Seminar on Machine-Assisted Decision-Making
Saarland University, SS 2021

Seminar Overview | Schedule

Seminar Description

Machine learning (ML) based systems are frequently used to assist human decision makers, in scenarios ranging from medical diagnostics, to bail decision-making. To understand and moderate the effects of using ML based decision aids in practice, it is necessary to understand how human decision makers interact with such decision aids. In this seminar, we will review recent advances in the area of machine-assisted decision-making. We will focus on four main topics: (i) Background in psychology of decision-making, (ii) Comparing human and machine decisions, (iii) Human perceptions of machine decision aids, and (iv) Machine-assisted human decision-making.

Basic Details

Instructors

Teaching Assistants
  • Ayan Majumdar


Contact Details

To reach out to the seminar instructors and teaching assistants, you can send an email to madm-ss21-staff@lists.mpi-sws.org. Please, email individual staff members only when the communication is personal, and is not related to the seminar in general.


Seminar Schedule

The seminar schedule and the reading list can be found here. The seminar takes place every third Thursday from 16:00 until 18:00, starting on April 15, as detailed in the schedule.


Announcements
  • Students interested in participating in the seminar should register by 23:59 CET, April 5, 2021 using the UdS Seminar Assignment portal at https://seminars.cs.uni-saarland.de.

Seminar Structure

Seminar Format

This seminar will cover 4 topics related to machine-assisted decision-making:

  • Background in psychology of decision-making
  • Comparing human and machine decisions
  • Human perceptions of machine decision aids
  • Machine-assisted human decision-making

There will be no weekly classes. The seminar will consist of 5 lectures: an introductory lecture + 1 lecture per topic. The seminar will take place every third Thursday from 16:00 until 18:00, starting on April 15, as detailed in the schedule. The lectures will consist of student presentations of the papers listed in the reading list, followed by a group discussion.


Intended Audience / Prerequisites

This seminar is open to Master's students. Background in machine learning is required to take part in this seminar. Students must have passed one or more of the following courses, or equivalent:

  • Machine Learning
  • Optimization
  • Information Retrieval and Data Mining
  • Elements of Machine Learning
  • Probabilistic Machine Learning
  • Neural Networks

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


Reading Assignments

Students will be required to read all papers from the reading list, in order to be able to participate in group discussions.

Additionally, before each lecture, students will need to submit a review of one of the papers that will be covered during the lecture (as detailed in the reading list). Students can choose which of the three papers they wish to review. In total, students will need to submit 4 reviews (1 per topic).

Assignments should be sumbitted via Moodle. The assignments consist of filling out a review form on the assignment submission website. The review form consists of the following questions:

  • Paper summary (short, 5-6 sentences)
  • Three strengths of the paper
  • Three weaknesses of the paper
  • At least one way in which the paper could be improved
  • At least one way in which the paper could be extended

Assignment deadlines: Paper reviews must be submitted by 11:59 AM (noon) on the day of the relevant lecture (as detailed in the schedule).

Late submissions: We will apply a flexible slip date policy for late reading assignment 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.


Presentations

Each student will need to present one of the papers from the reading list to successfully pass the course. The presentations will be assigned based on student preferences, using a fair allocation algorithm. After the introductory lecture (April 15), students will be asked to state their paper preferences.

The presentation should be 20 minutes long. The presentations will take place according to the schedule.


Grading

The reading assignments will count towards 50% of the score (12.5% per paper), and the presentation will count towards the remaining 50%. To pass the seminar, a student must score at least 50% of the maximal possible points for both the paper summaries and presentation.