Moed B exam grades published | |
|
Dear all (and primarily those of you who have take moed B), Moed B exam grades were published, and the the notebooks were sent to the secretary for scanning. Appeals are by mail to me, not after Monday next week, 30.3.15 Yuval |
| פורסם ב-22/3/2015, 17:35:28 Created on 22/3/2015, 17:35:28 Создано22/3/2015, 17:35:28 تم النشر ب-22/3/2015, 17:35:28 |
Moed B 1-question poll - I'll be happy if you fill it | |
|
Hi, Added a 1-question poll that asks whether you plan to do Moed B. This is not obligatory: you can change your mind anytime till the exam. Link: form Yuval |
| עדכון אחרון ב-17/2/2015, 16:05:44 Last updated on 17/2/2015, 16:05:44 Последняя модификация17/2/2015, 16:05:44 تمت الحتلنة الأخيرة ب-17/2/2015, 16:05:44 |
The exams were checked | |
|
Tomorrow the final grade will be published, and the appeals are until next week, 17/2. Here is the grading: Q1: 2 points for each question (everyone got the whole grade for section 1.3) Q2: 2.1: 8 pts, 2.2: 6 pts. 2.3: 6 pts Q3: 3.1: 10 pts, 3.2: 10 pts Q4: 4.1: 6 pts, 4.2: 4 pts, 4.3: 10 pts Q5: 5 points each |
| פורסם ב-9/2/2015, 18:24:25 Created on 9/2/2015, 18:24:25 Создано9/2/2015, 18:24:25 تم النشر ب-9/2/2015, 18:24:25 |
Added exam solution under "course materail -> exams" |
| פורסם ב-7/2/2015, 17:41:44 Created on 7/2/2015, 17:41:44 Создано7/2/2015, 17:41:44 تم النشر ب-7/2/2015, 17:41:44 |
Corrections from ExamB of prev semeser - explanation | |
|
Section A3: I think that the suggested answer might not be correct. Adaboost increases the L1 margin, and SVM increases the L2 margin, and these might not be equivalent. Section B.4.f: The suggested answer is not correct. Section C.4: It seems that the answer is 2*(2^n - 1). For any x in {0,1}^20 and for any y in {0,1} we have to estimate P(x|y). These are 2*2^n parameters. But the sum over x of P(x | y=0) is 1. Thus, it is sufficient to estimate P(x | y=0) for all the x's except for one x, thus for 2^20 - 1 x's. And the same is true for y=1. |
| פורסם ב-4/2/2015, 15:13:16 Created on 4/2/2015, 15:13:16 Создано4/2/2015, 15:13:16 تم النشر ب-4/2/2015, 15:13:16 |
Correction of HW postponement | |
| The late submission HW3 was postponed to next week, Monday 9.2 and not 9.3 |
| פורסם ב-3/2/2015, 19:30:41 Created on 3/2/2015, 19:30:41 Создано3/2/2015, 19:30:41 تم النشر ب-3/2/2015, 19:30:41 |
Preparation to the exam | |
|
Hello, I was asked about how it is best to prepare to the exam. Here is all you should do, sorted from the type of preparation that should get the highest focus, to what should get the lowest preparation. Though everything which is in the list is important. 1) Understand all the material from the lecture notes (you can get help from the book if you don't understand something) 2) Understand the tutorial notes (the parts that were passed in class) 3) Understand the sections about linear regression and PCA from the book 4) Solve previous exams 5) Go over the demos and HW. This is also important to do, because some of the questions will be similar to or talk about topics from the HW/demos. |
| עדכון אחרון ב-1/2/2015, 22:16:45 Last updated on 1/2/2015, 22:16:45 Последняя модификация1/2/2015, 22:16:45 تمت الحتلنة الأخيرة ب-1/2/2015, 22:16:45 |
Review before the exam | |
|
Hi all, Nir will have a reception hour tomorrow (Monday) between 14:30 - 15:30. I will have a combined review tutorial with a reception hour on Tuesday, 3/3, between 15:30 - 17:30, in Taub 4. Prepare questions to ask me, and I will solve on board. Thanks, Yuval |
| פורסם ב-1/2/2015, 14:42:47 Created on 1/2/2015, 14:42:47 Создано1/2/2015, 14:42:47 تم النشر ب-1/2/2015, 14:42:47 |
HW3 checked | |
|
Hi all, HW3 was checked and will be returned tomorrow to the cells. Late submissions of previous homeworks were checked and will also be returned tomorrow. Late submissions of HW3 will be returned after the exam Yuval |
| פורסם ב-26/1/2015, 16:20:19 Created on 26/1/2015, 16:20:19 Создано26/1/2015, 16:20:19 تم النشر ب-26/1/2015, 16:20:19 |
Review tutorial before the exam + exam in english | |
|
Hi all, 1) There will be a review tutorial before the exam, combined with an office hour, next week. If you are intended to come, please fill in your availability in the following form until Saturday. Link: https://docs.google.com/forms/d/14AspJYVuAgtdkbP1YDdtIIS6sAwWaDZEhFwtK25jzl8/viewform 2) If any of you needs an exam in english, please mail me till the end of this week, and we'll create a translation. Thanks, Yuval |
| פורסם ב-25/1/2015, 13:46:27 Created on 25/1/2015, 13:46:27 Создано25/1/2015, 13:46:27 تم النشر ب-25/1/2015, 13:46:27 |
Tomorrow guest lecturers in class! | |
|
Tomorrow 21/1 I will host guest lectureres who do research in ML to talk about current trends in ML research.
