以下為系統擷取之英文原文
2.160 Identification, Estimation, and Learning
Spring 2006
The Mars rover relies on sophisticated identification and estimation techniques to navigate the Martian terrain. (Image courtesy of NASA.)
Course Highlights
This course features extensive lecture notes and many assignments.
Course Description
This course provides a broad theoretical basis for system identification, estimation, and learning. Students will study least squares estimation and its convergence properties, Kalman filters, noise dynamics and system representation, function approximation theory, neural nets, radial basis functions, wavelets, Volterra expansions, informative data sets, persistent excitation, asymptotic variance, central limit theorems, model structure selection, system order estimate, maximum likelihood, unbiased estimates, Cramer-Rao lower bound, Kullback-Leibler information distance, Akaike's information criterion, experiment design, and model validation.
Technical Requirements
Special software is required to use some of the files in this course: .zip. The .txt files in the assignments section are used for MATLAB®.
Syllabus
Help support MIT OpenCourseWare by shopping at Amazon.com! MIT OpenCourseWare offers direct links to Amazon.com to purchase the books cited in this course. Click on the Amazon logo to the left of any citation and purchase the book from Amazon.com, and MIT OpenCourseWare will receive up to 10% of all purchases you make. Your support will enable MIT to continue offering open access to MIT courses. |
Course Description
This course covers the following topics, with the aim of providing students with a broad theoretical basis for system identification, estimation, and learning:
Least squares estimation and its convergence properties, Kalman filter and extended Kalman filter, noise dynamics and system representation, function approximation theory, neural nets, radial basis functions, wavelets, Volterra expansions, informative data sets, persistent excitation, asymptotic variance, central limit theorems, model structure selection, system order estimate, maximum likelihood, unbiased estimates, Cramer-Rao lower bound, Kullback-Leibler information distance, Akaike's information criterion, experiment design, and model validation.
Prerequisites
Advanced System Dynamics and Control (2.151)
Textbooks
There is no primary textbook for this course. Most of the course materials have been developed based on the following references:
Ljung, Lennart. System Identification: Theory for the User. 2nd ed. Upper Saddle River, NJ: Prentice-Hall, 1999. ISBN: 9780136566953.
Goodwin, Graham, and Kwai Sang Sin. Adaptive Filtering, Prediction, and Control. Englewood Cliffs, NJ: Prentice-Hall, 1984. ISBN: 9780130040695.
Burnham, Kenneth, and David Anderson. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach. 2nd ed. New York, NY: Springer, 2003. ISBN: 9780387953649.
Brown, Robert, and Patrick Hwang. Introduction to Random Signals and Applied Kalman Filtering. 3rd ed. New York, NY: Wiley, 1996. ISBN: 9780471128397.
Grading Policy
There will be 7 homework assignments and 2 two-hour exams. Each student is also expected to complete 1 project.
Activities | Percentages |
---|---|
First Exam | 30% |
Second Exam | 30% |
Homework Assignments | 20% |
Term Project | 20% |
Calendar
Lec # | Topics | Key Dates |
---|---|---|
1 | Introduction | |
Part I: Estimation | ||
2 | Recursive Least Square (RLS) Algorithms | |
3 | Properties of RLS | |
4 | Random Processes, Active Noise Cancellation | |
5 | Discrete Kalman Filter-1 | Problem set 1 due |
6 | Discrete Kalman Filter-2 | |
7 | Continuous Kalman Filter | Problem set 2 due |
8 | Extended Kalman Filter | |
Part 2: Representation and Learning | ||
9 | Prediction Modeling of Linear Systems | Problem set 3 due |
10 | Model Structure of Linear Time-invariant Systems | |
11 | Time Series Data Compression, Laguerre Series Expansion | Problem set 4 due |
12 | Non-linear Models, Function Approximation Theory, Radial Basis Functions | |
13 | Neural Networks | Problem set 5 due |
Mid-term Exam | ||
14 | Error Back Propagation Algorithm | |
Part 3: System Identification | ||
15 | Perspective of System Identification, Frequency Domain Analysis | |
16 | Informative Data Sets and Consistency | Problem set 6 due |
17 | Informative Experiments: Persistent Excitation | |
18 | Asymptotic Distribution of Parameter Estimates | |
19 | Experiment Design, Pseudo Random Binary Signals (PRBS) | |
20 | Maximum Likelihood Estimate, Cramer-Rao Lower Bound and Best Unbiased Estimate | Problem set 7 due |
21 | Information Theory of System Identification: Kullback-Leibler Information Distance, Akaike's Information Criterion | |
Final Exam |
Lecture Notes
Lecture Notes Table of Contents (PDF)
Available lecture notes are listed below.
