General
ELEC-E7870 Advanced Topics in User Interfaces (5 ECTS, Period II)
The link to the syllabus is outdated. This page provides all the relevant information regarding the course.
DL4HCI - Deep Learning for Human-Computer Interaction
Deep Learning (DL) for Human-Computer Interaction (HCI) features independent study modules that sit at the "practical middle" between theoretical DL on the one hand, and practical HCI applications on the other hand. Students coming from or interested in HCI with background in computer science, information science, engineering, and design will benefit from this self-paced 6-week course, which covers most of the major topics necessary to become proficient in both understanding and applying DL methods in HCI applications, striking a balance between theory and practice.
The DL4HCI course offers a practical view to HCI students who wants to prototype their own DL-based systems and applications. Students learn to identify challenging HCI problems and build the most adequate DL solution to solve those problems. The course covers DL models broadly, starting from discriminative modeling (classification, regression) and continuing to self-supervised learning (dimensionality reduction, clustering) and generative modeling. The lectures introduce DL applications and models through practical use cases, whereas the exercises apply the models to realistic HCI problems using Keras and Tensorflow DL libraries.
Learning objectives
Understanding of DL models for applications in HCI; Knowledge of foundational architectures used in DL applications; Ability to formulate and solve realistic problems in HCI using DL.
Teachers
- Responsible teacher: Luis A. Leiva, PhD
- Teaching assistant: Oleg Vlasovetc
- Guest lecturers: Pekka Ahtonen & Juho Kerttula (DAIN Studios), Peter Li (Silo.AI), Tomas Heiskanen (Fourkind), Kseniia Palin (Digital Workforce)
Contents
The DL4HCI course consists of lectures, exercises, assignments, readings, and a final project.
In the lectures, students will learn high-level concepts and the main ideas underlying DL methods. The exercises will train hard skills which are necessary to actually implement a DL solution in real-world applications. The assignments and readings are optional, aimed to complement the concepts introduced in the lectures and contribute to improving the student skills on DL for HCI. The final project covers the full DL4HCI application process, starting with problem identification, finding a DL solution to solve it, implementation of the solution, and evaluation.
Grading
Grading is based on points earned in assignments and the final project:
- Assignments: max 50 points; minimum for passing: 25
- Final project: max 50 points; minimum for passing: 25
Workload
- Contact teaching: 24h (lectures + course work)
- Independent work: 48h (final project + presentation)
- Bonus of 2 ECTS available
Study Material
Lecture notes and readings in MyCourses.
Participation
Due to the type of teaching and materials, participation in lectures is mandatory. The lack of attendance in some classes may be supplemented with an extra assignment, though you should not skip more than two classes.
The maximum number of students is limited. Students of University of Helsinki are welcome join the course but are asked to contact the teacher in advance.
Prerequisites
Python is the mandatory programming language for the course. Basic knowledge in math (algebra, calculus) and statistics is required.
A previous course on human-computer interaction and machine learning is recommended but not mandatory. From Aalto University, we recommend CS-C3210 Human-Computer Interaction and CS-E3210 Machine Learning: Basic Principles.
Tentative schedule
Week 1: 28.10.2019 [14:15 - 16:00 @ AS4] Introduction
Week 1: 30.10.2019 [16:15 - 18:00 @ AS6] Coding session
Week 2: 04.11.2019 [14:15 - 16:00 @ AS4] Discriminative modeling I: Classification
Week 2: 06.11.2019 [16:15 - 18:00 @ AS6] Guest lecture by Pekka Ahtonen & Juho Kerttula (DAIN Studios) and coding session
Week 3: 11.11.2019 [14:15 - 16:00 @ AS4] Discriminative modeling II: Regression
and coding session
Week 3: 13.11.2019 [16:15 - 18:00 @ AS6] Guest lecture by Peter Li (Silo.AI) and coding session
Week 4: 18.11.2019 [14:15 - 16:00 @ TU3] Self-supervised learning
and coding session
Week 4: 20.11.2019 [16:15 - 18:00 @ AS6] Guest lecture by Tomas Heiskanen (Fourkind) and coding session
Week 5: 25.11.2019 [14:15 - 16:00 @ AS4] Generative modeling
and coding session
Week 5: 27.11.2019 [16:15 - 18:00 @ AS6] Guest lecture by Kseniia Palin (Digital Workforce) and coding session
Week 6: 02.12.2019 [14:15 - 16:00 @ TU3] Model deployment and project preparations
Week 6: 04.12.2019 [16:15 - 18:00 @ AS6] Wrap up and project presentations
All lectures will be at TUAS building, Maarintie 8. Notice that Monday lectures are from 14:15 to 16:00 and Wednesday lectures are from 16:15 to 18:00. The classroom number also changes. We apologize for these changes, but since the course has been moved from period 3 to 2, it has been difficult to find a suitable room and time slot.
Final project presentations
Group 1 - Wednesday 11.12.2019 @ 16:00 (sharp!) AS4 room
- Neža Đukić
- Lukas Brückner
- Michelle Marabelli
- Xirui Fu + Xing Liu
Group 2 - Monday 16.12.2019 @ 10:00 (sharp!) AS4 room
- Yi-Chi Liao
- Laura Ham
- Luna Ansari
- Yu Zhang
- Xinyi Tu
- Asutosh Hota
- Morteza Shiripour