I ask that all students show up in class because this is based on YOUR request!!! The topics that will be presented: Structured prediction NLP (Natural Language Processing) Deep learning Active learning |
| עדכון אחרון ב-20/1/2015, 13:55:21 Last updated on 20/1/2015, 13:55:21 Последняя модификация20/1/2015, 13:55:21 تمت الحتلنة الأخيرة ب-20/1/2015, 13:55:21 |
Added tutorial slides - exam preparation |
| פורסם ב-17/1/2015, 14:28:54 Created on 17/1/2015, 14:28:54 Создано17/1/2015, 14:28:54 تم النشر ب-17/1/2015, 14:28:54 |
Demo 10 + A question asked in the tutorial today | |
|
Demo 10 was published, to submit by 22/1. (I did tell that there will be no more demos, but we wanted to demonstrate the NN too). The formula for calculating the grade for the demos that I published in a previous announcement still applies, which means that it is enough to submit 9 demos. About a question that I was asked today in the tutorial: Question: How many samples are needed in order to get a reliable estimator? Answer: for each value of m, you can calculate the MSE. It is defined as the expectancy of (t - s)^2, where s is the real value of the parameter, and t is the value estimated, based on m random samples. See http://en.wikipedia.org/wiki/Mean_squared_error |
| עדכון אחרון ב-13/1/2015, 00:04:59 Last updated on 13/1/2015, 00:04:59 Последняя модификация13/1/2015, 00:04:59 تمت الحتلنة الأخيرة ب-13/1/2015, 00:04:59 |
HW Checked | |
|
Hello, HW2 was checked and will be returned to the cells tomorrow. Next time, please print all you work, including the graphs. Yuval |
| פורסם ב-11/1/2015, 18:00:24 Created on 11/1/2015, 18:00:24 Создано11/1/2015, 18:00:24 تم النشر ب-11/1/2015, 18:00:24 |
Tutorial Notes + Demo 9 | |
|
A new demo was uploaded to site, to submit by 22/1. This is the last demo. The calculation of the grade for the demos is as follows: three 0.5 grades will be converted into 1. And then the grade will be multiplied by 10/9. Tutorial notes were uploaded Yuval |
| פורסם ב-7/1/2015, 08:42:33 Created on 7/1/2015, 08:42:33 Создано7/1/2015, 08:42:33 تم النشر ب-7/1/2015, 08:42:33 |
corrections | |
|
The file neuralNet.py in HW3 was corrected. The tutorial notes of tutorial 10 were corrected |
| פורסם ב-5/1/2015, 20:14:49 Created on 5/1/2015, 20:14:49 Создано5/1/2015, 20:14:49 تم النشر ب-5/1/2015, 20:14:49 |
Question 2e from HW2 is not for submission |
| פורסם ב-5/1/2015, 18:40:42 Created on 5/1/2015, 18:40:42 Создано5/1/2015, 18:40:42 تم النشر ب-5/1/2015, 18:40:42 |
Uploaded tutorial 10 + lecturer's poll | |
|
Uploaded tutorial 10, about backpropagation in Neural networks. Please review the last lecture to prepare for class. If we'll have time then we will also talk about PCA. The lecturer's poll will be open soon. I'll be happy if you will write at least one point for improvement about the tutorial. (and if you will have time, then also about the lecture, HW, demo or anything that you think that can be improved) Thanks, Yuval |
| עדכון אחרון ב-1/1/2015, 23:31:13 Last updated on 1/1/2015, 23:31:13 Последняя модификация1/1/2015, 23:31:13 تمت الحتلنة الأخيرة ب-1/1/2015, 23:31:13 |
HW2 error correction + postponement | |
|
In Q4 (b), there was an error, where X_I(i) was written instead of X_(i). It has been corrected. Two more days of postponement are given to the HW |
| פורסם ב-1/1/2015, 11:38:45 Created on 1/1/2015, 11:38:45 Создано1/1/2015, 11:38:45 تم النشر ب-1/1/2015, 11:38:45 |
HW Directions + Bug in the Demo server | |
|
Added a background for HW2 Question 4. There is a bug in the demo server giving 1/2 a point for the submission of demo7. This issue will be fixed. You can keep submitting. Yuval |
