I will be removing access to the google drives for lecture/tutorial/homework etc materials on Sunday night (1st September). If you would like a copy of the material to keep for yourself, please make sure to download them before then.
Update: To make it easier to download I've put them all in one folder accessible HERE
A number of issues on the Q&A forum have been raised regarding the final exam. Some of these are about possible errors on the paper, others about the design of the questions and concerns about the marking. We will make all efforts to address these and ensure the assessment is fair and reasonable.
In the meantime, if you have specific queries or concerns please email the LiC or course admin and we will address them for you, rather than further posts on the forum.
Although most of your submissions have been correctly auto-marked, it appears that some submission files were corrupted during auto-marking. We're working to fix this and will update you as soon as that happens.
I understand some of you have questions about your marks, but as I have stated before, I am not responsible for marking of the homework, so please do not email me about this. You can contact the LiC or post your issue on the Homework2 forum.
Marks for homework2 should now be viewable via sturec / give.
Hi Everyone, sorry for the late notice but I've had to make my consultation slightly earlier, it will now be from 2-5pm on Friday the 16th in 4.03. Here is a summary of all consultation hours available before the final exam:
Peng Yi, Thursday 15th Aug, 3-5pm, K17 203
Omar Ghattas, Friday 16th Aug, 2-5pm, K17 403
Mike Bain, Monday 19th Aug, 12:30-1:30, K17 403
One final reminder to please fill out your MyExperience survey.
Peng Yi will also have consultation hours on Thursday 15th Aug from 3-5pm in K17 203
I hope your revision is going well - I wasn't able to secure a large enough room to hold a tutorial but I will make myself available for an extended consultation in room 4.03 in CSE from 2-5pm on Friday the 16th. Hopefully this is enough time to answer all your revision questions, though I'm happy to stick around for a bit longer if needed.
Please also remember to fill out your MyExperience surveys before the Friday deadline
Please take the time to provide feedback for the course/your tutors by filling out the MyExperience survey - which may be accessed HERE . The more detailed your feedback the better we can make this course in future years. Keep in mind that evaluations close on Friday
Hi – just a reminder we have consultation at the usual time tomorrow (Monday 12 August) and next week (Monday 19 August) for questions about the exam, etc.
The exam will be of 2 hours duration. It will count for 60% of the total grade for the course. The focus of the exam questions will be the lecture material, although any material covered in the course is potentially examinable.
Owing to reduced time for marking as a result of moving to the new trimester-based teaching schedule, the exam format will be multiple-choice answers. However, with the aim of assessing problem-solving as well as knowledge, the structure of the exam will adopt aspects of short-answer design, as follows.
The exam will be in two sections. The first section will have a number of multiple-choice questions, with each question on a particular topic we covered in the course.
The second, longer, section will comprise a selection of "problem-solving" type exercises on particular topics. You will need to complete each exercise, which will be immediately followed by a number (from 2--5) of multiple-choice questions on aspects of your solution.
All questions must be answered. For all questions only ONE answer is required.
Only UNSW-approved calculators may be be taken into the exam. No other materials or notes are permitted.
A sample of questions (with sample answers) from previous papers is available for download under "Lectures" for Week 10. Although the exam format is different from this term's exam, it should give you a sense of the topics that are likely to be covered in the exam.
Hi Everyone, I've been getting an increasing number of emails from students requesting an extension to the deadline because they have had a lot of deadlines on this week. I just want to make it clear that there will be no extensions granted for this reason, as the majority of students in this course would be facing the same problem and so an extension to any one team would be unfair on the rest of the students. We released the assignment very early on to give you enough time not to have left it to the last minute. Similarly, if you are requesting an extension based on last minute changes in your team, the answer is also no.
Extensions will only be granted for extraordinary circumstances, such as illness (if the request is made with some proof such as a medical certificate).
Hi – due to numerous requests, the deadline for Homework 2 has been extended.
The REVISED deadline is: Tuesday 6 August 23:59:59, 2019.
Please note the revised time as well, reverting to the traditional time of just before midnight !
Hi – the submission for homework 2 has been updated. Give will now run a syntax check on your submission and give you feedback on whether there are any syntax errors. If you previously submitted and are confident that your submission was syntactically correct you do not have to do anything further. Otherwise, please resubmit to have the syntax checked.
Hi Everyone, please take note of the following:
1. Tutorial material for next week has been released.
2. An explanation of how to submit your assignment has been added to the Assignment page, please see the last update "Submission Update".
3. Please also see the last update on the Homework 2 page about submission
has been released, the due date is 17:59:59, Monday August 5, 2019.
has been released, solutions for ensemble methods tutorial have also been uploaded to the drive.
now available, solutions for Kernel Methods tutorial now available. I've also added an extended solutions for Questions 2 and 4 for those of you that were still confused/wanted to dig deeper into the concepts. Enjoy the weekend
has been released, solutions for neural learning have also been uploaded to the drive.
