KTH Mathematics  


Modern Methods of Statistical Learning Theory SF2935

The aim of the course is to introduce some of the basic algorithms and methods of statistical learning theory at an intermediate level. These are essential tools for making sense of the vast and complex data sets (c.f. big data) that have emerged in fields ranging from biology to marketing to astrophysics in the past decades. The course presents some of the most important modeling and prediction techniques, along with some relevant applications. Topics presented include classification, artificial neural networks with exponential families of distributions, Bayesian learning, resampling methods, tree-based methods, and clustering, highdimensional data. This is a good part of the background required for a career in data analytics. The course is lectured and examined in English.

Recommended prerequisities:

  • SF 1901 or equivalent course of the type 'a first course in probability and statistics (for engineers)'
  • Multivariate normal distribution
  • Basic differential and integral calculus, basic linear algebra.
  • Proficiency in R (optional)

Lecturers:

Course literature::

  • G. James, D. Witten, T. Hastie, R. Tibshirani: An introduction to Statistical Learning web page for the book (acronym below: ISL) by the publisher Springer
  • some sections of: Avrim Blum, John Hopcroft and Ravindran Kannan: Foundations of Data Science pdf from the authors
  • Supplementary reading and material from the lectures web page

The textbook ISL can be bought at THS Kårbokhandel, Drottning Kristinas väg 15-19.


Examination:

  • Computer homework (3.0 cu): there are two compulsory computer projects/home work that are to be submitted as written reports. Each report should be produced by a group of two (2) students. The reports are examined at the Project presentation seminars on TBA of November and TBA of December, 2017. The computer homework will be graded with Pass/Fail.
  • There will be a written exam (4.5 cu), consisting of five (5) assignments, on Thursday 11th of January, 2018, 08- 13.00 hrs.

  • Bonus for summaries of the guest lectures and papers An individually written summary (max. 2xA4) of the scientific contents of a guest lecture (2 x E.A), (LK) (SV) will provide one (1) bonus point for the exam. In addition can bonus points be gained by written summaries of at most two scientific articles (TBA). The summary is expected to be based on the students' own notes taken during the lecture or reading of a paper. The summaries must be submitted with deadline Fri 16th of December at 15 hrs. The bonus points are valid for the ordinary Exam on Thursday 11th of January, 2018, and in the re-examination on (TBA). The maximum number of bonus points to be gained is five (5).


  • Important: Students, who are admitted to a course and who intend to attend it, need to activate themselves in Rapp . Log in there using your KTH-id and click on "activate" (aktivera). The codename for sf2935 in Rapp is statin17.
    Registration for the written examination via "mina sidor"/"my pages" is required.
    Grades are set according to the quality of the written examination. Grades are given in the range A-F, where A is the best and F means failed. Fx means that you have the right to a complementary examination (to reach the grade E). The criteria for Fx is a grade F on the exam, and that an isolated part of the course can be identified where you have shown a particular lack of knowledge and that the examination after a complementary examination on this part can be given the grade E.

  • Supervision for computer projects
    Teaching assistant Daniel Berglund will be available for advice and supervision for computer projects at times to be announced.

    Plan of lectures
    KTH Social .
    (TK=Timo Koski, JO= Jimmy Olsson TP=Tetyana Pavlenko, DB= Daniel Berglund, EA= Erik Aurell, LK= Lukas Käll, SV= Sara Väljamets, ISL = the textbook, FoDSc= Foundations of Data Science ) 

    The addresses of the lecture halls and guiding instructions are found by clicking on the Hall links below


    Day Date Time Hall Topic Lecturer
    Tue 31/10 13-15 Q2 Lecture 1: Introduction to statistical learning (perceptrons, feedforward neural nets) and the course work. Introduction to computer projects Chapter 2 in ISL.
    TK
    Thu 02/11
    08-10 Q2 Lecture 2:
    Supervised Learning Part I.
    Chapter 4 in ISL

    TP
    Fri
    03/11
    10-12 Q2 Lecture 3: Supervised Learning Part II.
    Chapter 4 in ISL
    TP
    Tue
    07/11 14-16 Q2 Lecture 4: Bootstrap
    TP
    Thu
    09/11 08-10 Q2 Lecture 5: Introduction to R in a computer class Chapter 2 in ISL DB
    Fri
    10/11 10-12 Q2 Lecture 6: feedforward neural networks as statistical models I, handouts.

    TK
    Tue
    14/11 13-15 Q2 Lecture 7: feedforward neural networks as statistical models II, Support vector machines (SVM) I Chapter 9 in IS TK
    Thu
    16/11 08-10 Q2 Lecture 8: SVM II Chapter 9 in ISL
    TK
    Fri
    17/11 10-12 Q2 Lecture 9: Bayesian Learning I, Handouts TK
    Tue
    21/11 13-15 D3
    Lecture 10:Project presentation seminar 1
    TK
    Thu
    23/11 08-10 Q2 Lecture 11:Bayesian Learning II Handouts
    TK
    Fri
    24/11 10-12 E3 Lecture 12: Guest Lecture: TBA SV
    Tue
    28/11 13-15 E3 Lecture 13: Unsupervised learning part I. Chapter 10 in ISL
    TK
    Thu
    30/11 08-10 Q2 Lecture 14: Unsupervised learning part II. Chapter 10 in ISL
    TK
    Fri
    01/12 10-12 E3
    Lecture 15: GUEST LECTURE: An insight into computational and statistical mass spectrometry-based proteomics LK
    Tue
    05/12 13-15 E3 Lecture 16: Random Trees and Classification. Chapter 8 in ISL JO
    Fri
    06/12 15-17 Q2
    Lecture 17: Guest Lecture: Inferring protein structures from many protein sequences I
    EA
    Thu
    07/12 08-10 Q2 Lecture 18: Geometry of High-Dimensional Spaces, Gaussians in high Dimensions, Johnson -Lindenstrauss Lemma, Separating Gaussians.Part I, Chap.2 in FoDSc. TK
    Tue
    12/12 13-15 E3
    Lecture 19: Guest Lecture: Inferring protein structures from many protein sequences II
    EA
    Fri
    14/12 08-10 Q2
    Lecture 20: Geometry of High-Dimensional Spaces, Gaussians in high Dimensions, Johnson -Lindenstrauss Lemma, Separating Gaussians, Part II. Chap.2 in FoDSc. TK
    Fri
    15/12 10-12 E51
    Lecture 21:Project presentation seminar 2 TK, TP
    Thu
    11/01/2018 TBA Q24, Q26, Q22 Exam TK
    Xy
    xx/xx/2018 TBA TBA Re-exam TK

    Welcome, we hope you will enjoy the course (and learn (sic) a lot)!

    Tetyana, Jimmy & Timo


    To course web page




    Published by: Timo Koski
    Updated:20176-10-12