| פורסם ב-31/12/2014, 21:40:07 Created on 31/12/2014, 21:40:07 Создано31/12/2014, 21:40:07 تم النشر ب-31/12/2014, 21:40:07 |
Wikipedia page about projections | |
|
For HW2 Question 4. Here are some definitions from wikipedia: Projections: http://he.wikipedia.org/wiki/הטלה_%28מתמטיקה%29 Orthogonality: http://he.wikipedia.org/wiki/אורתוגונליות |
| פורסם ב-31/12/2014, 18:43:15 Created on 31/12/2014, 18:43:15 Создано31/12/2014, 18:43:15 تم النشر ب-31/12/2014, 18:43:15 |
HW3 + Question from the last tutorial | |
|
HW3 was published. Submit by 22.1. For the first question: the backpropagation algorithm that you have to implement will be reviewed in the tutorial. You can start implementing some of the parts though. Second question: It is about MLE and will be learned in the next lecture and in the tutorial after the next one. It is not expected to take long. Enjoy working for HW, Yuval --- Regarding to the question that I was asked in the last tutorial, about why the fact that adaboost is a coordinate descent is good for us. Citing Nir's answer: "it is not 'good' or 'bad', it is just another way of viewing things." And in addition, its an exercise. |
| עדכון אחרון ב-31/12/2014, 18:11:16 Last updated on 31/12/2014, 18:11:16 Последняя модификация31/12/2014, 18:11:16 تمت الحتلنة الأخيرة ب-31/12/2014, 18:11:16 |
This Wednesday's lecture slides have been uploaded to the site. | |
|
Please prepare for class! |
| פורסם ב-29/12/2014, 22:36:11 Created on 29/12/2014, 22:36:11 Создано29/12/2014, 22:36:11 تم النشر ب-29/12/2014, 22:36:11 |
Office hours | |
|
Hi all, Till the end of the semester, there will not be a fixed time for my office hour. Instead, anyone who wants to meet can just mail me and we will schedule a time for that. Yuval |
| פורסם ב-29/12/2014, 10:36:04 Created on 29/12/2014, 10:36:04 Создано29/12/2014, 10:36:04 تم النشر ب-29/12/2014, 10:36:04 |
added demo 8, submit by 6/1 |
| עדכון אחרון ב-28/12/2014, 19:02:40 Last updated on 28/12/2014, 19:02:40 Последняя модификация28/12/2014, 19:02:40 تمت الحتلنة الأخيرة ب-28/12/2014, 19:02:40 |
Added tutorial notes for tutorial 9 |
| עדכון אחרון ב-26/12/2014, 13:15:38 Last updated on 26/12/2014, 13:15:38 Последняя модификация26/12/2014, 13:15:38 تمت الحتلنة الأخيرة ب-26/12/2014, 13:15:38 |
Added hints for Q3 |
| פורסם ב-26/12/2014, 00:32:08 Created on 26/12/2014, 00:32:08 Создано26/12/2014, 00:32:08 تم النشر ب-26/12/2014, 00:32:08 |
HW hints and postponement | |
|
HW2 is postponed to 4/1, Sunday, though we plan to publish HW3 at 1/1. Added hints for question 2 d+e. You are encouraged to report errors in any material supplied by the course (though first check if they were already corrected). Yuval |
| פורסם ב-25/12/2014, 23:26:30 Created on 25/12/2014, 23:26:30 Создано25/12/2014, 23:26:30 تم النشر ب-25/12/2014, 23:26:30 |
Video Lectures of Shai Shalev Shwartz | |
|
Hello, Added to the course material the video lectures of Shai Shalev Shwartz. There is not a 1 to 1 correspondence between the material in our course and between these video lectures, but there is a big intersection. Link: http://www.cs.huji.ac.il/~shais/IML2014.html Yuval |
| עדכון אחרון ב-25/12/2014, 22:42:12 Last updated on 25/12/2014, 22:42:12 Последняя модификация25/12/2014, 22:42:12 تمت الحتلنة الأخيرة ب-25/12/2014, 22:42:12 |
Added slides to yesteday's presentation | |
|
Hi all Do to confusion in slide related to "why L_1 promotes sparsity" I double checked and the slide is ok (there was an error in what I wrote on the board). I added another picture slide to explain better, please download the presentation from the site. Please prepare for next tutorial as it will be dense with little time for questions. Nir |
