MA8701 Advanced statistical methods in inference and learning - Spring 2021
This is a phd course in statistics - and requires active participation. Zoom-link to activity sent by email to participants.
Course coordinator/lecturer: Mette Langaas
Guest lecturers:
- Benjamin Dunn (in L3) on algorithms for lasso and lasso friends
- Berent Å. S. Lunde (L6, 15.02) on boosting tre ensembles and hyperparameter tuning
- Samia Touileb and Jeremy Barnes: Analysing text with neural networks (L10, 15.03)
- Kristian Gundersen: Uncertainty in neural networks with Bayesian neural networks and Variational infence (L11, 22.03)
- Kjersti Aas (Part 4, L12+L13) on explainable AI/interpretable machine leaning
Scheduled teaching: Lectures are Mondays at 9.15-12.00 (zoom). On Monday 01.03, 08.03 and 26.04 we were hybrid in S21 and zoom.
Mette has office hours in 1236 and on zoom on Wednesdays at 14.15-15.00.
The same zoom link is used for both lectures and office hours.
Formal NTNU course description: https://www.ntnu.edu/studies/courses/MA8701#tab=omEmnet
Graphical overview of the course (click on the image for higher resolution version).
Course parts:
- Part 1: Shinkage
- Part 2: Ensembles
- Part 3: Neural nets
- Part 4: Explainable AI
Grading elements:
The grade for this course is pass/fail (has "always" been this way), and 70/100 score is required to pass each of the two elements: portfolio and final oral exam. The pass/fail at 70/100 is the standard rule for PhD courses at NTNU)
Evaluation plan (approved by IE faculty 17.02.2021)
1) Portfolio [75%] consisting of three element that will count equally:
- Project 1 (Part 1 topics) data analysis short report (in group of size 1-3) with peer review followed by evaluation by lecturer [25%].
- Project 2 (Part 2-4 topics) data analysis short report (in group of size 1-3) [25%].
- Oral presentation of chosen/assigned scientific article or manuscript (in group of size 1-3) [25%].
2) Individual oral exam [25%]
Course prerequisites:
- TMA4267 Linear statistical models
- TMA4295 Statistical inference
- TMA4300 Computer intensive statistical methods,
- TMA4315 Generalized linear models
- TMA4268 Statistical learning
Helpful knowledge
- TMA4180 Optimization
Programming/IT-knowledge
Good programming skills in either R or Python.
Reference group
Dates for meetings with reference group: 28.01, 11.03, 29.04.