graduate Institute of Computer Science & Information Engineering http://www.csie.ntu.edu.tw/main.php?lang=en
Tentative Course Outline: 1. Experiments, Models, and Probabilities 1.1. Applying Set Theory to Probability 1.2. Probability Axioms 1.3. Some Consequences of the Axioms 1.4. Conditional Probability 1.5. Independence 1.6. Sequential Experiments and Tree Diagrams 2. Random Variables 2.1. Definitions 2.2. Probability Mass Function 2.3. Families of Discrete Random Variables 2.4. Cumulative Distribution Function (CDF) 2.5. Probability Density Function 2.6. Families of Continuous Random Variables 3. Random Variables and Expected Value 3.1. Conditional Probability Mass/Density Function 3.2. Probability Models of Derived Random Variables 3.3. Average 3.4. Variance and Standard Deviation 3.5. Expected Value of a Derived Random Variable Midterm exam 4. Random Vectors 4.1. Probability Models of N Random Variables 4.2. Vector Notation 4.3. Joint Cumulative Distribution Function 4.4. Joint Probability Mass/Density Function 4.5. Marginal PMF/PDF 4.6. Functions of Two Random Variables (Jacobian Transformation) 4.7. Conditioning by a Random Variables 4.8. Bivariate Gaussian Random Variables 4.9. Correlation Matrix 5. Sums of Random Variables 5.1. Expected Values of Sums 5.2. PDF of the Sum of Two Random Variables 5.3. Moment Generating Functions 5.4. MGF of the Sum of Independent Random Variables 5.5. Random Sums of Independent Random Variables 5.6. Central Limit Theorem 5.7. Applications of the Central Limit Theorem 5.8. The Chernoff Bound 6. Parameter Estimation Using the Sample Mean 6.1. Sample Mean: Expected Value and Variance 6.2. Deviation of a Random Variable from the Expected Value 6.3. Point Estimates of Model Parameters 6.4. Confidence Intervals 7. Hypothesis Testing 7.1. Significance Testing 7.2. Binary Hypothesis Testing Final exam To introduce to students the theory, models and analysis of probability and basic statistics and their applications with emphasis on electrical and computer engineering problems. College of Electrical Engineering & Computer Science Main Campus Calculus (A) 1 & 2
Grading: Homework : 20%, Midterm : 40%, Final : 40%, Participation 5% Shi Chung Chang 50 Monday 4 Thursday 8,9 EE2007 (901E21000) 3 (College of Electrical Engineering and Computer Science) Department of Electrical Engineering http://www.ee.ntu.edu.tw/en/
[Course description] Control is the action of causing a system variable to approach some desired value. It is also a fundamental and universal problem-solving approach in many traditional and interdisciplinary fields. A control system, in a very general sense, is a system with an (reference) input that can be applied per the desired value and an output from which how well the system variable matches to the desired value (e.g., errors) can be determined. It can be found in daily life, almost all engineering disciplines, and even biological and social studies. For examples, bicycle riding involves with a control system comprising of a bicycle and a rider, with inputs and outputs associated with the desired attitude, speed, and direction of the bicycle. Temperature control systems have applications in household, automobile, aerospace, office, factory, and agriculture environments. Motion control systems are critical to factory automation and precision instruments, such as industrial robots, atomic-force microscopes, and step-and-scan photolithography exposure systems. Many modern cameras equip with autofocus and vibration compensation systems to minimize image blur. Many kinds of circuits such as phase lock loops, operational amplifiers, and voltage regulators rely on control to ensure their functions and performance. A living body is a complex control system where many critical variables such as heart beat rate, blood pressure, and body temperature are regulated constantly for health. Central banks of most countries around the world set interest rates as a way to control inflation. This undergraduate course is designed for junior and senior (3rd/4th yr.) students to apprehend basic modeling, simulation, analysis, and design techniques for control systems. It intends to cover fundamentals of “classical control” that primarily focuses on frequency domain feedback control approaches for single-input-single-output linear dynamical systems. When time permits, some essential elements in modern-day control engineering such as state-space approaches, discrete-time digital control, and numerical methods will also be introduced. [Course goals] Basic: – Awareness of the strength and the importance of control systems, especially the effectiveness of feedback – Ability of deriving dynamic models and simulating dynamic responses – Ability of analyzing and designing feedback controllers for linear SISO systems in the frequency domain using root locus and frequency