| פורסם ב-25/12/2014, 13:09:34 Created on 25/12/2014, 13:09:34 Создано25/12/2014, 13:09:34 تم النشر ب-25/12/2014, 13:09:34 |
Required reading for the next tutorial + note | |
|
Hi all, The next tutorial will be consisted of two halves: The first half will be about PCA. You are required to read: Chapter 23 (page 278): intro and section 23.1 (excluding: lemma 23.1, the proofs, section 23.1.1, and the pseudocode) The second half will be about viewing Adaboost as a Coordinate Descent. Make sure that you understandadaboost (section 10.2) and coordinate descent (taught in the previous tutorial) before the next tutorial. Also note that the subjects of the following lectures and tutorials is published, in order to allow you to prepare for them. |
| עדכון אחרון ב-24/12/2014, 18:23:56 Last updated on 24/12/2014, 18:23:56 Последняя модификация24/12/2014, 18:23:56 تمت الحتلنة الأخيرة ب-24/12/2014, 18:23:56 |
Demo Troubleshooting | |
|
Hello, If any of you had a problem with demo, that you cannot click and see the result, you may do one of the following: 1) Open the file using Spyder, open the code, switch to Console (instead of IPython Console), and then click the play button on top of the screen. Now it should work. 2) Add the following line to your code: matplotlib.use('TkAgg') |
| עדכון אחרון ב-23/12/2014, 13:27:02 Last updated on 23/12/2014, 13:27:02 Последняя модификация23/12/2014, 13:27:02 تمت الحتلنة الأخيرة ب-23/12/2014, 13:27:02 |
Added a new demo, for submission by 1.1.2015 |
| פורסם ב-23/12/2014, 04:24:41 Created on 23/12/2014, 04:24:41 Создано23/12/2014, 04:24:41 تم النشر ب-23/12/2014, 04:24:41 |
Tutorial Notes + Homework notes + Office hour | |
|
Tutorial notes were uploaded. The next tutorial will be about AdaBoost and coordinate Descent. Make sure that you understand AdaBoost (taught in lecture 7) before the tutorial. It is also recommended to look at "tutorial 8 intro" (though not compulsory). It is not long. In the tutorial we will solve a question from the homework of previous year. You can read the question ("tutorial 8" file) if you want to be prepared. In homework 2, there is a use of the term "infimum" (or "inf"). A wikipeia link for its definition was uploaded to the HW section. My office hour has moved to Tuesday 12:30, only for this week. Some notes about the homework: - When you are required to give a prove, it has to be formal. Don't leave holes in it. Intuitions cannot come instead of a formal prove. - The prove doesn't have to be long as long as it contains all the necessary details. It is not measured by its length. |
| פורסם ב-21/12/2014, 18:14:44 Created on 21/12/2014, 18:14:44 Создано21/12/2014, 18:14:44 تم النشر ب-21/12/2014, 18:14:44 |
HW1 checked and HW2 submission clarifications | |
|
HW1 was checked, and the grades appear in GR. If you submitted electronically, then mail me with the subject: "HW1 <id>", and I will reply with your checked file. The works that were submitted to the cell will be returned next week. For HW2: You have to submit in the cell all of your work (including the python code), and, in addition, submit electronically only the code. |
| עדכון אחרון ב-14/12/2014, 21:52:42 Last updated on 14/12/2014, 21:52:42 Последняя модификация14/12/2014, 21:52:42 تمت الحتلنة الأخيرة ب-14/12/2014, 21:52:42 |
Demo 6 + Tutorial Notes + HW calrifications | |
|
Demo 6 was published, to submit by 23.12 Tutorial notes of yesterday were uploaded HW2 Question 4 requires material you will learn only after Hanukkah |
