Difference between pages "CDS110 2016" and "GP SSM"

From Robotics
(Difference between pages)
Jump to: navigation, search
(Lecture Schedule)
 
m (Papers on Scattering Networks)
 
Line 1: Line 1:
This is the course homepage for CDS 101/110, Fall 2016.
+
This page gathers references and materials related to the study of
 +
* Gaussian Process (GP) State Space Models (SSM)
 +
* Deep Learning
 +
* Koopman Spectral Methods.
  
== Course Staff, Hours, Location ==
+
== Gaussian  Process Approaches ==
  
{| border=1 width=100%
+
=== Basic Gaussian Process Info ===
|-
+
* Rasmussen and Williams
| '''Position''' || '''Name''' || '''Office''' || '''Office Hours''' (changing weekly) || '''Email''' || '''Phone'''
+
|-
+
| '''Instructor'''
+
| Joel Burdick
+
| 245 Gates-Thomas
+
| ''send mail for an appointment''
+
| [mailto:jwb@robotics.caltech.edu jwb at robotics dot caltech dot edu]
+
| 626-395-4139
+
|-
+
| '''Teach Asst.'''
+
| Richard Cheng
+
| 205 Gates-Thomas
+
| TBD
+
| [mailto:georgiev@caltech.edu georgiev at caltech dot edu]
+
| 626-395-????
+
|-
+
| '''Teach Asst.'''
+
| Yoke Peng Leong
+
| Annenberg
+
| TBD
+
| [mailto:ypleong@caltech.edu ypleong at caltech dot edu]
+
| 626-395-????
+
|-
+
| '''Administrative'''
+
| Sonya Lincoln
+
| 250 Gates-Thomas
+
| 7:30am-noon; 1:00pm-4:30pm
+
| [mailto:lincolns@caltech.edu lincolns at caltech dot edu]
+
| 626-395-3385
+
|}
+
  
 +
=== Web Links ===
 +
* [http://dsc.ijs.si/jus.kocijan/GPdyn/ Bibliography on GP Models in Dynamical Systems]
  
