Beskjeder
Here is the promised solutions to the sample exam PDF.
With the correct solution to 1 f) and 2 b), but not all solutions PDF.
On friday 02.06.2017 in seminar room Caml 10:15 (usual room for friday lectures), we will give a rough overview of the solutions and be available for questions regarding the exam and sample exam.
Here we post some sample exam questions sample exam (PDF fixed)
This is supposed to give some examples of the types of questions you can get on the exam. We have tried to cover a good share of the curriculum, but we have of course not covered everything.
OBS! This sample is larger and more comprehensive the actual exam.
Now Mandatory exercise 2 is available at
https://github.com/sigmunjr/INF5860_Oblig2
Deadline: may 16, 2017
During the group session on monday 03.04, I will be holding a Introduction to Tensorflow lecture.
For those who cannot attend, I recommend following the guides in https://www.tensorflow.org/get_started/get_started
See here for hints when implementing backpropagation of a two-layer net
For those you did not attend today's lecture: you can submit your feedback on the course by filling in this form . Fill in on paper and hand it in to the student expedition desk on 4th floor.
This friday I will give a review of image convolution, edge detection, designing filters in the image- and Fourier-domain, and how filter banks can be used to extract features in images.
This is indended as background for convolutional nets.
No weekly exercises given this week, you should focus on Mandatory 1
Now Mandatory exercise 1 is available at
https://github.com/anneschistad/INF5860_Mandatory1
Hints
Dearline: March 31, 2017
The lecture today (jan 27) will be held in English.
Starting from February 3, lectures will be in Norwegian. But an hour with questions/discussion in English will be set up after the lectures, and links to relevant youtube lectures given.
All lecture foils and exercises will be in English.
Kurset vil åpne inntil 10 plasser for etteranmelding. Disse åpnes tirsdag kl. 9. Du må melde deg opp i studentweb tirsdag og plassene fylles fortløpende.
The course will open up for 10 more students. It opens Tuesday at 9.00 in Studentweb.
Since the course is very crowded, but we expect that quite a few will decide that this course is not for them, we will allow some overbooking also if you have not yet registered for the course.
If you have not registered yet, send a mail to anne@ifi.uio.no and explain why you want this course, and your briefly your field of research, background in programming, math, and image analysis. We will open up to those with a decent background and good motivation.
Today's lecture moved to room C due to an exam
Velkommen til et morsomt, men krevende kurs!
Kurset er populært og fulltegnet, tildels overbooket fram til første oblig, da vi regner med endel faller fra. Vennligst møt på første forelesning hvis du vet du vil ta kurset. Send en mail til anne@ifi.uio.no hvis du er forhindret.
Minner om at kurset krever kunnskaper i programmering (inkl. Python), bildeanalyse, spesielt filtere og konvolusjon, lineær algebra og en liten bakgrunn i klassifikasjon.
Come to the first lecture as the course is full! Send an e-mail to anne@ifi.uio.no if you cannot attend the first lecture.