| עדכון אחרון ב-9/12/2014, 17:23:55 Last updated on 9/12/2014, 17:23:55 Последняя модификация9/12/2014, 17:23:55 تمت الحتلنة الأخيرة ب-9/12/2014, 17:23:55 |
HW2 published | |
| to submit by 1/1/15 |
| פורסם ב-8/12/2014, 04:36:36 Created on 8/12/2014, 04:36:36 Создано8/12/2014, 04:36:36 تم النشر ب-8/12/2014, 04:36:36 |
New demo | |
|
A new demo. Submit by 11/12/2014. The demo is about SVM and kernels. Recall the polynomial kernel of dergee 2 we saw in class yesterday. In this demo, polynomial kernels of higher degrees are also used (which are used to find polynomial classifiers of higher degrees). In addition, there is another kernel there. This is called the Gaussian, or RBF kernel. Let K be the gaussian kernel. Then K(x, x'), is a dot product of phi(x) with phi(x'), and phi(x) is a vector with infinite number of entries. In the lecture slides you can see what if phi(x), if x is a vector with one entry. Another thing to notice there is the difference between hard magin and soft margin. Hard margin is when we want to find the largest margin, and not allow any vector to neither be inside the margin, nor be classified wrong. Soft margin is when do allow vectors to either be inside the margin or be classified wrong, and we minimize a weighted sum of the norm of w and the hinge loss (this is what we've done yesterday). This demo is a good opportunity to get a feel of how these things actually work. |
| פורסם ב-2/12/2014, 18:02:04 Created on 2/12/2014, 18:02:04 Создано2/12/2014, 18:02:04 تم النشر ب-2/12/2014, 18:02:04 |
Going back to room 201 Wednesday 10:30-12:30 | |
|
Starting tomorrow!! See you there. |
| פורסם ב-2/12/2014, 11:01:54 Created on 2/12/2014, 11:01:54 Создано2/12/2014, 11:01:54 تم النشر ب-2/12/2014, 11:01:54 |
Presentation for this Wednesday's class were uploaded | |
| Please prepare for the class. |
| פורסם ב-1/12/2014, 17:42:19 Created on 1/12/2014, 17:42:19 Создано1/12/2014, 17:42:19 تم النشر ب-1/12/2014, 17:42:19 |
Tutorial notes for tomorrow were uploaded |
| פורסם ב-30/11/2014, 21:52:08 Created on 30/11/2014, 21:52:08 Создано30/11/2014, 21:52:08 تم النشر ب-30/11/2014, 21:52:08 |
Hint for HW1 Question 4c | |
| Hint: for all x greater than some global constant, log(x) <= x/5 |
| פורסם ב-29/11/2014, 16:14:20 Created on 29/11/2014, 16:14:20 Создано29/11/2014, 16:14:20 تم النشر ب-29/11/2014, 16:14:20 |
Submission time for HW1 postponed + Reception Hours | |
|
Dear all, - The submission time of HW1 is postponed to Monday, 1.12.14. - Nir will have a reception hour next week on Sunday, 12:30-13:30. If you want to come, then make an appointment by mail. - Next week, my reception hour will be on Thursday, 10:30, and from the week after next week, it will be on Mondays, 15:00-16:00. Yuval |
| פורסם ב-26/11/2014, 14:46:47 Created on 26/11/2014, 14:46:47 Создано26/11/2014, 14:46:47 تم النشر ب-26/11/2014, 14:46:47 |
A new demo + Tutorial notes uploaded | |
|
Hello, The tutorial notes of today's class were uploaded. In addition, there is a new demo, about gradient descent, to submit by Wednesday, next week, 3.12.14. Have a good second and third thirds of the weekend! Yuval |
| פורסם ב-24/11/2014, 23:14:40 Created on 24/11/2014, 23:14:40 Создано24/11/2014, 23:14:40 تم النشر ب-24/11/2014, 23:14:40 |
Lecture today, tutorial tomorrow, lecture wednesday | |
|
Hello, This is a reminder for today's makeup class at 16:30-18:30 Taub 9. Before tomorrow and Wednesday, please read the chapters mentioned in material/lectures in sections "lecture 23/11" and "lecture 26/11". See you, Nir |
| פורסם ב-23/11/2014, 15:31:31 Created on 23/11/2014, 15:31:31 Создано23/11/2014, 15:31:31 تم النشر ب-23/11/2014, 15:31:31 |
Homework 1 - submission clarifications | |
|