== Announcements ==
+
=== Papers on GP-SSMs ===
 +
* J.M. Wang, D.J. Fleet, A. Hertzmann, [[Media:GPDynamicModels.pdf | Gaussian Process Dynamical Models]]
 +
* R. Turner, M.P. Deisenroth, C.E. Rasmussen, [[Media:StateSpaceInferenceLearningGPs.pdf | State-Space Inference and Learning with Gaussian Process]];
 +
* A. McHutchon, [[Media:NonlinearModellingControlUsingGPs.pdf | Nonlinear Modelling and Control Using Gaussian Processes]] (Ph.D. thesis, Cambridge University)
 +
* J. Ko, D. Fox, [[Media:GPBayesFilters.pdf | GP-BayesFilters: Bayesian filtering using Gaussian Process Prediction and Observation Models]]
 +
* F. Perez-Cruz, S.V. Vaerenbergh, J.J. Murrillo-Fuentes, M. Lazarro-Gredilla, and I. Santamaria, [[Media:GPforNonlinearSignalProcessing.pdf | Gaussian Processes for Nonlinear Signal Processing]];
 +
* A. Svensson, A. Solin, S. Sarkka, T.B. Schon, [[Media:EfficientBayesianLearmingGPSSMs.pdf | Computationall Efficient Bayesian Learning of Gaussian Process State Space Models]]
 +
* A.C. Damianou, M.K. Titsias, N.D. Lawrence, [[Media:VariationalGPDynSystems.pdf | Variational Gaussian Process Dynamical Systems]]
 +
* M.P. Deisenroth, D. Fox, C.E. Rasmussen, [[Media:GPsDataEfficientLearning.pdf | Gaussian Processes for Data-Efficient Learning in Robotics and Control]];
 +
* K. Jocikan, [[Media:DynamicGPModelsOverview.pdf | Dynamic GP Models: An Overview and Recent Developments]];
 +
* A. Solin, S. Sarkka, [[Media:ReducedRankGPR.pdf | Hilbert Space Methods for Reduced-Rank Gaussian Process Regression]]; (ArXiv.1401.5508)
 +
* C.L.C. Mattos, Z. Dai, A. Damianou, J. Forth, G.A. Barreto, N. Lawrence, [[Media:RecurrentGaussianProcesses.pdf | Recorruent Gaussian Processes]]
 +
* N.D. Lawrence, A.J. Moore, [[Media:HierarchicalGPLatentVariableModels.pdf | Hierarchical Gaussian Process Latent Variable Models]]
 +
* M.K. Titsias, N.D. Lawrence, [[Media:BayesianGPLatentVariableModel.pdf | Bayesian Gaussian Process Latent Variable Model]]
 +
* R. Calandra, J. Peters, C.E. Rasmussen, M.P. Deisenroth, [[Media:ManifoldGPR.pdf | Manifold Gaussian Processes for Regression]]
 +
* F. Berkenkamp and A.P. Schoellig, [[Media:SafeRobustLearningControlwithGPs.pdf | Safe and Robust Learning Control with Gaussian Processes]]
 +
* E.B. Fox, E.B. Sudderth, M.I. Jordan, A.S. Willsky, [[Media:SharingFeaturesDynamicalSystems.pdf | Sharing Features Among Dynamical Systems with Beta Processes]]
 +
* J.M. Wang, D.j. Fleet, A. Hertzmann, [[Media:GPDynamicalModelsHumanMotion.pdf | Gaussian Process Dynamical Models for Human Motion]]
 +
* E.D. Klenske, P. Hennig, [[Media:DualControlApproxBayesianRL.pdf | Dual Control for Approximate Bayesian Reinforcement Learning]]
 +
* Y. Pan and E.A. Theodorou, [[Media:DataDrivenDDPUsingGPs.pdf | Data-Driven Differential Dynamic Programming Using Gaussian Processes]]
 +
* F. Berkenkamp, R. Moriconi, A.P. Schoellig, A. Krause, [[Media:SafeLearningRegionsAttractionWithGPs.pdf | Safe Learning of Regions of Attraction for Uncertain, Nonlinear Systems with Gaussian Processes]]
 +
* M.P. Deisenroth, J. Peters, C.E. Rasmussen, [[Media:ApproximateDPwithGaussianProcesses.pdf | Approximate Dynamic Programming with Gaussian Processes]]
 +
* R. Frigola, F. Lindsten, T.B. Schon, C.E. Rasmussen, [[Media:IdentificationGPSSMwithParticalEM.pdf | Identification of Gaussian Process State-Space Models with Particle Stochastic Approximation EM]]
 +
<hr />
 +
* Z.Y. Wan and T.P. Sapsis, [[Media:ReducedSpaceGPR.pdf | Reduced-Space Gaussian Process Regression for Data-Driven Forecast of Chaotic Dynamical Systems]], ''arXiv:1611.01583''
 +
* T. Beckers, J. Umlauft, and S. Hirsche, [[Media:ModelBasedGPRControl.pdf | Stable Model-Based Control with Gaussian Process Regression for Robot Manipulators]],
 +
* A. Marco, P. Hennig, S. Schaal, S. Trimpe, [[Media:DesignLQRKernels.pdf | On the Design of LQR Kernels for Efficient Controller Learning]], ''arXiv:1709.07089v1''
 +
* N. Gorbach, S. Bauer, J. Buhmann, [[Media:ScalableVariationalInference.pdf | Scalable Variational Inference for Dynamical Systems]], NIPS 2017, Long Beach, CA, 2017.
 +
* J. Umlauft, T. Beckers, M. Kimmel, S. Hirsche, [[Media:FeedbackLinearlizatingUsingGPs.pdf | Feedback Linearization Using Gaussian Processes]]
 +
* F. Lindsten, M.I. Jordan, T.B. Schon, [[Media:ParticleGibbsWithAncestorSamping.pdf | Particles Gibbs with Ancestor Sampling]], ''J. Machine Learning Research'', vo. 15, pp. 2145-2184.
  
