DATA SCIENCE FOR DYNAMICAL SYSTEMS - 2022/3
Module code: MATM062
In light of the Covid-19 pandemic the University has revised its courses to incorporate the ‘Hybrid Learning Experience’ in a departure from previous academic years and previously published information. The University has changed the delivery (and in some cases the content) of its programmes. Further information on the general principles of hybrid learning can be found at: Hybrid learning experience | University of Surrey.
We have updated key module information regarding the pattern of assessment and overall student workload to inform student module choices. We are currently working on bringing remaining published information up to date to reflect current practice in time for the start of the academic year 2021/22.
This means that some information within the programme and module catalogue will be subject to change. Current students are invited to contact their Programme Leader or Academic Hive with any questions relating to the information available.
The presentations of the module will focus on data-driven methods for the analysis of dynamical systems and time-series data and on related machine learning problems such as dimensionality reduction, manifold learning, regression, and classification.
Example topics include:
Learning governing equations from data; Transfer operators; Dynamic mode decomposition; Reproducing kernel Hilbert spaces; Support vector machines; Kernel ridge regression.
Python or Matlab will be used to implement data-driven methods. The methods will then be applied to typical benchmark problems such as chaotic dynamical systems, metastable stochastic systems, and fluid dynamics problems, but also, for instance, to image classification problems to highlight similarities with classical supervised learning applications.
KLUS Stefan (Maths)
Number of Credits: 15
ECTS Credits: 7.5
Framework: FHEQ Level 7
Module cap (Maximum number of students): 20
Overall student workload
Independent Learning Hours: 106
Seminar Hours: 11
Laboratory Hours: 9
Guided Learning: 11
Captured Content: 13
Prerequisites / Co-requisites
Indicative content includes:
General dimensionality reduction, classification, and clustering:
We will first introduce methods such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Canonical Correlation Analysis (CCA), k-means, and Spectral Clustering.
Kernel-based methods and statistical learning theory:
Reproducing kernel Hilbert space theory, Support Vector Machines, kernel-based variants of methods such as PCA and CCA, Kernel Ridge Regression, manifold learning.
Transfer operators and applications:
Koopman and Perron-Frobenius operators and their generators (for the detection of metastable sets and the global analysis of dynamical systems), Markov State Models, Ulam's method, Dynamic Mode Decomposition (and extensions), Sparse Identification of Nonlinear Dynamics (learning governing equations from data), kernel-based methods for transfer operators.
|Assessment type||Unit of assessment||Weighting|
|Examination Online||ONLINE EXAM||80|
The assessment strategy is designed to provide students with the opportunity to demonstrate:
Analytical ability by solution of unseen problems in the exam.
Subject knowledge through the recall of key definitions, theorems, and their proofs.
An understanding of practical considerations when completing the coursework.
The ability to analyse data, to select suitable methods, and to interpret the numerical results.
Thus, the summative assessment for this module consists of:
i) One two hour examination (students have the choice of three questions out of four to contribute to exam mark) at the end of the semester; weighted at 80% of the module mark.
ii) One coursework; weighted at 20% of the module mark.
Formative assessment and feedback
Students receive written feedback via a number of marked unassessed coursework assignments over an 11 week period. Formative guidance is given on the coursework.
- Give an overview of machine learning problems related to dynamical systems and equip students with a solid understanding of the mathematical tools required to analyse high-dimensional data sets.
- Provide students with hands-on experience in data science and time series analysis.
- Give students the opportunity to apply state-of-the-art machine learning tools to real-world problems.
|001||Demonstrate systematic understanding of key aspects of selected topics within dynamical systems theory, data science, and statistical learning theory.||KC|
|002||Demonstrate the capability to apply data-driven methods to simulation and measurement data and to interpret the results.||KCPT|
|003||Ability to understand and assess related machine learning methods and their applications and limitations.||KCPT|
|004||Capability to implement machine learning algorithms in Matlab/Python and to use established machine learning libraries.||KPT|
C - Cognitive/analytical
K - Subject knowledge
T - Transferable skills
P - Professional/Practical skills
Methods of Teaching / Learning
The learning and teaching strategy is designed to provide:
i) A comprehensive treatment of principles and methods required for model reduction, system identification, and regression or classification problems.
ii) Experience in problem solving and machine learning in general.
iii) Practical experience in data science and time-series analysis.
The learning and teaching methods include:
i) 3 x 1 hour lectures per week x 11 weeks. 1 x 1 hour computer lab session per week for 9 weeks. Students will gain experience in using Matlab/Python to analyse benchmark problems from different application areas such as molecular dynamics, fluid dynamics, or biology.
ii) Assessed coursework to give students practical experience in implementing techniques covered in lectures and lab sessions.
iii) Several unassessed exercise sheets to give students experience in using techniques introduced in the module and to receive feedback.
Indicated Lecture Hours (which may also include seminars, tutorials, workshops and other contact time) are approximate and may include in-class tests where one or more of these are an assessment on the module. In-class tests are scheduled/organised separately to taught content and will be published on to student personal timetables, where they apply to taken modules, as soon as they are finalised by central administration. This will usually be after the initial publication of the teaching timetable for the relevant semester.
Upon accessing the reading list, please search for the module using the module code: MATM062
Programmes this module appears in
|Mathematics MMath||1||Optional||A weighted aggregate mark of 40% is required to pass the module|
|Mathematics with Statistics MMath||1||Optional||A weighted aggregate mark of 40% is required to pass the module|
|Mathematics MSc||1||Optional||A weighted aggregate mark of 50% is required to pass the module|
|Mathematics and Physics MPhys||1||Optional||A weighted aggregate mark of 50% is required to pass the module|
|Mathematics and Physics MMath||1||Optional||A weighted aggregate mark of 50% is required to pass the module|
Please note that the information detailed within this record is accurate at the time of publishing and may be subject to change. This record contains information for the most up to date version of the programme / module for the 2022/3 academic year.