In this course, the students are introduced to the basic mathematical concepts required to understand deep learning. They start from the general ideas of applied mathematics that allow to define multivariable functions, find the maximum and minimum points of these functions as well as to quantify the degrees of certainty. Next, the fundamental goals of machine learning are analyzed. The students examine how to achieve these goals by defining a model which has specific targets, designing a cost function which measures how well those targets correspond to reality whilst using a training algorithm in order to minimize this cost function. This fundamental framework is the basis for a wide range of machine learning algorithms, including non-deep learning approaches. However, the main aim is to develop deep learning algorithms within the defined framework.

• Linear Algebra
• Probability and information theory
• Numerical Calculations
• Introduction to pattern recognition and machine learning 
• Statistical data analysis

This course is using project-based assessment. During the semester small (individual) projects are given to students on the following (indicative) topics, and final grade is based on the total sum of all small project grades (100% / number of small projects, with a number of 4-5 small projects given during the course):

• Algorithms
• Image recognition using matrices
• Quaternions or Hamiltonian Quaternions (use case: animation)

• Utilize fundamental mathematical principles to address intricate challenges in deep learning, encompassing tasks such as handling multivariable functions and identifying extremum points.
• Formulate machine learning models with precise targets and objectives, showcasing proficiency in designing models tailored to tackle real-world challenges..
• Design cost functions that adeptly gauge the alignment between machine learning models and real-world data, demonstrating skill in quantifying model performance.
• Execute training algorithms for the minimization of cost functions and the optimization of machine learning models, illustrating practical expertise in algorithm implementation.
• Advance the development of deep learning algorithms within the established mathematical and machine learning framework, demonstrating the ability to construct, train, and fine-tune neural networks for advanced tasks.
• Evaluate the performance of machine learning models, employing quantitative metrics to measure model accuracy and its effectiveness in addressing intricate problems.