== Course Syllabus, Mechanics, and Grading ==
+
== Deep Learning ==
  
CDS 101/110 provides an introduction to feedback and control in physical,
+
=== Papers on Deep Learning ===
biological, engineering, and information sciences. The course will introduce students to the basic principles of
+
* This is the paper that we seek to understand in this reading group.
feedback and its use as a tool for altering the dynamics of systems, meeting systems specifications, and
+
** A. Achile and S. Soatto, [[Media:InvarianceDisentanglement.pdf | Emergence of Invariance and Disentanglement in Deep Representations]], ''arXiV:1706.01350v2'', Oct. 2017.
managing system uncertainty. Key themes include: linear system theory
+
* P. Chauderi, A.Oberman, S. Osher, S. Soatto, G. Ccarlier, [[Media:DeepNetPDEs.pdf | Deep Relaxation: Partial Differential Equations for Optimizing Deep Neural Networks]], ''arXiV:1704.04932v2''
input/output response, closed loop behavior, linear versus nonlinear
+
* Here are some background papers
models, and local versus global behavior.
+
** N. Tishby and N. Zaslavsky, [[Media:DeepLearningBottleneck.pdf | Deep Learning and the Information Bottleneck Principle]], ''arXiv:1503.02406'', March, 2015.
  
CDS 101 is a 6 unit (2-0-4) class intended for science
+
==== Review-like papers ====
and engineering students who are interested in the principles and tools of feedback
+
* S. Mallat, [[Media:UnderstandingDeepCNNs.pdf | Understanding Deep Convolutional Networks]], ''Phil. Trans. Royal Society, A, vol. 374, May 15, 2017.
control, but not necessarily the engineering and analytical techniques for design and synthesis of control
+
* M.D. Zeiler and R. Fergus, [https://arxiv.org/abs/1311.2901  Visualizing and Understanding Deep Convolutional Networks]
systems. CDS 110 is a 12 unit class (3-0-9) that provides a traditional first
+
course in control for engineers and applied scientists. It assumes a working knowledge of linear algebra and
+
ODEs as a prerequisite (e.g., as found in ACM 95). Familiarity with complex variables (Laplace transforms, residue theory)
+
is helpful but not required. The basics of these topics will be reviewed during the course.
+
  
=== Grading ===
+
==== Papers on Scattering Networks ====
The final grade will be based on homework sets, a midterm exam, and a final exam:  
+
''Scattering Networks'' are proposed by Stephan Mallat (of wavelet fame) to understand why deep nets work so well.
 +
* S. Mallat [[Media:GroupInvariantScattering.pdf | Group Invariant Scattering]], 2012.
 +
* J. Anden, S. Mallat, [[Media:DeepScatteringSpectrum.pdf | Deep Scattering Spectrum]], 2015.
 +
* J. Bruna, S. Mallat, [[Media:InvariantScatteringCNNs.pdf | Invariant Scattering Convolution Networks]]
  
*''Homework (50%):'' Homework sets will be handed out weekly and due on Wednesdays by 2 pm either in class or in the labeled box across from 107 Steele Lab. Each student is allowed up to two extensions of no more than 2 days each over the course of the term.  Homework turned in after Friday at 2 pm or after the two extensions are exhausted will not be accepted without a note from the health center or the Dean.  MATLAB/Python code and SIMULINK/Modelica diagrams are considered part of your solution and should be printed and turned in with the problem set (whether the problem asks for it or not).
+
==== Classics ====
 +
* K. Hornik, M. Stinchcombe, H. White, [[Media:HSH.pdf | Universal Approximation of an Unknown Mapping and Its Derivatives Using Multi-Layer Feedforward Networks]], ''Neural networks'', vol. 3, 1990.
  
* ''Midterm exam (20%):'' A midterm exam will be handed out at the beginning of midterms period (28 Oct) and due at the end of the midterm examination period (3 Nov). The midterm exam will be open book and computers will be allowed (though not required).  
+
=== Web Links ===
 +
** [http://deeplearning.net Deeplearning.net]
  
* ''Final exam (30%):''  The final exam will be handed out on the last day of class (4 Dec) and due at the end of finals week. It will be an open book exam and computers will be allowed (though not required).
+
== Koopman Spectral Method ==
  