Hello, Question 4 you can write on paper (no need to use word or lyx), but make sure it is readable. For the other questions, you can use a regular text editor, and there is no need to use an equation editor (for most of the cases). Just write it in a way that it is clear what you meant. (for instance, you can write Hn,k in question 3(c)). You can submit either printed or electronic. Thanks, Yuval |
| פורסם ב-20/11/2014, 00:16:17 Created on 20/11/2014, 00:16:17 Создано20/11/2014, 00:16:17 تم النشر ب-20/11/2014, 00:16:17 |
Makeup class | |
|
Hello all, Due to a strike in El-Al I will unfortunately not make it in time for tomorrow's class, as you've been informed by Yuval. A makeup class will be given on Sunday November 23rd 16:30-18:30, room Taub 9. I'm sorry but I cannot accommodate requests for alternative dates because we must meet before the next tirgul. I apologize for the inconvenience. Sincerely, Nir Ailon |
| פורסם ב-18/11/2014, 20:33:02 Created on 18/11/2014, 20:33:02 Создано18/11/2014, 20:33:02 تم النشر ب-18/11/2014, 20:33:02 |
A practice you can do in order to get a feel of the material taught today | |
|
Hello, If you want to get a feel of the material taught today, you can do the example in the tutorial notes, in the end of the second page. Even if you don't implement, but just do a few iterations with pencil, it might help. Before solving the example, try to think which w is argmin(L_S hinge). Yuval |
| עדכון אחרון ב-17/11/2014, 21:17:09 Last updated on 17/11/2014, 21:17:09 Последняя модификация17/11/2014, 21:17:09 تمت الحتلنة الأخيرة ب-17/11/2014, 21:17:09 |
The lecture on wednesday is cancelled | |
|
Hello, This week there will not a lecture on Wednesday. Nir is trying to get a classroom for Sunday, in the official time for supplementary lectures. An announcement about the final lecture time and place will be sent later. Yuval |
| פורסם ב-17/11/2014, 20:42:03 Created on 17/11/2014, 20:42:03 Создано17/11/2014, 20:42:03 تم النشر ب-17/11/2014, 20:42:03 |
Reading material (that we won't cover in class) and recommended reading for class | |
|
Hello, 1) In your free time, please read the section about linear regression in the book. It is a part of the course material, and may appear in the HW/exam. This is a basic concept of machine learning that we haven't had time to give in class. 2) tutorial notes + recommended reading for the tutorial were uploaded. 3) recommended reading for the next lecture: Sections: 15.1.1 15.1.2 15.2.1 15.2.2 in the book Have a good weekend :*) Yuval |
| עדכון אחרון ב-16/11/2014, 10:23:02 Last updated on 16/11/2014, 10:23:02 Последняя модификация16/11/2014, 10:23:02 تمت الحتلنة الأخيرة ب-16/11/2014, 10:23:02 |
Reception hour moved | |
| moved permanently to Monday, 15:30 |
| פורסם ב-14/11/2014, 15:17:09 Created on 14/11/2014, 15:17:09 Создано14/11/2014, 15:17:09 تم النشر ب-14/11/2014, 15:17:09 |
Demo 3 postponed instead of demo 2 | |
|
Hello, Sorry for the mistake. Demo 3 is postponed (by a week), instead of demo2. If somebody has planned to do the demo today, and he/she can't submit it on time because of my false announcement, then tell me. Yuval |
| פורסם ב-12/11/2014, 22:09:26 Created on 12/11/2014, 22:09:26 Создано12/11/2014, 22:09:26 تم النشر ب-12/11/2014, 22:09:26 |
Presentation for tomorrow (Lecture 4) published on material section |
| פורסם ב-11/11/2014, 23:10:00 Created on 11/11/2014, 23:10:00 Создано11/11/2014, 23:10:00 تم النشر ب-11/11/2014, 23:10:00 |
HW1 + Nir's Course | |
|
Added HW1. Submit by 29.11.14. For a different subject: Nir is giving a course called: "Algorithms and Theory for (Big) Data and Matrices". For those of you who are interested, it is on wednesdays, 14:30-16:30. Link: http://webcourse.cs.technion.ac.il/236602/Winter2014-2015/en/news.html |
