ACS-205-B Machine Learning Tools

Instructors: Dmitry Kropotov

Type: Online Lab

Org-unit: ACS

Course Name Abbreviation: ACS-205-B

Credits: 2.50

Min. | Max. participants: - | -

Partial Grades:

Further Grading Information:
Examination Type: Module Component Examinations
 
Module Component 1: Lecture & Project
 
Assessment Type: Written examination - Weight: 67%                                               

Duration: 120 min
                                                                                                      
Scope: All intended learning outcomes of the excluding the practical aspects.
 
Assessment Type: Project - Weight: 33%
                                                                                                      
Scope: All practical aspects of the intended learning outcomes.
 
Module Component 2: Lab

Assessment Type: Lab Assignments - Weight: 100%                                                   
 
Scope: All intended learning outcomes of the module. 
 
Completion: To pass this module, the examination of each module component has to be passed with at least 45%.

Official Course Description:
Machine learning (ML) concerns algorithms that are fed with (large quantities of) real-world data, and which return a compressed “model” of the data. An example is the “world model” of a robot; the input data are sensor data streams, from which the robot learns a model of its environment, which is needed, for instance, for navigation. Another example is a spoken language model; the input data are speech recordings, from which ML methods build a model of spoken English; this is useful, for instance, in automated speech recognition systems. There exist many formalisms in which such models can be cast, and an equally large diversity of learning algorithms. However, there is a relatively small number of fundamental challenges that are common to all of these formalisms and algorithms. The lectures introduce such fundamental concepts and illustrate them with a choice of elementary model formalisms (linear classifiers and regressors, radial basis function networks, clustering, online adaptive filters, neural networks, or hidden Markov models). Furthermore, the lectures also (re-)introduce required mathematical material from probability theory and linear algebra. The ML lecture is complemented in this module by an online tutorial where the application-oriented side of software development in the context of ML is considered.

Modern machine learning in industry and research requires the knowledge of a comprehensive stack of tools and systems that allow to store and administrate data (e.g. Amazon S3, Kaggle, Dataverse, GIT LFS), extract features for various applications (e.g. Word2Vec, TSFEL), build up machine learning pipelines of training, testing, and hyperparameter optimization (e.g. skit-learn, Keras, TensorFlow, PyToarch) and ultimately deploy finalized models (e.g. TensorFlow Serving, MLFlow). This module gives exposure to a regularly updated latest state of the art set of tools that are relevant for the practical use of Machine Learning. It thereby complements the more theoretical and methods-driven module “Machine Learning” with market-oriented skills. 

Additional Information:

Usability and Relationship to other Modules

• This module gives a thorough introduction to the basics of machine learning. It complements the Artificial Intelligence module.
  

Syllabus:
Intended Learning Outcomes

By the end of this module, students should be able to:

1. Understand the notion of probability spaces and random variables.
   
2. Understand basic linear modeling and estimation techniques.
   
3. Understand the fundamental nature of the “curse of dimensionality”.
   
4. Understand the fundamental nature of the bias-variance problem and standard coping strategies.
   
5. Use elementary classification learning methods (linear discrimination, radial basis function networks, multilayer perceptions).
   
6. Implement an end-to-end learning suite, including feature extraction and objective function optimization with regularization based on cross-validation.
   
7. Deploy ML tools in an application context.
   

1. T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd edition, Springer, 2008.
   
2. S. Shalev-Shwartz, Shai Ben-David: Understanding Machine Learning, Cambridge University Press, 2014.
   
3. C. Bishop, Pattern Recognition and Machine Learning, Springer, 2006.
   
4. T.M. Mitchell, Machine Learning, Mc Graw Hill India, 2017.