DELETED: AI for Technical PhysicsLaajuus (5 cr)

Objective

After completing the course, the student - Knows how deep learning works - Can train models with appropriate architectures according to the data types - Can understand how to estimate the performance of the neural network The course develops the following generic skills: Digitalization, Leadership and development, Internationality, Sustainability and responsibility, Critical thinking, Identification and development of expertise, Interaction and communication

Content

Course content: - Introduction - Probability - Machine learning - Neuralnetworks - Training - Symmetries and acrhitectures - Images - Timeseries - Reinforcement learning - Self-supervised learning - Hyperparameters - Transformers - Adversial systems(GAN) - On the horizon.

Qualifications

- Python skills - LS00DB28 Probability Theory - LS00DB45 Linear Algebra - LS00DB36 Advanced Course in Linear Algebra b

Further information

The course is held every year.

Execution methods

Lectures 28 h, exercises 28 h, project work 20 h, exam 3 h, independent studies 56 h. Possibility to attend lectures and exercises remotely.

Accomplishment methods

Exam 60%, excersices 30% , projectwork 10%. At least 50% of the exercises done.