Machine learning on graphs is an important and ubiquitous task with applications ranging from drug design to friendship recommendation in social networks. The primary challenge in this domain is finding a way to represent, or encode, graph structure so that it can be easily exploited by machine learning models. Traditionally, machine learning approaches relied on user-de_ned heuristics to extract features encoding structural information about a graph (e.g., degree statistics or kernel functions). However, recent years have seen a surge in approaches that automatically learn to encode graph structure into low-dimensional embeddings, using techniques based on deep learning and nonlinear dimensionality reduction. In this course, we provide a technical introduction to key advancements in this area of representation learning on graphs, including matrix factorization-based methods, random-walk based algorithms, graph neural networks, and graph generation. In doing so, we develop a unied framework to describe these recent approaches, and we emphasize real-world applications involving large-scale social and biological networks.
Organizer: Professor Bin Yang, e-mail: byang@cs.aau.dk
Lecturers: Professor Jian Tang, Montreal Institute for Learning Algorithms
ECTS: 2,0
Time: May or June 2020 (delayed due to COVID-19. new dates will be decided.)
Place: Aalborg University
Zip code: 9220
City: Aalborg
Number of seats: 30
Deadline: April 14, 2020
Important information concerning PhD courses: We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3,000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.
- Teacher: Bin Yang
Welcome to Advanced Program Verification
Description:
With the growing dependence on and use of IT in society, the correct functioning of software applications is a primary concern for both developers and users. In this course students will become acquainted with a number of state-of-the-art techniques and tools for formal analysis and verication of software including type systems, static analysis, model checking, and automated theorem proving. The main focus of the course is to give an overview of the various tools and techniques, but will also dive deep into specific topics, e.g., secure information flow, programming languages for distributed computing or the use of theorem proving assistants. The course will take place over six (6) full days with the first four (as 2x2 full days) focusing primarily on an overview and the last two on a more specialised topic. Each course session will use a mix of lectures, exercises, hands-on projects, discussions and (student) presentations.
Prerequities:
Mathematical maturity, good working knowledge of programming languages (semantics, compilers, interpreters); advantage with basic knoweldge of type systems
Learning objectives:
Upon completion of this course the student will:
(*) know and be able to correctly apply the relevant general terminology used within the verication community
(*) know the fundamental theories and structures within the topics covered in the course
(*) be able to dene and apply basic program verication tools and techniques to simple/intermediate projects/programs
(*) be able to nd and further study relevant advanced literature within the eld of program verication and apply it to their own research.
Organizer: Associate Professor Rene Rydhof Hansen
Lecturers: Flemming Nielson (DTU), Aslan Askarov (AU)
ECTS: 2
Time: October-November 2020
Place:
Zip code:
City:
Number of seats: 25
Deadline:
Important information concerning PhD courses: We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3,000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.
- Teacher: René Rydhof Hansen
Welcome to Graph Powered Machine Learning
Description:
Many powerful Machine Learning algorithms are based on graphs, e.g., Page Rank (Pregel), Recommendation Engines (collaborative ltering), text summarization and other NLP tasks. There are even more applications once we consider data pre-processing and feature engineering, which are both vital tasks in Machine Learning Pipelines.
In this course, we will therefore investigate the intersection of graphs and Machine Learning and cover 3 parts:
(1) Graph-based Feature Engineering and Graph Algorithms:
We will start by looking at popular graph algorithm and their value for feature engineering for Machine Learning models.
(2) Graph Embeddings and Graph Neural Networks:
Utilizing graphs as input to Neural Networks is a very powerful combination. In this part we will consider different Embedding strategies and the eld of Graph Neural Networks.
(3) Machine Learning Metadata:
We know about the value of high quality and quantity for building high quality machine learning models, but for operating a production grade machine learning pipeline Metadata is equally important. In this part we will look at leveraging graphs to capture metadata and provenance information of our machine learning ecosystem.
Prerequities:
- A general background in computer science
- Familiarity with database management at undergraduate level
- Participants who have taken a graduate database course will benet from this additional background.
Learning Objectives:
The goal of this course is for students to familiarize themselves with recent algorithmic and methodological advancements in the area of graph powered machine learning. Particular focus is given to highlighting how the areas of graph data management and machine learning can benet each other.