| פורסם ב-11/11/2014, 21:24:22 Created on 11/11/2014, 21:24:22 Создано11/11/2014, 21:24:22 تم النشر ب-11/11/2014, 21:24:22 |
Added demonstration 3 | |
|
Hello, Demo 3 was uploaded. Submit by 20.11.14. Before doing that, read the section "Converting H into a set of homogenous half spaces", in the tutorial notes of tutorial 2. Yuval |
| פורסם ב-11/11/2014, 14:45:29 Created on 11/11/2014, 14:45:29 Создано11/11/2014, 14:45:29 تم النشر ب-11/11/2014, 14:45:29 |
Uploaded tutorial 3 | |
|
Tutorial notes were updated. We will go over the definitions from there, and do some exercises if there'll be time |
| פורסם ב-9/11/2014, 20:09:58 Created on 9/11/2014, 20:09:58 Создано9/11/2014, 20:09:58 تم النشر ب-9/11/2014, 20:09:58 |
lecture room changed | |
| It moved to Taub 301 starting from this week. The tutorials remain in room 201 |
| עדכון אחרון ב-9/11/2014, 15:06:10 Last updated on 9/11/2014, 15:06:10 Последняя модификация9/11/2014, 15:06:10 تمت الحتلنة الأخيرة ب-9/11/2014, 15:06:10 |
Algebra, and other material - optional | |
|
Hello, Here is a link to a machine learning course in stanford. It contains algebra review (under section notes). You can also look at other things there if you'd like to, but the book by Shai Shalev Shwartz is the recommended material. |
| פורסם ב-7/11/2014, 16:50:09 Created on 7/11/2014, 16:50:09 Создано7/11/2014, 16:50:09 تم النشر ب-7/11/2014, 16:50:09 |
office hours | |
|
Hello all, Nir's reception hours next week will be on Sunday, by appointment. Have a nice weekend. |
| פורסם ב-7/11/2014, 15:08:09 Created on 7/11/2014, 15:08:09 Создано7/11/2014, 15:08:09 تم النشر ب-7/11/2014, 15:08:09 |
Short probability review | |
|
Hello, Here is a short probability review, taken from a machine learning class in stanford university. It contains 12 pages, but worth reading (it is better to read some of it, than not read at all). In the class we don't have time to review probability, and it is important to understand that subject, in order to get a good understanding. If you've already read the material from the moodle that I published, then you don't need to read this. Link: http://cs229.stanford.edu/section/cs229-prob.pdf Yuval |
| עדכון אחרון ב-5/11/2014, 16:31:51 Last updated on 5/11/2014, 16:31:51 Последняя модификация5/11/2014, 16:31:51 تمت الحتلنة الأخيرة ب-5/11/2014, 16:31:51 |
Lecture 3 was changed | |
|
Hello, In order to reduce the amount of material, we decided to change lecture 3. It is now on the site. In addition, the exams from the previous semester were uploaded. |
| פורסם ב-4/11/2014, 16:24:20 Created on 4/11/2014, 16:24:20 Создано4/11/2014, 16:24:20 تم النشر ب-4/11/2014, 16:24:20 |
Uploaded lecture presentation + recommended reading for the lecture |
| פורסם ב-4/11/2014, 13:20:46 Created on 4/11/2014, 13:20:46 Создано4/11/2014, 13:20:46 تم النشر ب-4/11/2014, 13:20:46 |
Demo 2 Uploaded | |
|
Demo 2 was uploaded to the website. The submission is by 17/11. You can either do it now, or wait till the next tutorial, when we will talk about some concepts that appear in the demo. Yuval |
| פורסם ב-3/11/2014, 22:03:40 Created on 3/11/2014, 22:03:40 Создано3/11/2014, 22:03:40 تم النشر ب-3/11/2014, 22:03:40 |
Tutorial Notes Added + Review before tutorial | |
|
Hello, The notes for tutorial 2 were added to the course website. And another thing: Make sure you understand the following concepts before the tutorial: Sample Space (X), Label Set (Y), Hypothesis Class (H) and Training Set (S), VC Dimension. Yuval |
