Course info page of DAT550 (Data Mining and Deep Learning, Spring 2021)

Overview

• Course Structure
• Labs
• Schedule
• Updates

Updates

• Jan 15 2021
  • Lecture 5 (Evaluation) and Lecture 6 (Regression Trees and Ensemble methods) are posted
  • Zoom recording for the lab session from week 2 is posted
  • Assignment 2 is already public!
• Jan 02 2021
  • Lectures 3 (Data Exploration) and Lecture 4 (Decision Trees) slides are videos are posted
• Dec 30 2020
  • First zoom session on Jan 4th 2021 at 12.15 pm is confirmed
  • Lectures 1 (Introduction) and Lecture 2 (Data) slides and videos are posted
  • Autograder is ready and assignment 1 is published

Structure

• Lectures:
  • The lectures are currently fully online until further notice
  • Video lectures will be published on canvas with a link in Schedule at least 1 week before the discussion
  • Video discussions will happen twice a week from 12.15 to 14.00 on Mondays and Wednesdays
  • You must watch the video lectures and come to the live zoom call discussions (zoom link will be shared in canvas)
  • First dicussion will happen live on 4th Januaray 2021 12-2 pm (video lectures are already published)
• Labs:
  • The labs are online in week 1 and 2 and the rest will be hybrid (you can show up physically with a prior appointment with the TAs)
  • Make an appointment with the student assistants for the help with lab
  • 3 ungraded (pass/fail) mandatory assignments for qualification to final exam
• Project:
  • 40% of the final grade is assigned to the group project (working code + written report + presentation)
  • 60% of the final grade is based on a written home exam

Labs

Autograder

This course uses Autograder, a tool developed at the University of Stavanger for students and teaching staff to manage the submission and validation of lab assignments. All lab submissions from students are handled using Git, a source code management system, and GitHub, a web-based hosting service for Git source repositories. Thus, basic knowledge of these tools are necessary.

Students push their updated lab submissions to GitHub. Every lab submission is then processed by a custom continuous integration tool. This tool will run several test cases on the submitted code. Autograder generates feedback that let the students verify if their submission implements the required functionality. This feedback is available through a web interface. The feedback from the Autograder system can be used by students to improve their submissions.

Grading and Collaboration Policy

Lab Assignments

All the three lab assignments must be completed to satisfaction (passed) in order to sit for the exam. Cooperating for assignments is tolerated only at the high level ideas but copying code is not allowed.

Projects

The project accounts the 40 % of your grade. You will get a grade based on the novelty, quality and complexity of the project, along with the ability to show use of appropriate models, preprocessing techniques and evaluation strategies. The projects are done in groups of 2-3.

Plagiarism Warning

Any form of cheating, plagiarism, i.e. copying of another student’s text or source code, will result in the grade F, and immediate disqualification from the course and may be reported to the university for administrative processing. Copying code from fellow students and borrowing code from past year also counts as cheating. You may collaborate ideas with your fellow students but do not copy the code. It is perfectly fine to get help from fellow students to debug something or get tips if you are stuck somewhere but copying code is forbidden. We will do plagiarism detection using a software which takes into account same with only minor changes to structure of the code, method names or variables. Committing acts that violate Student Conduct policies that result in course disruption are cause for suspension or dismissal from UiS.

Don’t cheat. It’s not worth it!

Git and GitHub

Git is a distributed revision control and source code management system. Basic knowledge of Git is required for handing in the lab assignments. There are many great resources available online for learning Git. A good book is Pro Git, which is available for free here. Chapter 2.1 and 2.2 should contain the necessary information for delivering the lab assignments.

GitHub is a web-based hosting service for software development projects that use the Git revision control system. An introduction to Git and GitHub is available in this video.

You need to sign up for a GitHub account to get access to the needed course material.

Autograder Registration

Follow the steps below to register and sign up for the course on Autograder. Here are two short videos recorded by Prof. Hein Meling (inventor of Autograder) describing these steps in the context of a different course DAT320 (same steps should work with DAT550): Part 1 and Part 2.

1. Go to GitHub and register. A GitHub account is required to sign in to Autograder. You can skip this step if you already have an account.

2. Click the "Sign in with GitHub" button in Autograder to register. You will then be taken to GitHub's website.

3. Approve that our Autograder application may have permission to access to the requested parts of your account. It is possible to make a separate GitHub account for only this (and other) courses if you do not want Autograder to access your personal one with the requested permissions.

Signing up for the Course on Autograder

1. Click the Plus (+) menu and select “Join course”. Available courses will be listed.

2. Find the course and click Enroll.

3. Wait for the teaching staff to confirm your Autograder registration.

4. You will then be invited to the course organization on GitHub and two separate repositories. You will need to navigate to each of these links and accept these invitations:

   • Navigate to the course organization dat550-2021 accept the invitation.
   • Navigate to the assignments repository and accept the invitation.
   • Navigate to your private https://github.com/dat550-2021/username-labs repository and accept the invitation. Remember to replace username in this link with your own GitHub username.