=== Collaboration Policy ===
+
=== Papers on Koopman Spectral methods ===
 +
* S. Brunton, J. Proctor, N. Kutz, [[Media:DiscoveringEquationsFromData.pdf | Discovering Governing Equations from Data: Sparse Identification of Nonlinear Dyanmical Systems]], ''arXiv:1509.03580v1'' (also, [[Media:BPK_PNAS.pdf | PNAS Version of the paper]]).
 +
* S. L. Brunton, B.W.Brunton, J.L. Proctor, J.N.  Kutz, [[Media:KoopmanInvariantSubspaces.pdf | Koopman Invariant Subspaces and Finite Linear Representations of Nonlinear Dynamical Systems for Control]], ''PLOS One'', vol. 11, no. 2, 2016.
 +
* M. Budisic, R. Mohr, I. Mezic, [[Media:AppliedKoopmanism.pdf | Applied Koopmanism]], ''Chaos'', vol. 22, 2012.
 +
* M.O. Williams, C.W. Rowley, I. G. Kevrekidis, [[Media:KernalBasedMethod.pdf | A Kernel-Based Method for Data-Driven Koopman Spectral Analysis]], ''arXiv:1411.2260v4''
 +
* J.L. Proctor, S.L. Brunton, J.N. Kutz, [[Media:DMDwithControl.pdf | Dynamic Mode Decomposition with Control]], SIAM J. Applied Dynamical Systems, vol. 15, no. 1, pp. 142-161, 2016.
 +
* I. Mezic, [[Media:ApplicationsSpectralTheoryKoopman.pdf | On the Applications of the Theory of the Koopman Operator in Dynamical Systems and Control Theory]], ''Proc. IEEE Conf. Decision Control'', 2015
 +
* M.O. Williams, C. Rowley, I.G. Kevrekidis, [[Media:DataDrivenApproximation.pdf | A Data-Driven Approximation of the Koopman Operator: Extending Dynamic Mode Decomposition]], ''J. Nonlinear Science'', vol. 25, 2015.
 +
* D. Giannakis, [[Media:DataDrivenSpectralDecomposition.pdf | Data-Driven Spectral Decomposition and Forecasting of ergodic dynamical systems]], ''arXiv:1507.02338v2''
 +
* I. Mezic and A. Surana, [[Media:PeriodicKoopmanModeDecomposition.pdf | Koopman Mode Decomposition for Periodic/Quasi-Periodic Time Dependence]], ''IFAC Papers Online, 48-18, pp. 690697, 2016.
 +
* H. Schaeffer, [[Media:LearningPDEs.pdf | Learning Partial Differential Equations via Data Discovery and Sparse Optimization]], ''J. Royal Society, Proceedings A'', 2017
 +
* J.N. Kutz, X. Fu, S.L. Brunton, [[Media:MultiresolutionDMD.pdf | Multi-resolution Dynamic Mode Decomposition]], ''arXiv:1506.00564''
  
Collaboration on homework assignments is encouraged. You may consult
+
==== Papers which are more oriented toward control ====
outside reference materials, other students, the TA, or the
+
* D. Goswami and D.A. Paley, [[Media:GlobalBilinearization.pdf | Global Bilinearization and Controllability of Control-Affine Nonlinear Systems: A Koopman Spectral Approach]],
instructor, but you cannot consult homework solutions from
+
* A. Surana, [[Media:KoopmanObserverBasedSynthesis.pdf | Koopman Based Operator Synthesis for Control-Affine Noninear Systems]], ''Proc. IEEE Conf. Decision and Control'', 2016
prior years and you must cite any use of material from outside
+
* A. Surana and A. Banaszuk, [[Media:LinearObserverSynthesis.pdf | Linear Observer Synthesis for Nonlinear Systems Using Koopman Operator Framework]], ''IFAC Papers Online'', 49-18, pp. 716-723, 2016.
references. All solutions that are handed in should be written up
+
individually and should reflect your own understanding of the subject
+
matter at the time of writing. MATLAB/Python scripts and plots are
+
considered part of your writeup and should be done individually (you
+
can share ideas, but not code).
+
  
No collaboration is allowed on the midterm or final exams.
+
==== Papers which are more oriented toward fluids ====
 +
* I. Mezic, [[Media:AnalysisFluidFlows.pdf | Analysis of Fluid Flows via Spectral Properties of the Koopman Operator]], ''Annual Review of Fluids,'' vol. 45, 357-378, 2013.
 +
* C.W. Rowley and S.T.M. Dawson, [[Media:FlowModelReductionReview.pdf | Model Reduction for Flow Analysis and Control]], ''Annual Review Fluids'', 49:387-417, 2017.
 +
* M.S Hemati, C.W. Rowley, E.A. Deem, L.N. Cattafesta, [[Media:DebiasingDMD.pdf | De-biasing the Dynamic Mode Decomposition for applied Koopman spectral analysis of noisy data sets]], ''arXiv:1502.03854v2''
 +
* J.H. Tu, [[Media:DMDApplicationsTheory.pdf | Dynamic Mode Decomposition, Theory and Applications]], Ph.D. Thesis, Princeton, 2013.
 +
* S Bagheri, [[Media:ShearFlowsThesis.pdf | Analysis and Control of Transitional Shear Flows Using Global Modes]], Ph.D. Thesis, Royal Inst. Technology, Sweden, 2010.
  