response techniques Bonus: – Awareness of some advanced control topics (e.g., state-space methods, digital control, and nonlinear systems) – Development of technical writing skills in English College of Electrical Engineering & Computer Science Main Campus [Prerequisites] Linear algebra, ordinary differential equations, Laplace transforms, fundamental circuit and mechanics analysis — which should have been well covered by several freshman and sophomore (1st/2nd yr.) courses in most electrical and mechanical engineering curriculums. Prior exposure to the analysis of signals and systems will be beneficial but not absolutely required. Kuen-Yu Tsai 60 Thursday 7,8,9 EE3024 (901E43100) 3 Non-degree Program: Education Program For Agricultural Automation,
(College of Electrical Engineering and Computer Science) Department of Electrical Engineering,
Non-degree Program: Transprotation Electrification Technology Program http://www.ee.ntu.edu.tw/en/
This course is on discrete mathematics. It covers combinatorics, boolean logic, computation theory, analysis of algorithms, probability, algebra, number theory, graph theory, set theory, and many other fields. Parts of the book should have been covered in high school and will be skipped or only briefly reviewed. I have in mind basic combinatorics, logic, and basic set theory. This courses prepares students for foundations of computer science and analysis of algorithms. It is also useful for many applications of computers and mathematics, even social sciences. College of Electrical Engineering & Computer Science Main Campus Homeworks. Examinations. Yuh-Dauh Lyuu 50 Thursday 2,3,4 CSIE2122 (902E25200) 3 *Majors-only (including minor and double major students).
(College of Electrical Engineering and Computer Science) Department of Computer Science & Information Engineering http://www.csie.ntu.edu.tw/main.php?lang=en
Adaptive Control SystemsThis course is mainly for graduted students (but not restricted to). We will provide techniques to estimate unknown system parameters, and design the controller for such systems. The main topics are: -Introduction -Identification of System Parameters -Adaptive Control of Linear Systems -Adaptive Control of a Class of Nonlinear Systems -Adaptive Neural Network Control -Adaptive Sliding Mode Control The main objectives are: – Estimate unknow system parameters – Design Adaptive controllers for linear and nonlinear sytems – Analysis of system properties for systems with unknown parameters – Apply the adaptive control techniques to various systems College of Electrical Engineering & Computer Science Main Campus Evaluation: -Homework (every 2~3 weeks) -Final term report -Final oral presentation Li-Chen Fu 28 Wednesday 2,3,4 EE7005 (921EM1380) 3 (College of Electrical Engineering and Computer Science) Graduate Institute of Electrical Engineering http://www.ee.ntu.edu.tw/en/
Nonlinear OpticsPrinciples of nonlinear optics with emphasis on the fundamental aspects of nonlinear optical theory and techniques. To understand the principles of nonlinear optics. To be equipped with the basic ability to analyze a nonlinear optics problem. College of Electrical Engineering & Computer Science Main Campus To understand the basic principles behind different nonlinear optics phenomena. Chi-Kuang Sun 30 Thursday 7,8,9 EE5050 (921EU2310) 3 (College of Electrical Engineering and Computer Science) Graduate Institute of Electrical Engineering,
(College of Electrical Engineering and Computer Science) Graduate Institute of Electro-Optical Engineering,
(College of Electrical Engineering and Computer Science) Graduate Institute of Biomedical Electronics and Bioinfornatics,
Non-degree Program: Program of Photonics Technologies http://www.ee.ntu.edu.tw/en/
.1.Humanoid Robotics — Definition of walking, dynamic and static gaits — ZMP (zero moment point) — Lagrange’s
Equations — Control Algorithms of Humanoid Robotics — Sensors and Sensing Aspects of Humanoid Robotics
2.Mutisensor Fusion and Integration –Sigal Level Fusion –Pixel Level Fusion –Feature Level Fusion –Decision Level
Fusion –Kalman Filter –Extended Kalman Filter –Particle Filter –Covariance Intersection –Covariance Union —Dempster-Shafer Evidence Theory 3.Sensing and Control for Robot Motion 4.Interactive Service Robotics 5.Advanced Topics on Robot Sensing and Control 6.Practical examples of robot sensing and control through photos and video demonstrations. The objectives of this course are to let students who have had the basic background of the robot sensing and control issues, approaches with more in depth understanding of theories and practical applications in robot sensing
and control. The idea of this course is to convey the concept that usually sensing and control should not be
separated and they are interdependent in dealing with an intelligent systems, such as an intelligent robotics system.