| עדכון אחרון ב-2/11/2014, 16:21:16 Last updated on 2/11/2014, 16:21:16 Последняя модификация2/11/2014, 16:21:16 تمت الحتلنة الأخيرة ب-2/11/2014, 16:21:16 |
Recommended Reading | |
|
Hello, The recommended reading material was added to course web-site under "Syllabus". Currently, it contains chapters 1-6, excluding some proofs. It is really recommended to go over it, either before the next tutorial, or before the next lecture. In the next tutorial we are going to do some exercises about VC dimension. Yuval |
| עדכון אחרון ב-30/10/2014, 16:56:12 Last updated on 30/10/2014, 16:56:12 Последняя модификация30/10/2014, 16:56:12 تمت الحتلنة الأخيرة ب-30/10/2014, 16:56:12 |
Probability material | |
|
Hello, Here is the recommended material about probability. (probably, we will not make a review in the next tutorial). It is taken from the moodle web-page of the course: 104222. Link: http://moodle.technion.ac.il/course/view.php?id=1339 Under the link, open the lecture notes. The first four chapters are the relevant ones. |
| פורסם ב-28/10/2014, 21:33:00 Created on 28/10/2014, 21:33:00 Создано28/10/2014, 21:33:00 تم النشر ب-28/10/2014, 21:33:00 |
Added demonstration 1 | |
|
Added demo1. The submission is by 11.11.14. It contains 2 parts. A link to the site was added under "Assignments". Note that the lecture and tutorial notes are added to the website before the class. Yuval |
| עדכון אחרון ב-28/10/2014, 01:28:49 Last updated on 28/10/2014, 01:28:49 Последняя модификация28/10/2014, 01:28:49 تمت الحتلنة الأخيرة ب-28/10/2014, 01:28:49 |
Books for the course | |
|
I am working on buying more books for the library, will update you if I make any progress. FYI: the Kindle edition on Amazon costs $29.49. Even if you don't own a Kindle device, both Android and Ipad tablets have a free Kindle app that allows you to read Kindle content. |
| פורסם ב-22/10/2014, 14:07:01 Created on 22/10/2014, 14:07:01 Создано22/10/2014, 14:07:01 تم النشر ب-22/10/2014, 14:07:01 |
Homework 0 was uploaded - Python Tutorial and Exercise | |
|
Added HW0. It includes links to relevant tutorials and an exercise. Not for submission. It may take some time, thus you are recommended not to postpone it. |
| עדכון אחרון ב-21/10/2014, 01:23:54 Last updated on 21/10/2014, 01:23:54 Последняя модификация21/10/2014, 01:23:54 تمت الحتلنة الأخيرة ب-21/10/2014, 01:23:54 |
Lecture 0 was uploaded to the course website (to be taught today) |
| פורסם ב-20/10/2014, 14:37:38 Created on 20/10/2014, 14:37:38 Создано20/10/2014, 14:37:38 تم النشر ب-20/10/2014, 14:37:38 |
Welcome to Introduction to Machine Learning | |
|
The course site was updated. Do not use Google Chrome for surfing the site because it might not show the updates. It is recommended to take a look at the General Info and at the Syllabus. In order to do the demos and the HW it is necessary to know how to program in python, and in its libraries: Numpy, Scipy and Matplotlib. Under course material there are relevant docs and tutorials. If you have forgotten the basic concepts of probability, it might be useful to go over it. All the required material in the topic of Machine Learning will be taught in class. However, you might want to take a look at the book "Understanding Machine Learning: From Theory to Algorithms". The CS library contains copies. On 20/10, between 16:30-17:30 there will be an introductory lecture by Nir Ailon instead of a tutorial. Have a successful semester, and feel free to consult if you have questions, Yuval |
| עדכון אחרון ב-7/10/2014, 13:57:57 Last updated on 7/10/2014, 13:57:57 Последняя модификация7/10/2014, 13:57:57 تمت الحتلنة الأخيرة ب-7/10/2014, 13:57:57 |