Research at IvI and the ILLC

Our research institutes are separated into individual Research Groups:

The Amsterdam Machine Learning Lab (AMLab) conducts research in the area of large scale modelling of complex data sources. This includes the development of new methods for probabilistic graphical models and non-parametric Bayesian models, the development of faster (approximate) inference and learning methods, deep learning, causal inference, reinforcement learning and multi-agent systems and the application of all of the above to large scale data domains in science and industry (‘Big Data problems’). 

AMLab is co-directed by Max Welling and Joris Mooij. Other faculty in AMLab include Ben Kröse (professor at the Hogeschool Amsterdam) doing research in ambient robotics, Dariu Gavrila (Daimler) known for his research in human aware intelligence and Zeynep Akata (scientific co-director of Delta Lab and co-affiliated with Max Planck Institute for Informatics) doing research on machine learning applied to the intersection of vision and language. 

Market places to share data in a trustable and transparent way.

The Complex Cyber Infrastructure (CCI) group is part of the Informatics Institute at the University of Amsterdam. CCI focuses on the complexity of man-made systems on all scales. This scale can be small, like the devices that you carry with you, or the apps they are running, or the communication protocols these apps use to interact. It can be also comprehensive, as in large systems such as data centres or multi-domain networks.

The complexity of these systems is caused by the fact that more and more cyber infrastructure – e.g. routers, switches, the cloud – is reprogrammable nowadays. This offers many possibilities, but it also makes the equipment more difficult to operate and less transparent. Further, there is the complexity of mapping in computational terms the data sharing requirements which are defined at societal level, through legislation, organizational policies, private data-sharing agreements, and consents.

The Computational Science Lab, led by Mike Lees, tries to understand how information is processed in natural settings through the study of a large variety of dynamic multi-scale complex systems with a focus on – but not limited to – biomedicine.

We study this ‘natural information processing’ in complex systems by computational modelling and simulation. An example is the spreading of the HIV virus: many processes on a large range of spatiotemporal scales play a role, from the molecular scale (e.g. the details of the entry of the virus into a cell) to the organism level (the sequence of events leading from an initial infection to the development of AIDS, and medication to keep the infection under control), and even to the population level (the actual spreading of the virus).

We rely on a variety of modelling approaches (such as Agent Based models, Cellular Automata, Dynamic Complex Networks, particle methods, and models based on differential equations), on multiscale modelling methods that capture the transmission and transformation of information up – and down the scales, on formal methods (theories of natural information processing) and on Problem Solving Environments (workflows, visualisation, multiscale coupling libraries and e-science infrastructures for distributed multiscale computing).

The Institute for Logic, Language and Computation (ILLC) hosts the COMSOC Group, which carries out research in the areas of computational social choice, algorithmic game theory, multiagent systems, and knowledge representation. Related courses include Computational Social Choice, Game Theory, Knowledge Representation and Reasoning (previously called Symbolic Systems), and Computational Complexity. The senior members of staff in this group are Ulle Endriss and Ronald de Haan.

The mission of the Computer Vision research group is to study core computer vision technologies and in particular colour processing, 3D reconstruction, object recognition, and human-behaviour analysis.

The aim is to provide theories, representation models and computational methods which are essential for image and video understanding. Research ranges from image processing (filtering, feature extraction, reflection modeling, and photometry), invariants (color, descriptors, scene), image understanding (physics‐based, probabilistic), object recognition (classification and detection) to activity recognition with a focus on human‐behavior (eye tracking, facial expression, head pose, age and gender).

Working on search engines with a modern artificial intelligence perspective.

IRLab focuses on bringing the right information to the right people in a fair and transparent way.

IRLab works on data-driven methods to understand content, to analyse and predict user behaviour and to make sense of context and information, all in the setting of search engines, recommender systems and conversational assistants. Applications are ubiquitous: tools to find documents on the web, recommend products, discover music and much more.

IRLab positions itself in the AI and Data Science research themes of the Informatics Institute.

The INtelligent Data Engineering Lab (INDElab), led by Prof. Paul Groth, investigates intelligent systems that support people in their work with data and information from diverse sources. This includes addressing problems related to the preparation, management, integration and reuse of data.

We perform both applied and fundamental research informed by empirical insights into data science practice. Topics of interest include: data supply chains, data provenance, transparency, information integration, automated knowledge base population, knowledge graph construction, and data semantics.

The Language Technology Lab (LTL) is a research group within the Informatics Institute at the University of Amsterdam.

One of the main goals of computer science in general, and information technology in particular, is to process, analyze, and present information through formal, algorithmic means. A tremendous amount of information is encoded in human language, e.g., news articles, books, scientific reports, social media postings, etc. The research area of language technology aims to bridge the gap between the practical value of developing concrete information technologies and the challenge to adequately model the complexities of human language.