=== Course Text and References ===
+
==== Some Early Papers ====
 +
* Koopman's original paper: [[Media:KoopmanPNAS.pdf | Dynamical Systems of Continuous Spectra]], ''PNAS'' , vol. 18, 1932.
 +
* J. Ding, [[Media:PointSpectrum.pdf | The Point Spectrum of Frobenius-Perron and Koopman Operators]], ''Proc. AMS,'' vol. 126, no. 5, 1998.
 +
* I. Mezic and A. Banaszuk, [[Media:ComparisonSystems.pdf | Comparison of Systems with Complex Behavior]], ''Physica D'', vol. 197, pp. 101-133, 2004.
 +
* Y. Lan and I. Mezic, [[Media:LinearizationInTheLarge.pdf | Linearization in the Large of Nonlinear Systems and Koopman Operator Spectrum]], ''Physica D'', 2013
  
The primary course text is
+
==== Other Papers ====
* K. J. Astrom and Richard M. Murray, [http://fbsbook.org ''Feedback Systems: An Introduction for Scientists and Engineers''], Princeton University Press, 2008
+
* E.Berger, M. Sastuba, D. Vogt, B. Jung, H.B. Amor, [[Media:EstimationRobotPerturbations.pdf | Estimation of Perturbations in Robotic Behavior using Dynamic Mode Decomposition]], ''Advanced Robotics,'' vol. 25, no. 5, 2015.
This book is available via the Caltech online bookstore or via download from the [http://fbsbook.org companion web site].  Note that the second edition of this book is in preparation for publication and will serve as the primary text for the course (but almost all of the material we will cover is also in the first edition).
+
* S. Wang, Z. Qiao, [[Media:NuclearNormDMD.pdf | Nuclear Norm Regularized Dynamic Mode Decomposition]], ''IET Signal Processing'', 2016.
 
+
The following additional references may also be useful:
+
 
+
* A. D. Lewis, ''A Mathematical Approach to Classical Control'', 2003. [http://www.mast.queensu.ca/~andrew/teaching/math332/notes.shtml Online access].
+
* J. Distefano III, A. R. Stubberud and Ivan J. Williams (Author), ''Schaum's Outline of Feedback and Control Systems'', 2nd Edition, 2013. 
+
 
+
In addition to the books above, the textbooks below may also be useful.  They are available in the library (non-reserve), from other students, or you can order them online.
+
 
+
* B. Friedland, ''Control System Design: An Introduction to State-Space Methods'', McGraw-Hill, 1986.
+
* G. F. Franklin, J. D. Powell, and A. Emami-Naeni, ''Feedback Control of Dynamic Systems'', Addison-Wesley, 2002.
+
 
+
== Lecture Schedule ==
+
 
+
The following is a '''tentative''' schedule for the class, based on previous years' experience.
+
 
