Scribd Logo
Upload

Welcome to Scribd!

  • 79 views

    Proposal For A New Course

    Uploaded by

    rnkaushal
    mathematics for machine learning

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Proposal For A New Course

    Uploaded by

    rnkaushal
    79 views2 pages
    mathematics for machine learning

    Original Title

    cs698MathForML

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    mathematics for machine learning

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    79 views2 pages

    Proposal For A New Course

    Uploaded by

    rnkaushal
    mathematics for machine learning

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Proposal for a new course

    Title: Special topics in CS: Mathematics for Machine Learning Course No: CS698AB Units: 3-0-0-9 Proposer: Harish Karnick, Prateek Jain, Adjunct Faculty (Microsoft Research, Bangalore) Others interested in teaching course: Pre-requisites: CS201, CS203, (or CS602), CS210 (or equivalent). MSO201 (or equivalent knowledge) is desirable. About the course: The course will cover the important mathematical concepts, methods and results that are widely used and needed in machine learning. It has three major sub-parts. Matrix analysis and introductory functional analysis: Vector spaces, metrics, norms, inner products, linear transformations and properties. Banach spaces, inner product spaces, reproducing kernel Hilbert spaces, kernels, PSD matrices and properties, elementary properties of linear transformations. Basic spectral theory, eigenvalue decompositions, SVD and applications, Lanczos method. Matrix norms, spectral norms, elementwise and mixed norms, induced norms, matrix inversion, least squares and pseudo-inverse. Basic matrix analysis. Probability theory: Probability as a measure, sample space, algebra, pmf, pdf, elementary properties, conditional probabily, independence. Random variables, expectation, moments, law of large numbers, central limit theorem. Concentration inequalities - Markov, Tchebyshev, Bennett, Cherno, Azuma-H oding, McDiarmid etc. Johnson-Lindenstraus lemma. Matrix concentration bounds. Rademacher complexity, generalization bounds, uniform convergence. Design of experiments and basic hypothesis testing.

    Convex analysis and programming: Convex sets and functions, lower and upper semi-continuous sets, Jensens inequality. Lagrange multipliers, duality theory, Fenchels duality, dual norms, KKT conditions. LP, Farkas lemma, QP and S-lemma, cone programming, SDP. Gradient descent, Newtons method, conjugate gradient descent. References: 1. Lloyd N Trefethen, David Bau III, Numerical Linear Algebra, SIAM, 1997. 2. Gene H Golub, Charles F Van Loan, Matrix Computations, 3rd Ed., John Hopkins Univ. Press, 1996. 3. David A Harville, Matrix Algebra From a Statisticians Perspective, Springer, 1997. (Reference for results from Matrix Algebra). 4. Erwin Kreyszig, Introductory Functional Analysis with Applications, Wiley, 1978. 5. William Feller, An Introduction to Probability Theory and its Applications, 3rd Ed., Wiley, 1968. 6. David Stirzaker, Probability and Random Variables: A Beginners Guide, Cambridge Univ. Press, 2003. 7. Georey R Grimmet, David Stirzaker, Probability and Random Processes, 3Ed., Oxford Univ. Press, 2001. 8. Stephen Boyd, Lieven Vandenberghe, Convex Optimization, Cambridge Univ. Press, 2004. 9. A Ben-Tal, Arkadi Nemirovski, Lectures on Modern Convex Optimization: Analysis, Algorithms, Engineering Applications, SIAM, 2001. 10. Various resources on the internet - lecture notes, videos, papers etc.

    Proposers signature: Harish Karnick, Prateek Jain (Adjunct Faculty).

    Convenor, DPGC

    Chairman, SPGC

    You might also like