Students will in particular deepen their knowledge of state-of-the-art popular graph algorithms, machine learning over graphs, and metadata management for machine learning pipelines. Researchers working with (large-scale) graph data will nd inspiration and tools for further improving their approaches and systems.
Participants at the end of the course should be able to:
- Describe and explain the main methods of graph empowered machine learning
- Compare and assess the limitations and advantages of the different methods
- Understand the opportunities offered in terms of research and industrial application by graph powered machine learning
Organizer: Professor Katja Hose
Lecturers: Jörg Schad, ArangoDB, Germany
ECTS: 2
Time: September 21-22, 2020
Place:
Zip code:
City:
Number of seats: 30
Deadline:
Important information concerning PhD courses: We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3,000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.
- Teacher: Katja Hose
Welcome to Aspects of Advanced Analytics
Description:
Big Data is being collected in ever larger amounts, e.g., from the (geo-social) web, social media, sensors/IoT devices in cyber-physical systems, or scientic experiments. However, it is necessary to go beyond merely storing and querying data to get a full overview and understanding. Instead, advanced analytics (data mining, prediction, forecasting,..) is applied to the huge data volumes to extract trends and patterns, and use historical data to predict future events, so-called predictive analytics. Recently, optimization has been added on top, resulting in so-called prescriptive analytics that prescribes the best course of action given the data and associated predictions and optimization goals. Other types of advanced analytics concern specic types of data such as sensor time series.
Traditional data analytics systems do not scale, and/or support only some of the tasks, or do not support the deep requirements for specic types of data, resulting in poor scalability, poor developer productivity and/or lack of functionality.
This course will cover selected aspects of advanced analytics including concepts, algorithms, and systems, with focus on 2 emerging areas: prescriptive analytics and time series analytics. The course will feature a mix of theoretical concepts and algorithms with practical hands-on exercises using specic advanced analytics systems on a number of realistic case datasets.
Prerequities:
A general background in computer science, and general familiarity with database management and analytics, as can be achieved through undergraduate courses in databases and data mining/machine learning, is expected. Participants who have taken graduate data management and/or analytics courses will benet from this additional background.
Organizer: Professor Torben Bach Pedersen
Lecturers: Postdoc Laurynas Siksnys, Postdoc Nguyen Ho, Professor Torben Bach Pedersen
ECTS: 2
Time: August - September 2020
Place:
Zip code:
City:
Number of seats: 20
Deadline:
Important information concerning PhD courses: We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3,000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.
- Teacher: Torben Bach Pedersen
- Description:
- Organizer:
- Lecturers:
- ECTS:
- Time:
- Place:
- Zip code:
- City:
- Number of seats:
- Deadline:
- Teacher: Enrique Encinas
- Teacher: Dimitrios Raptis
- Teacher: Mikael B. Skov
Welcome to Advanced Topics of Machine Learning: Reinforcement and Online Learning
Description:
This course covered one or more selected topics of recent advances in machine learning. In particular, the course covered reinforcement learning and online learning.
After this course, you will be able to have the knowledge on
1. Underlying mathematical and algorithmic principles of reinforcement and online learning 2. The key factors that have made reinforcement and on-line learning successful for various applications
In particular, the course covers basic reinforcement (e.g., TD learning, Q learning and State Space Models), online learning (e.g., regret minimisation, stochastic vs. adversarial, full information, semi-bandit, and bandit feedback), and Monte Carlo Tree Search.
Prerequities:
Machine Learning Basics
Organizer: Professor Bin Yang, Aalborg University, DK
Lecturers: Long Tran-Thanh, Southampton University, UK
ECTS: 2
Time: September 15 and 16, 2020
Place: Aalborg University, Selma Lagerlöfs Vej 300, 0.2.90 Cassiopeia Seminar Room
City: Aalborg
Number of seats: 60
Deadline: 25 August 2020
Important information concerning PhD courses: We have over some time experienced problems with no-show for both project and general courses. It has now reached a point where we are forced to take action. Therefore, the Doctoral School has decided to introduce a no-show fee of DKK 3,000 for each course where the student does not show up. Cancellations are accepted no later than 2 weeks before start of the course. Registered illness is of course an acceptable reason for not showing up on those days. Furthermore, all courses open for registration approximately four months before start. This can hopefully also provide new students a chance to register for courses during the year. We look forward to your registrations.
- Teacher: Bin Yang