Human-Centered Machine Learning
Saarland University Advanced Course, WS 2018

Course Webpage | Schedule

Basic Details

In recent years, machine learning have been increasingly used to predict, enhance, and even replace human decision making in a wide variety of off-line and on-line applications. In this course, you will learn about recent advances in human-centered machine learning. More specifically, in the first half of the course, you will get to know about a growing set of techniques to ensure that machine learning methods fueling algorithmic decisions are fair to all and its outputs are interpretable. In the second half of the course, you will learn about machine learning methods specifically designed to understand, predict and enhance human decision making, with a particular emphasis on online social and information systems.

October 16 '18 — February 7 '19
Tuesdays 14:00 — 16:00, Thursdays 14:00 — 16:00
E 1 5, Room 005 (SB), Building G26, Room 113 (KL)

Timeline

October 16, 2018 — December 4, 2018
Fairness and Interpretability in Machine Learning
December 6, 2018
Midterm
December 11, 2018 — February 6, 2019
Machine Learning for Understanding, Predicting and Enhancing Human Decision Making
Feb 7, 2019
Final

Course Overview

Instructors

Instructors

Email

Krishna Gummadi

gummadi@mpi-sws.org

Manuel Gomez Rodriguez

manuelgr@mpi-sws.org


Teaching Assistants

TAs

Email

Office Hours

Abhijnan Chakraborty

achakrab@mpi-sws.org

Abir De

ade@mpi-sws.org

Mondays 14:30 — 15:30,
Skype: abirdeee

Junaid Ali

junaid@mpi-sws.org

Fridays 15:00 — 16:00,
Room 309, Building E1.5, SB

Nina Grgic-Hlaca

nghlaca@mpi-sws.org

Wednesdays 14:30 — 15:30,
Room 312, Building E1.5, SB

Preethi Lahoti

plahoti@mpi-inf.mpg.de

Fridays 16:00 — 17:00,
Room 414, Building E1.4, SB

Utkarsh Upadhyay

utkarshu@mpi-sws.org

Wednesdays 14:30 — 15:30,
Skype: utkarsh.upadhyay


Announcements
  • The total number of points earned in the first part of the course can be found here: link. The total number of points is a weighted sum of the midterm (max 30 points) and reading assignments (max 30 points). The maximum number of points you can earn in the first part of the course is 50 points.

  • The results of the midterm can be found here: link.

  • The reading assignment scores for the first part of the course can be found here: link.

  • The course mailing list is now set up. Please subscribe at https://lists.mpi-sws.org/listinfo/hcml-ws18 using the same email address you use to send messages (postings from non-members will be rejected).

  • Students interested in participating in the course should register by 23:59 CEST, October 17, 2018 using the sign up form.


Mailing Lists
  • hcml-ws18@lists.mpi-sws.org: includes everyone involved with the course, the teaching staff as well as the students. Important announcements, such as exam schedules and assignment deadlines, will be posted on this list. Students can also use it to discuss assignments, and to exchange ideas and experience. Everyone should join and read this group mailing list daily. To subscribe, please visit https://lists.mpi-sws.org/listinfo/hcml-ws18.

  • hcml-ws18-staff@lists.mpi-sws.org: includes all members of the teaching staff, the instructors as well as the teaching assistants. Students should use this for all communication with the course staff.

  • Please, email individual staff members only when the communication is personal, and is not related to the course in general.

Course Schedule

The course schedule, recommended readings, and assignments can be found here.

Course Description

Intended Audience / Prerequisites

This advanced course is open to Bachelor and Master students. Background in machine learning is required to take this course. The students must have passed one or more of the following courses:

  • Machine Learning
  • Convex Optimization
  • Optimization
  • Information Retrieval and Data Mining
  • Data Mining and Matrices
  • Neural Networks: Implementation and Applications
  • The Elements of Statistical Learning
  • Probabilistic Graphical Models
  • Adversarial Machine Learning
  • Machine Learning for Natural Language Processing
  • Machine Learning for Image Analysis

The language of the course is English. All lectures, office hours, tutorials and communication with the course staff will be conducted exclusively in English.


Exams

There will be a midterm exam (covering material from the first half of the course), a final exam (focusing on the second half of the course), and an optional repeat exam (covering the entire course). The exams carry equal weight. All exams will be based on the material covered in lectures and readings. All exams will be open book and based on the material covered in lectures, readings, and projects.


Grading

To pass the course, a student must pass either the final or the repeat exam. Students may choose to take the repeat exam, even if they have passed the final exam, to try and improve their grade. To be admitted to the final and repeat exams, a student must (i) pass the reading and coding assignments and (ii) pass the midterm exam. To pass the reading and coding assignments, the sum of all points earned by a student in both the reading and coding assignments must be at least 50% of the maximal possible points. To pass the midterm exam, a student must score at least 30% of the maximum possible points in the exam. To pass the final exam or repeat exam, a student must score at least 50% of the maximal possible points in the exam.

Your course grade will be based on a weighted score computed from the points you earn in your successful examinations and your reading and coding assignments. If a student takes all three examinations, then the exam with the lowest result will not be considered when computing the course grade. Reading and coding assignment scores count towards 50% of the weighted score (25% for the reading assignments from the first part of the course, and 25% for the reading and coding assignments from the second part of the course) and exam scores account for the remaining 50% of the weighted score (25% each).


Reading Assignments

Before or after each lecture, you will be required to read and review a relevant paper.

Your reading assignments will consist of reading a paper and filling out a review form on the assignment submission website. The review form consists of the following questions:

  • Please summarize the paper in 5-6 sentences.
  • Write down three strengths of this paper.
  • Write down three weaknesses of this paper.
  • Write down at least one way in which this paper could be improved.
  • If this paper inspired you to think of new ideas, please describe them. (Optional field)

To access your reading assignments, please create an account on the course's assignment submission website. After you create an account, we will give you access to the reading assignments.

Assignment deadlines: Paper reviews must be submitted by 11:59 AM (noon) on their due dates (posted on the schedule website). Assignments should be submitted using the course's assignment submission website.

Late submissions: We will apply a flexible slip date policy for late submissions. Each student is allocated an automatic extension of 4 calendar days for the entire semester. Students can use the extension on any reading assignment during the semester in hourly increments. For instance, you can hand in one reading assignment 4 days late, or one assignment 2 days late and two assignments 1 day late.

Grading: Your reading assignments will be reviewed and graded. For each assignment, you can receive 0 points (fail) or 1 point (pass). For exceptionally well written assignments, you can get 1 additional bonus point. The results will updated on Sundays and posted on this website.


Coding Assignments

After the midterm (in the second part of the course), there will be two short coding assignments.