Lec # | Topics | |
---|---|---|
1 | Introduction (PDF 1) (PDF 2) | |
Part I: Estimation | ||
2 | Recursive Least Square (RLS) Algorithms (PDF) | |
3 | Properties of RLS (PDF) | |
4 | Random Processes, Active Noise Cancellation (PDF) | |
5 | Discrete Kalman Filter-1 (PDF) | |
6 | Discrete Kalman Filter-2 (PDF) | |
7 | Continuous Kalman Filter (PDF) | |
8 | Extended Kalman Filter (PDF) | |
Part 2: Representation and Learning | ||
9 | Prediction Modeling of Linear Systems (PDF) | |
10 | Model Structure of Linear Time-invariant Systems (PDF) | |
11 | Time Series Data Compression, Laguerre Series Expansion (PDF) | |
12 | Non-linear Models, Function Approximation Theory, Radial Basis Functions (PDF) | |
13 | Neural Networks (PDF) | |
14 | Error Back Propagation Algorithm (PDF) | |
Part 3: System Identification | ||
15 | Perspective of System Identification, Frequency Domain Analysis (PDF) | |
16 | Informative Data Sets and Consistency (PDF) | |
17 | Informative Experiments: Persistent Excitation (PDF) | |
18 | Asymptotic Distribution of Parameter Estimates (PDF) | |
19 | Experiment Design, Pseudo Random Binary Signals (PRBS) (PDF) | |
20 | Maximum Likelihood Estimate, Cramer-Rao Lower Bound and Best Unbiased Estimate (PDF) | |
21 | Information Theory of System Identification: Kullback-Leibler Information Distance, Akaike's Information Criterion (PDF) |
Assignments
Special software is required to use some of the files in this section: .zip. The .txt files are used for MATLAB®. This section contains documents that could not be made accessible to screen reader software. A "#" symbol is used to denote such documents.
Each set of related files includes a "Read Me" document, which details how those files relate to the problem set.
Assignments | Related Files |
---|---|
Problem Set 1 (PDF) | ps1files.zip (ZIP) (The ZIP file contains: 13 .txt files.) |
Problem Set 2 (PDF) | ps2files.zip (ZIP) (The ZIP file contains: 6 .txt files.) |
Problem Set 3 (PDF) | ps3files.zip (ZIP) (The ZIP file contains: 6 .txt files.) |
Problem Set 4 (PDF) | |
Problem Set 5 (PDF) | ps5files.zip (ZIP) (The ZIP file contains: 3 .txt files.) |
Problem Set 6 (PDF)^{#} | |
Problem Set 7 (PDF) |
Exams
Mid-term Exam (PDF)
Final Exam (PDF)
Projects
Students are required to complete one project, which counts towards 20% of their final grade.
Topics
Find a project topic of your interest. Use identification, estimation, and learning techniques relevant to the course material. Possible topics include, but are not limited to:
A good term project will contain a mixture of theory and practice. Please include an interesting context or application background, address how you achieved the goal, and discuss specific technical details. Don't forget that the focus of this subject is data; let the data speak about the system.
Requirements
The project should be a maximum of ten pages long, excluding figures. Please include a title, name, and an abstract. The paper should contain the following sections:
留下您對本課程的評論 |
標籤 現有標籤：1 |
有關本課程的討論
目前暫無評論，快來留言吧！