   Several invitation emails will also be sent to the email address associated with your GitHub account. However, emails from GitHub can sometimes take a while to arrive.

5. Once you have accepted the invitations, you will get your own repository under dat550-2021, which is the course's organization on GitHub.

Submitting the assignment once it is working

The procedure used to submit your lab assignments is explained in the lab submission process.

Group Signup on Autograder

Projects will be done in groups. So this step is only needed once you are ready to work on your project.

1. Read the policy about group assignments. Find and agree with another student to form a group. We prefer groups of two, but will allow groups of three. It is important that all group members agree to contribute equally to the group assignments.

   If you prefer to work alone, you must still create a group.

2. Agree on a name for the group. The name will be used as the group's GitHub repository. We prefer group names that identifies the persons in the group. The group name cannot be changed later.

3. Navigate to the course's left menu bar and select “New Group”.

4. Enter the name of the group in the textbox above the list of students.

5. In the dialog, find your own name via the “Search for students” text box. Click the Plus (+) symbol to add yourself to the group.

6. Repeat the above step for the other group members.

7. Click the “Create” button.

Discord DAT550 Server Registration

1. Go to Discord and register. A Discord account is required to sign in to communicate with the teaching staff during lab hours. You can skip this step if you already have an account.

2. To join our UiS DAT550 Discord server, navigate here.

3. Make sure you are registered in autograder and you have accepted all invitations from github to join the dat550-2021 GitHub organization before next step.

4. Once connected to the server, please register with our bot, @dat550_helperbot, by sending it a direct message:

   !register username
   

   where username is your GitHub username.

   Note that to register with the bot, you must previously have registered with Autograder. Hence, please make sure that you have joined the dat550-2021 GitHub organization and registered with Autograder first. Make sure you have accepted all the Github invitations before registering in Discord.

   If you need help with registering on the server, send a message in #new-users.

Schedule

(Lab submission deadlines are marked with bold.)

W	Date	Chap.	Topic	Resources	Travels / Remarks
1	4/1	1,2	Introduction	slides video
	6/1	2	Data processing	slides video
	8/1	Lab	Get started with in autograder, play with jupyter notebook, numpy, pandas, sklearn	hands-on-1
2	11/1	2	Data sampling, exploration, visualization	slides video, hands-on-2
	13/1	3	Classfication (Decision Trees)	slidesvideo
	15/1	Lab	Work on lab assignment 1		Zoom recording
3	18/1	3	Evaluation, Decision trees regression	regression tree slides regression tree video Evaluation slides Evaluation video
	20/1	6	Ensemble methods, Random forests	slides video
	22/1		Lab assignment 1 continued
4	25/1	ch 3 from MMDS book	Locality Sensitive Hashing (LSH)	slides video
	27/1		LSH Continued
	29/1		Lab assignment 1 Due and Lab assignment 2 announced
5	1/2		Dimensionality reduction SVD, PCA	slides video
	3/2		Recommender Systems (Collaborative Filtering)	slides video
	5/2	Lab	Work on Lab assignment 2 (LSH)
6	8/2	6	Rule based, Nearest Neighbors classifiers	slides video Exercise
	10/2		Bayesian classifiers (Naive Bayes and Bayesian Networks)	slides video
	12/2	Lab	Work on Lab assignment 2 (LSH)
7	15/2	6	Bayesian classifiers	hands-on-5
	17/2	6	Linear Regression	slides video
	19/2	Lab	Lab assignment 2 Due and Lab assignment 3 announced, projects announced
8	22/2		Logistic Regression	slides video
	24/2		Support Vector Machines (SVM)
	26/2	Lab	Work on Lab assignment 3 (SVD/Recommendation Systems) Project selection due
9	1/3		SVM (continued)	slides video
	3/3		Neural Networks, Deep feed forward networks, Backpropagation	slides video
	5/3	Lab	Work on Lab assignment 3 (SVD/Recommendation Systems) / Work on project
10	8/3		Convolutional Neural Networks (CNNs)	slides video
	10/3		CNN Applications	slides video
	12/3	Lab	Lab assignment 3 Due Work on project
11	15/3		Recurrent Neural Networks	slides video
	17/3		GRU, LSTM and Transformer models
	19/3	Lab	Work on project
12	22/3		Guest lecture 1 (Deep Reinforcement Learning)	slides video	Given by Diana Spencer
	24/3		Guest lecture 2 (Deep Reinforcement Learning)	slides video	Given by Diana Spencer
	26/3	Lab	Work on project
13	29/3		Easter (No lecture)
	1/4		Easter (No lecture)
	3/4		Easter (No lab)
14	5/4		Easter (No lecture)
	7/4		Deep learning on graphs (network embeddings)	slides video
	9/4		Work on project
15	12/4		Clustering (K-means, K-medoids, Spectral, EM)	slides video
	14/4		Guest lecture 3 (Time series analysis)	slides video	Given by Magnus Book
	16/4	Lab	Project + report due
16	19/4		Project presentation + Q&A
	21/4		Project presentation + Q&A