HW4&5 feedbacks and appeals
|The feedbacks are online.|
If you didn't receive, please send a mail to Guy
Appeals on the HW will be accepted until 29/8
Last date to apply appels is this Friday, 3.8
|Reminder: the appeals should be applied by email to Shai and Nadav.|
Exam solution and appeals
|Exam solution is available in the website.|
Appeals on the exam should be filled using the standard CS form and sent to Shai and Nadav via email (do NOT submit to the cells in Taub floor 5).
Final date for submitting appeals will be published soon.
minor correction and clarification
|QDA decision boundaries may be any second order curve (parabola, hyperbola, ellipse) not only a parabola (slide 18 of the linear models tutorial was updated).|
Exams written with pencils cause trouble to the scanning devices resulting in blackening of other scanned papers. Therefore their use in the exam is prohibited, instead a pen should be used.
This will be enforced in the exam.
|1. There's no need for a calculator in the exam.|
2. Material covered only in "Supplement to Bayesian Learning - Probabilistic Based Linear Models" wasn't covered in the lecture, therefore isn't part of the course material.
3. The summary slides presented in the last lecture after Deep learning are available on the website.
Office hours before exam
|Guy: Monday, 9:00|
Nadav: Tuesday, 9:30
Tutorial 9 correction
|Regarding the last slide:|
Gaussian Naive Bayes assumes independence of features -> the covariance matrix of each class is diagonal.
However, they aren't necessarily equal.
Therefore, Diagonal LDA is a private case of both LDA and GNB.
GNB can be considered as diagonal QDA.
Updated the last slide in tutorial 9.
Lecture 13 is now available
Exams from previous years are now available
Correction to tutorial 13
|Slides 10, 22 and 28 were corrected.|
minor correction uploaded to tutorial 13
|slide 16 was updated.|
Tutorial 13 is available
Lecture 12 is now available
HW3 feedbacks on website
|HW3 feedbacks are now available on the website.|
updated tutorial 12
|Slide 13 was fixed to use the correct indices.|
HW2 feedbacks on website
|HW2 feedbacks are now (finally) available on the website.|
If there's someone that's still missing his feedback for the exercise, please email Guy and Nadav.
Final exercise is out
|Due date is the last day of the semester, 1/7.|
|In the exercise you are required to "Train at least one generative model and one clustering model".|
This means that each one should be used for (at least) one of the tasks.
This does not imply that you must use a generative model for clustering.
Tutorial 12 is out
Lecture 11 is now available
Tutorial 11 is out
Lecture 10 is now available
Correction: due date of exercise is June 19th
Exercise 4 is now available
|Submission is due June 16th.|
Tutorial 10 is now available
Lecture 9 is now available
|In order to receive your checked excercise please send an email with your ID and with your partner Cc'd to the following address.|
All tutorials will be held as planned this week
|Just making sure that there is no confusion -|
Nadav is indeed in reserve duty, but Guy will replace him and deliver the tutorial for Nadav's group at their usual schedule
Bottom line - All tutorials will be held as planned this week
Lecture 8 is now available
Nadav not available this week
|I'll be in מילואים this week.|
Please address your issues to Guy if needed.
Correction in tutorial 8
|Slides 33-34 were modified with the correct term for the variance.|
Tutorial 8 is now available
Reminder: No tutorials this week
|Monday's (Nadav) group had a תרגול השלמה at the first week of the semester.|
Thursday (Guy) is יום הסטודנט so no studies.
Lecture 7 is now available
Correction to tutorial 7 and exercise
|In tutorial 7, slide 15, the expression for the gradient has been corrected.|
In the exercise, comments section, second bullet has been corrected: the term "stratification" refers to maintaining the same ratio, while the comments suggests that you may change the ratio between different classes.
You may refer to lecture 4 (Elements of statistical learning part 1) , slides 30-31.
Tutorial 7 available
Lecture 6 is now available
Exercise 3 is now available
Exercise 2 deadline postponed to tomorrow
Tutorial 6 available
Bonus assignment A canceled
|This part is actually required in the mandatory assignment, that is: pre-processing steps should not take test-set statistics into account, not should they apply class-conditioned steps on the test data.|
Apologies for the late notice.
Lecture 5 is now available
Update to HW2 selected features submission
|We've uploaded a file named SelectedFeatures.csv .|
When submitting the assignment, please use this file to indicate which features you've chosen.
The PDF was updated to reflect this change (under "please submit", first bullet).
Tutorial 5 is available
Lecture 4 is now available
|Please use the one that was uploaded after 13|
Reminder: Guy's group has a tutorial tomorrow
|This is because the Technion tomorrow works in מתכונת יום ה'.|
Tutorial 4 is available
|Nadav's office hour for today (only) moved to 15:30.|
Lecture 3 is now available
End of registration for course
|The registration for the course is now over.|
All students that were on the waiting list have been enrolled.
Tutorial 3 has been updated
|Apologies for the last minute notice|
Office hours change
|Nadav's office hour has moved from 13:30 to 14:30.|
Guy's office hour this week only is canceled.
The Election challenge has started!
|(HW2 is now available).|
So is tutorial #3.
Lecture 2 has been uploaded
Updated version of HW1 uploaded
|We've uploaded an updated version that fixes minor issues and with clarification following questions that have risen in Piazza questions.|
Additional slots for registration to the course
|Those who aren't registered to the course, since it is full, please turn to מזכירות לימודי הסמכה to sign up for the waiting list.|
Students that decided not to take the course, although they are registered, are requested to do so formally, in order to allow for other students to take your place.
Homework 1 is out
|Unlike future exercises, this exercise is to be done individually (not in pairs). Due date 12/4/2018.|
If you have questions regarding the exercise, please post them in the Piazza:
New material uploaded
|The lectures for today, as well as an corrected version of tutorial 1 were uploaded.|
The tutorial is also available in PDF format for your convenience.
Tutorial substitution for Monday's class
|Students attending Monday's tutorial (Nadav's class), will have an additional tutorial this Sunday, 18:30 at class 3 (after the first lecture).|
Those who find it inconvenient may go to Guy's tutorial this Thursday instead.
This tutorial will be replacing the tutorial of the 14th of May, in order to maintain the assure the entire class are on the same page during the lectures.
|Welcome to Introduction to Machine Learning course spring 2018!|
Please note that in order to enlist to the course you must have previously completed the following courses:
236501 - Introduction to Artificial Intelligence
094412 - Probability
Discussion groups will be held in the Piazza platform.
Please enroll using the following link: