Velkommen til OOP2

Welcome

Updated

ML (Machine Learning
4. Semester: Datamatician Course (Computer Science)

Welcome to ML

Links to literature requirements

ML is short for "Machine Learning"

Be aware that ML requires group discussion as well as reading by the individual student.

The major topics at 4. semester are:

Applications of Machine Learning
Real life applications and data mining
Machine Learning Principles
Supervised/unsupervised, model/instance based,

Simple Training Models
Linear and logistic regression
Classification
MNIST, Binary classifier, confusion-matrix

Decision trees
Probabilities, hyperparameters
Support Vector Machines (SVM)
Linear, non-linear, classification, regression
Unsupervised Learning
Clustering: K-Means, Gaussian mixtures
(Dimensionality reduction)
Probabilities, hyperparameters

Artificial Neural Nerworks (ANN)
Perceptron, Multilayers, Activation functions, Finetuning

Machine Learning frameworks
Anaconda, Jupyter, Spyder, TensorFlow, Keras, (Azure, Google)

Libraries
sklearn, pandas, numpy, matplotlib

Practical exercises:

Working on many smaller exercises and homeworks supporting the understanding of the theory.

Great emphasize is given to your understanding of the basic principles and to your technical skills, too. Therefore you have to solve some discussion assignments and many practical exercises. Some can be solved at the school using your teachers and class mates carefull guidance/help. Others you have to work with, and to solve at home or in your study group.

ML is really like swimming:
You can only learn by doing - But you will have to read about it, too !

The subjects have been chosen in accordance with the official datamatician curriculum as given by the Danish Ministry of Education and the syllabus at ZIBAT, Roskilde Computer Science..

Teaching Principles
The teaching will be based on classic lectures, discussions, smaller practical assignments, groupwork and individual work and team building/work on the on the mandatory assignments.
Class (10 weeks): Lectures 25%, Hands on Individual Assignments 30%, Group work 35%, Mandatory Projects 25%
Individual Synopsis (5 weeks research)

Relations with other subjects
ML is related to:

Object Oriented Programming, 3. semester.
C# programming
Databases, 4. semester.
Data mining

Mandatory assignments
There are 2 minor mandatory assignmets in ML solved on the fly.
This will involve groupwork of 4-5 students as well as individual solution.
The mandatory assignments must be passed.

Exam
There is one exam held after 4. semester.

An oral individual exam in the individual synopsis (5 weeks research)

To proceed to the 5th Semester, all mandatory assignments must be passed and a minimum of 02 must be achieved at the exam.

Program and developing enviroment
There are various developing enviroments for building and testing models in ML.
We shall use Anaconda for several reasons:

It includes the newest facilities in Pythons3, Jupyter and Spyder
Relative simple to handle
All necessary libraries are included in the download
It is free for students at our school.

Maintenance by micl@esj.dk