Modern language technology approaches overwhelmingly use machine learning methodologies to automatically learn statistical models. The introduction of deep learning based models has lead to remarkable improvements. However, current data-driven models unfortunately suffer from very poor performance if the amount of available training data is limited. At the same time, many, if not most, real-world application domains lack large amounts of manually annotated training data. Addressing this fundamental problem requires investigating a spectrum of possible solutions, including sophisticated data sampling, semi-supervised training, knowledge transfer, as well as invariant learning. 

The Language Technology Lab (LTL) focuses on the accuracy, robustness, generalizability, and wide applicability of language technology methodologies and contributes to the development of novel theories, methodologies, algorithms and tools in this area. In particular, addressing generalizability across languages plays an important role in our research. Achieving high degrees of generalizability is often seen as a hallmark of intelligence. It is extremely challenging and therefore not only of immense theoretical value but also significantly extends the range of applicability of language technology approaches across many languages as well as across many real-world applications. 

Our work ranges from basic research in natural language processing to key applications in human language technology, and covers areas such as machine translation, summarization, question answering, language modeling, and image captioning.

Tackling multimedia data with AI techniques.

Multimedia Analytics Lab Amsterdam (MultiX) is a research group within the Informatics Institute at the University of Amsterdam. The group develops artificial intelligence (AI) techniques that help people understand large collections of multimedia data. Multimedia data can be imagery, text, video, graphs, but also other informational context like geocoordinates.

The predominant question the group tries to answer: How can you bring together all this information in a way that users get a better understanding of it? How do you combine them in a proper way? And how can you improve machine intelligence by learning from the user?

MultiX positions itself in the Data Science and AI research themes of the Informatics Institute.

Multiscale systems that make a difference.

The MultiScale Networked System (MSN) group is part of the Informatics Institute at the University of Amsterdam. The group focusses its research on multiscale systems e.g. cloud systems or clusters that define themselves by their dynamic size and scale, and on the network connecting them. The MNS group explores the emerging architectures that can support emerging applications across the future internet.

The predominant question that the group tries to answer: How can these distributed systems work as efficiently as possible? And how do these systems need to evolve to satisfy the constantly new application requirements?

The Institute for Logic, Language and Computation (ILLC) includes a strong group of international researchers who do cutting-edge research on Natural Language Processing and Computational Linguistics. We are passionate about language and aim to develop computational systems powered by machine learning that lead to better language technology.

Related courses within the MSc AI include NLP 1&2 , Deep Learning for NLP , Computational Dialogue Modelling , and Advanced Topics in Computational Semantics . The NLP researchers at the ILLC are organised into different groups with many things in common but slightly different focus. See the website of each group for more details:

Probabilistic Language Learning Group led by Wilker Aziz
Dialogue Modelling Group led by Raquel Fernández
Amsterdam Natural Language Understanding Lab led by Ekaterina Shutova
Statistical Language Processing and Learning Lab led by Khalil Sima’an
Cognition, Language and Computation Lab led by Willem Zuidema

Extra-functional behaviour of computer systems in full glory.

The Parallel Computing Systems (PCS) group is part of the Informatics Institute at the University of Amsterdam. It is the foremost research group in The Netherlands in the field of system optimization of multi-core and multi-processor computer systems. The PCS group looks at system performance, power/energy consumption, reliability, security & safety, but also the degree of productivity to design and program these systems: the extra-functional behaviour of computer systems in full glory.

The top research of the PCS group is indispensable for developments within, for example, Artificial Intelligence. In order to be able to cope with the increasingly demanding calculations in computer science, it is essential that computer systems become faster and more efficient. Without the skills of researchers within computer systems, AI, amongst others, was certainly not where it is today.

Advancing society through inclusive AI technology.

The Socially Intelligent Artificial Systems (SIAS) group is part of the Informatics Institute at the University of Amsterdam. The group focuses on civic-centered and community-minded artificial intelligence (AI) that aims to reduce inequality and promote equal opportunity in society.

SIAS arose out of the concern that AI is increasing inequality in society. The predominant question the group tries to answer: How can we use AI, and in particular learning systems, to advance society? And how can we do that in such a way that people from all corners of society benefit from it?

SIAS positions itself in the AI and Data Science research themes of the Informatics Institute, with clear links to the Computational Science and Systems & Networking themes.

The Theory of Computer Science group, led by Alban Ponse, is concerned with the development of theoretical foundations of computer science, based on logic and mathematics.

The aim is to seek greater understanding of fundamental computational techniques and their inherent limitations. The emphasis is not only on the abstract aspects of computing, but also on the application of theory in the field of computer science.

The focus is on developing theory and tools in the field of algebraic specification which can be used to specify, analyse, and verify concurrent communicating and programmed systems.

At the VIS Lab we make sense of video and images with artificial and human intelligence. The lab studies computer vision, deep learning and cognitive science. We are based at the Informatics Institute of the University of Amsterdam.

The VIS Lab embeds four public-private AI labs. QUVA Lab with Qualcomm, Delta Lab with Bosch, Atlas Lab with TomTom and AIM Lab with the Inception Institute of Artificial Intelligence. Spin-off’s from the lab include Kepler Vision Technologies and Ellogon.ai.