Firstly, student will learn more advanced robot sensing and control issues in humanoid robot including definition of
walking, dynamics static gaiting issues, control algorithms and the need for robot sensors interact with different
control aspects. The second focus will be the study of advanced issues of sensor fusion and integration.
Synergistic use of multiple sensors by machines and systems enables greater intelligence to be incorporated into
their overall operation. Motivation for using multiple sensors can be considered as response to simple question:
if a single sensor can increase the capability of a system, would the use of more sensors increase it even further ?
In this course, theories of multisensory fusion and its applications to sensory controlled robotics systems which
involves mathematical and statistical issues including combining sensor uncertainty methods for sensor fusion
includes estimation methods, such as covariance Intersection (CI), Kalman Filtering; Classification methods,
such asSupport Vector Machine (SVM) etc. will be presented and discussed. The third focus will be the advanced
robot motionplanning and control issues. The fourth focus will be the advanced topics in interactive service robotics by using various sensing and control algorithms.Finally, a variety of practical examples of robot sensing and control
will be presented through photos and video demonstrations. After taking this course, it is expected that students
will l get the state of the art knowledge about the advanced core robotics technologies especially in robot sensing
and control. College of Electrical Engineering & Computer Science Main Campus This course is suitable for senior and graduate students. There is a take home project in addition to the weekly class meets. The final grade will be computed on the basis of the following weights: Take Home Project Report. 25% Project Presentation during the Class. 25% Midterm Exam. 25% Final Exam. 25% TOTAL 100% Ren C. Luo 20 Thursday A,B,C EE5155 (921EU4350) 3 (College of Electrical Engineering and Computer Science) Graduate Institute of Electrical Engineering http://www.ee.ntu.edu.tw/en/
This is an introductory course in system identification, the process of developing or improving a mathematical representation of a physical system using experimental data. This course focuses equally on theoretical and practical aspects of the subject. Students will learn key mathematical skills including linear time-invariant systems, random processes, and basic estimation techniques. Practical system identification skills such as input signal design, system excitation, and model validation will also be discussed. Students are required to integrate the knowledge into their works of final projects. 1. Review of linear systems 2. Random variables and random processes 3. Least-square estimation 4. Non-parametric model identification 5. Parametric model identification 6. State-space methods 7. System identification in practice 8. Advanced topics* (subspace identification, time varying or nonlinear systems) 9. Final project presentation College of Electrical Engineering & Computer Science Main Campus Undergraduate-level Control Systems, and/or Signal and Systems. Basic/working knowledge about linear algebra, linear dynamical systems, state-space models, and Fourier, Laplace, and Z-transforms. Kuen-Yu Tsai 15 Thursday 2,3,4 EE5129 (921EU8300) 3 (College of Electrical Engineering and Computer Science) Graduate Institute of Electronics Engineering,
(College of Electrical Engineering and Computer Science) Graduate Institute of Electrical Engineering,
Non-degree Program: Transprotation Electrification Technology Program http://www.ee.ntu.edu.tw/en/
The module will be delivered over one semester, as a blend of small group work and lectures, practical exercises, group project, presentation and in-class discussion of reading tasks. The aim of this course is to introduce concepts of study design, data collection and statistical analysis commonly used in public health research with a strong focus in global health. College of Public Health Downtown Campus-College of Public Health 1. Active participations in the discussion and presentation of reading tasks are requirements for all students. 2. On the completion of this course, students will identify a specific research topic related to global health and use the skills and knowledge taught in the course to undertake a critical review of the literature relating to the identified research topic/problem, design a study to investigate the problem, and write an appropriate protocol for conducting a research project on the topic, including ethical aspects of their research. 3. For the mid-term presentation, each student is required to do a 15-minutes presentation on her/his identified research topic. The content of presentation should include a preliminary report of background, literature search strategy and research hypotheses. 4. For the final presentation, each student is required to do a 15-minutes presentation on her/his research proposal for the identified topic. The content of presentation should include a report of background, literature review, research hypotheses, study design and statistical methods. 5. Each student is required to submit a final written report in the format of a research project proposal, including project title and sections on research background, literature review, materials and methods, and expected outcomes. Wei-Jane Chen 12 Wednesday 6,7,8 EPM8003 (849ED0400) 3 (College of Public Health) Graduate Institute of Epidemiology and Preventive Medicine
http://epm.ntu.edu.tw/?locale=en