+
{| class="mw-collapsible wikitable" width=100% border=1 cellpadding=5
+
|-
+
| '''Date'''
+
| '''Topic'''
+
| '''Reading'''
+
| '''Homework'''
+
|- valign=top
+
|- valign=top
+
| '''Week 1'''<br>
+
26 Sept <br>  28 Sept <br> 30 Sept.
+
| Introduction and Review
+
* Introduction to feedback and control
+
* Review of differential equation and linear algebra
+
* Feedback principles and examples
+
| FBS-1e 1.1-1.2, 1.4-1.5 <br> FBS-2e 1.1-1.5 (skim), ''2.1-2.4''
+
* {{cds110 fa15 pdf |bgsurvey.pdf | Background survey}}
+
* {{cds110 fa15 pdf |L1-1_intro-28Sep15_h.pdf | Mon lecture notes}},
+
* [[Media:ReviewLAODEs.pdf | Wed review session notes]] (PDF)
+
| {{cds110 fa15 pdf |hw1-fa15.pdf | HW 1}} <br> Due: 7 Oct, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw1-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 2'''<br>
+
3 Oct <br> 5 Oct <br> 7 Oct*
+
| Modeling, Stability
+
* State space models
+
* Phase portraits and stability
+
* Introduction to MATLAB
+
| FBS-1e 2.1-2.2, 3.1 4.1-4.3 <br> FBS-2e 3.1-3.2, 4.1, 5.1-5.3
+
* {{cds110 fa15 pdf |L2-1_modeling-05Oct15_h.pdf | Mon lecture notes}}
+
* {{cds110 fa15 pdf |L2-2_stability-07Oct15_h.pdf | Wed lecture notes}}
+
| {{cds110 fa15 pdf |hw2-fa15.pdf | HW 2}} <br> Due: 14 Oct, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw2-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 3'''<br>
+
10 Oct* <br> 12 Oct* <br> 14 Oct*
+
| Linear Systems
+
* Input/output response of LTI systems
+
* Matrix exponential, convolution equation
+
* Linearization around an equilibrium point
+
| FBS-1e  5.1-5.4 <br> FBS-2e 6.1-6.4
+
* {{cds110 fa15 pdf |L3-1_linsys-12Oct15.pdf | Mon lecture slides}}
+
* {{cds110 fa15 pdf |L3-3_recitation.pdf | Fri recitation slides}}
+
| {{cds110 fa15 pdf |hw3-fa15.pdf|HW 3}} <br> Due: 21 Oct, 2 pm
+
* Python: [http://www.cds.caltech.edu/~macmardg/courses/cds101/fa12/python/cartpend.py cartpend.py]
+
* MATLAB: [http://www.cds.caltech.edu/~macmardg/courses/cds101/fa10/matlab/cartpend.m cartpend.m], [http://www.cds.caltech.edu/~macmardg/courses/cds101/fa10/matlab/cartpend_model.m cartpend_model.m]
+
* SIMULINK: [http://www.cds.caltech.edu/~macmardg/courses/cds101/fa09/matlab/balance_simple.mdl balance_simple.mdl]
+
 
+
{{cds110 fa15 pdf |caltech/hw3-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 4'''<br>
+
17 Oct <br> 19 Oct <br> 21 Oct*
+
| State Feedback
+
* Reachability
+
* State feedback and eigenvalue placement
+
| FBS-1e 6.1-6.4 <br> FBS-2e 7.1-7.4
+
* {{cds110 fa15 pdf |L4-1_statefbk-19Oct15_h.pdf | Mon lecture slides}}
+
* MATLAB:  {{cds110 fa15 matlab|L4_1_statefbk.m}}, {{cds110 fa15 matlab |predprey.m}}, {{cds110 fa15 matlab |predprey_rh.m}}
+
* Python: {{cds110 fa15 python|L4_1_statefbk.py}}, {{cds110 fa15 python|predprey.py}}
+
|  {{cds110 fa15 pdf |hw4-fa15.pdf|HW 4}} <br> Due: 28 Oct, 2 pm
+
 
+
[http://www.cds.caltech.edu/~murray/amwiki/index.php/Bicycle_dynamics Bicycle dynamics]
+
* MATLAB: {{cds110 fa15 matlab|bike_linmod.m}}
+
* Python: {{cds110 fa15 python|bike_linmod.py}}
+
 
+
{{cds110 fa15 pdf |caltech/hw4-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 5'''<br>
+
24 Oct <br> 26 Oct <br> 28 Oct
+
| State space control design
+
* Trajectory generation, feedforward
+
* Integral feedback
+
* State estimation (if time)
+
* Midterm review
+
| FBS-1e 7.1-7.3 <br> FBS-2e 8.1-8.3
+
| Midterm exam <br> Due: 3 Nov, 5 pm
+
 
