The program at a glance
The basic principle of machine learning (ML) is the teaching of computers to learn from data, enabling them to execute useful tasks. There are many types of data, including images, text, tables and audio recordings. The tasks to be executed are as manifold and range from the classification of documents to risk assessment, to process optimization, to computer vision and to recommender systems. In this course, you will learn
- which ML models are the right ones for a host of tasks,
- how you can prepare your data to train these models,
- to train your ML system, to evaluate its performance and to scale the system to operate it productively.
After a series of remarkable breakthroughs in the past decade, we now have machine learning models that perform at an astonishing level. ML with many adjustable parameters (such as deep learning models) allow for the development of human-like chatbots and self-driving cars. The program teaches the neural network models on which these systems are based as well as other ML approaches such as decision trees and Bayesian learning.
The CAS in Machine Learning program is for everybody wishing to develop a deeper understanding of machine learning and to improve their skills in using it. Programming experience is welcome, but not required. To support participants who did not do much programming in recent years, the first module is fully dedicated to brushing up coding skills with Python. If you have any question concerning the admission requirements, please contact the head of program.
The course objective is to enable you to use ML systems directly in your day-to-day work environment. It will give you a valuable and sought-after professional qualification.
Module overview
Module 1 - Introduction
- Math refresher: Linear algebra, calculus, statistics
- Coding refresher: Python, Numpy, Matplotlib
- The history and development of machine learning
- Data management and feature engineering
Module 2 - Machine Learning
- Unsupervised learning
- Supervised learning
- Artificial neural networks
- Model validation
- Model diagnostics
Module 3 – Deep Learning
- Convolutional neural networks
- Computer vision
- Generative models
- Artificial neural networks
- Natural language processing (NLP)
- Transformers: Attention is all you need
Modul 4 – Other types of Machine Learning
- Recommender systems
- Decision trees, random forest and gradient boosting
- Bayesian learning and Bayesian networks
- Self-supervised learning
- Reinforcement learning
Module 5 – Production deployment and MLops
- The workflow of machine learning
- Model deployment in production
- MLops concepts and strategies
- Architectures of deployment: Edge, Cloud, Browser
- Monitoring of production models
The focus is on practical work: each topic will be studied through Python programming and exercise units in groups. Guest lecturers will come in to discuss advanced topics. The requirement to obtain the CAS certificate is course attendance for the duration of the program.
Transfer project
In the transfer project, participants work with a real-life dataset. The process comprises the following stages:
- The participants either choose an existing project/problem from their professional practice or define one based on their personal interests.
- They discuss the idea with the lecturers, who directly approve it or suggest changes.
- During the program, the participants are given the time to work their project supervised by their lecturers.
- The participants present their projects on the final day of the CAS program.
The goal is for the participants to gain experience with a real-life project, creating added value for their CV and their companies alike.
Technologies used:
- Python, Jupyter, Scikit Learn, Pandas, Numpy, Matplotlib
- Tensorflow, Keras
- GPUS and hardware acceleration
- AWS, MS-Azure, Google Cloud