Machine Learning: Introduction Jens Kauffmann MaxPlanckInstitut fr - - PowerPoint PPT Presentation

▶
machine learning introduction
SMART_READER_LITE
LIVE PREVIEW

Machine Learning: Introduction Jens Kauffmann MaxPlanckInstitut fr - - PowerPoint PPT Presentation

Sep 26, 2023 308 likes •444 views

Machine Learning: Introduction Jens Kauffmann MaxPlanckInstitut fr Radioastronomie Example I: Classification of handwritten Digits Jens Kauffmann MPIfR 2 Example I: Classification of handwritten Digits training of classifier

slide-1
SLIDE 1

Machine Learning: Introduction

Jens Kauffmann Max–Planck–Institut für Radioastronomie

slide-2
SLIDE 2

Jens Kauffmann ● MPIfR

Example I: Classification of hand–written Digits

2

slide-3
SLIDE 3

Jens Kauffmann ● MPIfR

Example I: Classification of hand–written Digits

3

training of classifier

slide-4
SLIDE 4

Jens Kauffmann ● MPIfR

Example I: Classification of hand–written Digits

4

free parameter => verification needed!

slide-5
SLIDE 5

Jens Kauffmann ● MPIfR

Example II: Classification just uses complex Boundaries

5

slide-6
SLIDE 6

Jens Kauffmann ● MPIfR

Example II: Classification just uses complex Boundaries

6

slide-7
SLIDE 7

Jens Kauffmann ● MPIfR

Machine Learning: Needs Verification!

7

different methods => different results no verification, no parameter adjustment => nonsensical results

slide-8
SLIDE 8

Jens Kauffmann ● MPIfR

Survey of Methods I

8

slide-9
SLIDE 9

Jens Kauffmann ● MPIfR

Survey of Methods II

9

slide-10
SLIDE 10

Jens Kauffmann ● MPIfR

Survey of Methods III

10

(a) well–separated categories (b) overlapping categories

slide-11
SLIDE 11

Jens Kauffmann ● MPIfR

References & Literature

11

short texts to get started:

  • Scikit Tutorial „An Introduction to Machine Learning“
  • Scikit Tutorial „A Tutorial on Statistical Learning“

free books:

  • „The Elements of Statistical Learning"
  • „An Introduction to Statistical Learning"
  • „An Introduction to Machine Learning“ (Springer Link; free at MPIfR)
  • „Python Data Science Handbook“, section „Machine Learning“

Python packages:

  • Scikit learn
  • emcee
  • pymc
  • Seaborn (for graphics)

rather for people writing ML algorithms?

slide-12
SLIDE 12

Jens Kauffmann ● MPIfR

Exercises

12

schedule: Thursday, Feb. 16, 13:00 Tuesday, Feb. 21, 13:00 Wednesday, Feb. 22, 13:30 preparations: get Python 3.6 via Anaconda (https://www.continuum.io/downloads) learn how to start a Jupyter notebook familiarize yourself with Scikit learn (http://scikit-learn.org)