+
{{cds110 fa15 pdf |caltech/midterm-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 6'''<br>
+
1 Oct <br> 2 Nov <br> 4 Nov*
+
| Transfer Functions
+
* Frequency domain modeling
+
* Block diagram algebra
+
* Bode plots
+
| FBS-1e 8.1-8.4 <br> FBS-2e 9.1-9.4
+
* {{cds110 fa15 pdf |L6-1_xferfcns-02Nov15_h.pdf | Mon lecture slides}}
+
* [[Media:Recitation_nov_6.pdf | Fri review session notes]] (PDF)
+
| {{cds110 fa15 pdf |hw5-fa15.pdf | HW 5}} <br> Due: 11 Nov, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw5-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 7'''<br>
+
7 Nov <br> 9 Nov <br> 11 Nov*
+
| Loop Analysis
+
* Loop transfer function and the Nyquist criterion
+
* Stability margins
+
| FBS-1e 9.1-9.3 <br> FBS-2e 10.1-10.3
+
* {{cds110 fa15 pdf |L7-1_loopanal-09Nov15_h.pdf | Mon lecture slides}}
+
* {{cds110 fa15 pdf |L7-3_delay+nyquist.pdf | Fri recitation notes}}
+
| {{cds110 fa15 pdf |hw6-fa15.pdf | HW 6}} <br> Due: 18 Nov, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw6-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 8'''<br>
+
14 Nov <br> 16 Nov* <br> 18 Nov
+
| PID Control
+
* Simple controllers for complex systems
+
* Integral action and anti-windup
+
| FBS-1e 10.1-10.4 <br> FBS-2e 11.1-11.4
+
* {{cds110 fa15 pdf |L8-1_pid-16Nov15_h.pdf | Mon lecture slides}}
+
* [[Media:Recitation_110_nov_17.pdf | Wed lecture slides]] (PDF)
+
* [http://www.cds.caltech.edu/~murray/courses/cds110/fa15/minsegpid.py Fri PID example] (python)
+
| {{cds110 fa15 pdf |hw7-fa15.pdf | HW 7}} <br> Due: 25 Nov, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw7-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 9'''<br>
+
21 Nov <br> 23 Nov <br> ''Thanksgiving Holiday''
+
| Loop Shaping, I
+
* Sensitivity functions
+
* Feedback design via loop shaping
+
| FBS-1e 11.1-11.3 <br> FBS-2e 12.1-12.4
+
* {{cds110 fa15 pdf |L9-1_loopsyn-23Nov15_h.pdf | Mon lecture slides}}
+
* {{cds110 fa15 pdf |L9-2_recitation-25Nov15.pdf | Wed recitation slides}}
+
| {{cds110 fa15 pdf |hw8-fa15.pdf | HW 8}} <br> Due: 4 Dec, 2 pm
+
 
+
{{cds110 fa15 pdf |caltech/hw8-fa15_solns.pdf | Solutions}} (Caltech access only)
+
|- valign=top
+
| '''Week 10'''<br>
+
28 Nov <br> 30 Nov. <br> 2 Dec
+
| Loop Shaping II
+
* Fundamental limitations
+
* Modeling uncertainty
+
* Performance/robustness tradeoffs
+
| FBS-1e 11.4, 12.1-12.4 <br> FBS-2e 12.6-12.7, 13.1-13.3
+
* {{cds110 fa15 pdf |L10-1_limits-30Nov15_h.pdf | Mon lecture slides}}
+
* {{cds110 fa15 pdf |L10-2_pvtol-02Dec15_h.pdf | Wed lecture slides}}
+
* [http://www.cds.caltech.edu/~murray/courses/cds110/fa15/pvtol-nested.py Wed PVTOL example] (python)
+
| Final exam <br> Due 5 pm on last day of Final Exam Period
+
* To be posted on-line
+
|}
+

Revision as of 17:41, 16 January 2018

This page gathers references and materials related to the study of

  • Gaussian Process (GP) State Space Models (SSM)
  • Deep Learning
  • Koopman Spectral Methods.

Gaussian Process Approaches

Basic Gaussian Process Info

  • Rasmussen and Williams

Web Links

Papers on GP-SSMs


Deep Learning

Papers on Deep Learning

Review-like papers

Papers on Scattering Networks

Scattering Networks are proposed by Stephan Mallat (of wavelet fame) to understand why deep nets work so well.

Classics

Web Links

Koopman Spectral Method

Papers on Koopman Spectral methods

Papers which are more oriented toward control

Papers which are more oriented toward fluids

Some Early Papers

Other Papers