The aim of this course is to provide a general introduction to path analysis, factor analysis, structural equation modeling and multilevel analysis. The examples and data are extensively drawn from literature in health and medical sciences. Students will learn how to use Mplus and Lisrel software to undertake these analyses. After attending the course, students should be able to describe the relationship between commonly used statistical methods and structural equation modeling (SEM); define the statistical concepts behind factor analysis, path analysis, and structural equation modeling; understand the relation between SEM and multilevel modeling (MLM); explain the above statistical methods and properly interpret their results; and use a computer software package to undertake the statistical analyses and correctly specify the statistical models. SEM has been very popular among quantitative social scientists in the last two decades, and has started to draw attentions from epidemiologists. SEM is a very useful tool for testing causal models, and learning SEM theory is very helpful for students to understand the causal assumptions behind different models. SEM is also useful for explaining the concepts of confounding, mediation and moderation in epidemiological research. The course will start with basic concepts of SEM, such as model specification, fitness testing, interpretation of causality and model modification. Then, more advanced topics will be introduced, such as equivalence models, identification issues, and multiple groups testing. MLM will then be introduced for the analysis of clustered data, where random effects may be viewed as latent variables. Students will be assessed by their participation in the classroom discussion, one interim and one final report on the critical appraisal of literature and real data analysis. By the end of this course, students should be able to: Describe the relations between general linear models and structural equation models Explain the statistical theory of principal component analysis, exploratory and confirmatory factor analysis, path analysis and structural equation models Understand the concepts and rationales of causal models within the framework of structural equation models Understand the concept of mediation and the decomposition of total effects into direct and indirect effects Undertake structural equation modeling using statistical software packages and interpret the results properly Report the results from structural equation modeling properly College of Public Health Downtown Campus-College of Public Health Active participation in class discussion and practical session is required. Tu, Yu Kang 30 Friday 3,4 EPM7001 (849EM0850) 2 (College of Public Health) Graduate Institute of Epidemiology and Preventive Medicine,
Common General Education Center Master Program In Statistics of National Taiwan University
http://epm.ntu.edu.tw/?locale=en
The aim of this course is to provide a general introduction to the research methods and application in global health. The examples and data are drawn from published literatures related to evidence-based medicine and health data research. The course will start with basic concepts of global health and evidence-based approach. Then, more advanced topics will be introduced, such as selection of a topic of interest, setting up the search strategy for literature review, and formation of a synthesis. Introduction to the management of public health data, as well as the assessment for quality of care using health claims data, will also be provided. Three special lectures will also be provided by experts in the relevant fields. Students will be guided to conduct projects related to their research, and present their results at the end of this semester. By the end of this course, students should be able to: 1. Understand the concepts and rationales of evidence-based medicine within the framework of global health. 2. Understand the process of forming a synthesis from literature review, quality assessment, statistical analysis, to manuscript writing. 3. Understand the statistical theory and the application of different statistical theories of meta-analysis. 4. Understand the management and assessment of quality of care for health data. 5. Report the results from their personal project properly. College of Public Health Downtown Campus-College of Public Health 1. Students should have the basic concept of epidemiology and biostatistics. 2. Students should have the basic concept of systematic review and meta-analysis. 3. Active participation in class discussion and practical session. Hon-Yen Wu 20 Tuesday 8,9 EPM7007 (849EM0910) 1 (College of Public Health) Graduate Institute of Epidemiology and Preventive Medicine http://epm.ntu.edu.tw/?locale=en
Measuring burden of disease: methods and applicationsThe measurement and quantification of population health could assist health policy making and priority setting. In the past few years there have been major advancements in burden of disease research, mainly led and stimulated by the Global Burden of Disease Study (GBD). This course will give an overview on the concepts and methods used to quantify the burden of disease at the national and global level. The GBD will be a main focus of this course, but other alternative approaches will also be reviewed. The course consists of lectures, computer labs, a hands-on group-based project, and a field visit to the Department of Statistics of Ministry of Health and Welfare. At the end of the course the students are expected to: 1. Understand the key concepts and elements in burden of disease studies 2. Comment on the strengths and limitations of burden of disease studies 3. Understand the estimating procedures of the GBD study 4. Be familiar with and be able to use the major databases of GBD while acknowledging their limitations College of Public Health Downtown Campus-College of Public Health The course will be offered in English. Basic understanding of key concepts of epidemiology will be helpful but is not absolutely required. Lin Hsien-Ho 30 Tuesday 6,7 EPM5018 (849EU0460) 2 (College of Public Health) Graduate Institute of Epidemiology and Preventive Medicine http://epm.ntu.edu.tw/?locale=en