Messages
Hi
We have uploaded today's exam questions with suggested solutions to the folder with earlier exams.
Since this is a a subject with many students, we will need 3 weeks for grading.
- Kai and Jan Tore
Hi everyone!
This is the last week of lectures and group sessions, and we recommend that those of you who have questions related to the syllabus or things you want some final assistance with before the exam to show up. Two group sessions will be cancelled this week:
-Monday's 14-16 session
-Wednesday's 10-12 session
So those of you planning to go to either of those, we recommend to visit any of the other sessions - see the full list under the schedule.
Have a nice and sunny week!
-Kai
Some of the exam questions may ask for calculations. There is a calculator in Inspera. You are advised to familiarize yourself with the calculator prior to the exam. You find the calculator here.
The detailed syllabus is listed on the weekly pages. On request, we have collected all the readings on one page which you find under practical information and here.
We have been asked how carefully you should read the Wikipedia texts related to week 13, History and Philosophy of AI. We tried to explain how the texts are meant to elaborate on the slides in the interactive session, May 8, see the slides.
We have got questions regarding which formulas you are supposed to know related to supervised learning, and we have given an overview which you find under practical information and here. We have made several corrections compared to the 2022 version.
Jan Tore
May 12: I spotted a typo in slides no 24 and 27, lecture 11. At one place it was written w_{m} instead of w_{j, m}. This is now corrected and the link on the weekly page is to the corrected version.
There is a similar typo in Marsland, formula 4.9 where a w_{\zeta} should have been w_{\zeta \kappa}. Unfortunately, we did not see this when we included the text from Marsland in the 2021 trial exam. It is now corrected in the updated solution proposal.
Jan Tore
Hi everyone!
On Wednesday this week (May 10th), we will devote the group sessions to summarizing some of the topics from earlier in the course, to help you revise that content as you begin preparing for exams.
The topics will be:
-At 10:15, Tobias will cover Supervised Learning
-At 12:15, Victor will cover Evolutionary Algorithms and Deep Learning.
The location will be Seminarroom Sed, as usual for the Wednesday sessions. Hope this will be a useful summary/reminder of these topics for many of you!
We have gotten a few questions on what to expect regarding coding on the IN3050/4050 exam in 2023. Here are our thoughts:
- Our expectations are not very different from previous years. As earlier years we will in the exam test your understanding of algorithms. Parts of that involves testing that you know the steps of algorithms well enough to describe them in detail, in a step-by-step manner.
- We may (and could also earlier years have) ask you to describe algorithms in code or precise textual descriptions, or otherwise test your detailed understanding of the algorithms in the course. Note that we will in such cases not be too picky about the syntax you use for code. Focus will be on testing your understanding of algorithms, not your knowledge of specific coding languages.
- We will however expect you to know key algorithms in enough details to implement them yourselves. And we may ask you questions where you have to show this detail-understanding.
Tomorrow's group session for group 4 will unfortunately have to be cancelled. Feel free to show up at any of the other group sessions in the next week for assistance - or contact us on the course forum!
-Kai
Monday 17 April, usual time, we will get a guest lecture from the company Elliptic Labs on "Classification of Sensor data using Neural Networks". Last year they gave a demo combining physical real time sensor data with deep learning in a Jupyter notebook. Really cool!
Afterwards, we will repeat last weeks syllabus and also look at recent developments, like ChatGPT.
See you Monday!
For those of you who had problems installing the original environment on Macs because of tensorflow, Amadu has made a recipe for modifying the yaml-file and installing tensorflow on macs. You find it here. Enjoy!
Hi everyone!
We just posted Mandatory Exercise 3, which focuses on Unsupervised Learning. See the details and deadline in the Mandatory Assignments subpage.
There will also be a third math tutorial by group teacher Tobias, following the two on Supervised Learning - this time with the math behind Unsupervised Learning as focus. The time and place of the tutorial is Thursday April 13th at 12.15 in Sed.
Happy easter and good luck on the final assignment!
-Kai and Jan Tore
The group is cancelled due to illness. You have to attend one of the other groups this week.
Apologies for the late notice.
The group session today, Monday March 13, at 12.15, is cancelled. You may attend another group later this week.
There will be a second math tutorial, Thursday March 16 at 12.15 in Sed. The main topics will be loss functions and backpropagation. Slides for the tutorial and from last week's tutorial are published here.
We know your mathematical background knowledge varies. Some have problems with the formulas. Hence, we will offer a tutorial on relevant topics from math, in particular
- Vectors
- Matrices
- Functions of several variables
- Partial derivatives
The tutorial will be in Sed, Thursday March 9, 12.15-14 and given by the group teacher Tobias. This is an additional offer and will not replace the ordinary group sessions. It is the place where you may ask all the questions regarded math, which you have been afraid to ask so far.
Dear IN3050/4050 students,
Unfortunately, Tobias has gotten a fever and today's group session from 10 to 12 will be cancelled.
Those of you normally attending Tobias' group may go to one of the other sessions this week - see the overview in the course schedule.
-Kai
The first mandatory exercise is now posted online, here. The deadline is February 24th. Note that for master-level students (IN4050) there is an extra task, which is optional for IN3050-students.
Good luck!
Hi everyone!
There was unfortunately some problem with recording the first lecture today. We clicked "record", but when attempting to upload the lecture video, the program wouldn't respond. We're checking with IT support if the recording can be rescued, but if not, you can also find all the important information from today's lecture on the slides, found under the weekly resources.
Remember also that you can join our course forum at https://in3050-discourse.uio.no/ where you can also ask questions to us in the course staff and to your fellow students.
-Kai