Hi everyone, many of you are sending through your assignment proposals and I just wanted to make two quick notes:
1. I know many of you have done the deep learning course at UNSW and in that course, have covered topics such as Q-learning, DQN, NLP, and simple CNNs. There are no issues with you wanting to do deep learning projects for your assignment, but I just want you to be careful about reusing code that has already been submitted in previous assignments at the CSE. All code is checked against previous CSE submissions and this is an easy way to commit plagiarism - if you do use code taken from COMP9444 (or anywhere else for that matter) please be sure to reference it very clearly in your submission. The same goes for any material taken from COMP9318 or any of the MATH courses at UNSW.
2. If you are choosing one of the topics under 2:core topics: I'm sure you can see that these topics are extremely simple relative to other topics, and really are meant for those of you struggling heavily with the course material - after all we have pretty much implemented both models from scratch in the tutorials. If you are aiming to score highly in this assignment, I would urge you to consider a more challenging topic. Please also be aware that if you have a three or four person team, the expectations are higher (this is true of all projects) - so a three or four person team doing KNN will make it very difficult to achieve a high grade. Of course you will still be approved if you feel strongly about doing this topic.
Hi – seems the submission was down between about 6:30 and 10pm tonight. Looks like It is back up now.
I will extend the deadline to 11:59:59 tomorrow (Monday 8th July) to avoid anyone who tried to submit this evening being penalised.
Apologies for the inconvenience.
Some of you are having issues submitting, I’ve notified Mike and hopefully it should be sorted soon. Don’t panic, and please don’t send any more emails.
Materials have been released
Solutions have now been uploaded to the google drive.
I've added some further explanations of Naive Bayes and the spam email classifier question in the tutorial from last week. If you missed your tutorial due to the outage or if you were still confused by this question, hopefully this helps. Any mistakes are my own - please point them out on the forum. The file is on the google drive along with the rest of the material for tutorial 3 (NaiveBayesSupp.pdf)
Please check the Assignment and Homework 1 pages regularly (once a day), I will post updates there and don't want to make a course notice every time
I've been receiving a growing number of emails about course content related matters. Please do not email me unless your email contains something personal that cannot be shared on the forum, or if it is an assignment proposal. All other emails will be ignored from now on, please use the forum.
From tomorrow (Monday 1st July) there will be a consultation (office hours) slot each week:
Monday 12:30-1:30 in K17, Level 4, Rm 403.
See you there.
See 'Assignment' tab
1. Homework 1 has been released, see the Homework 1 tab
2. Tutorial materials for next week have been released, under 'Week 5/ Tutorial: Tree Learning '
3. Solutions to this weeks tutorial have been released (on the same drive where you found the questions).
Enjoy the weekend,
Power is back in Clancy so we are going to start now. If you miss the start (which will be about 2:35 pm) it should be recorded.
We will work out how to make up the lost time and let you know.
If the power outage has caused you to miss one of your tutorials, please feel free to attend any of the other tutorials offered today. You can find the list of available tutorials under the 'Timetable' heading on the left hand side tab.
See Tutorial: Classification (2)
The link has been posted in Tutorial: Classification (1)
The tutorials this week will be more practical, so it would be helpful if you can bring your laptop to the tutorial
Hi, apologies for the delay, hopefully you have enough time before your tutes to see what you can do with coding in this notebook !
I hope you're enjoying the first two weeks of the course and finding the tutorials helpful. There have been some requests for seeing derivations of least squares using matrices so I will make the notes I used for my tutorials available here . The notes contain solutions to the tutorial and a bit extra, I also include a notebook that should clarify the equivalence between the two approaches used in the notes to find the least squares estimates. Before anyone asks, you do not need to know matrix calculus for this course , this is just provided for those of you who are curious about it.
Some good reasons to go through this material:
1. Machine Learning literature heavily relies on vectors/matrices
2. When working with real world data in numpy, you are working with arrays which are just matrices, so it doesn't hurt to understand how to think about machine learning from this viewpoint.
Note that this is my own material, there will be an official solutions file for the tutorial released shortly, so any mistakes are my own.
To students in T12A, apologies today for the mixup in the room location, the room will be QUADG032 from now on.
If you missed out on the tutorial today my other tutes are T15A, T16A and F10A - or feel free to attend any of the other tutorials offered this week.
Just a reminder: tutorials start in Week 2 and run until Week 10 !
To keep things simple the tutorial or lab materials will be posted under "Course Work / Lectures" for the week in which the tutorial runs.
So, look for Week 2's tute stuff under "Course Work / Lectures / Week 2".
Just a reminder to everyone that our course schedule is (contrary to what the UNSW Timetable says):
Lectures – Week 1 to Week 9
Tutorials – Week 2 to Week 10
More information in the Course Outline.
Welcome to COMP9417 !
Please note: in the first week of the course (Week 1) we have NO TUTORIALS !
However, we do have some introductory practical exercises to get you started with Machine Learning in Python.
This is available by navigating to Course Work > Lectures > Week 1.
The activity will enable you to:
On Thursday of Week 1 we will provide sample solutions to the exercises. This will enable you to compare your solutions with those provided. Although we will not be formally assessing your solutions, this comparison will give you an indication of what will be expected of you as you go through the tutorials in this course.
Have fun, and please post your comments or questions under the activity rather than here !