We announce a new talk in the JAII Lecture Series. Our presenter, Hinrich Schütze (LMU Munich) is a renowned expert in computational linguistics who will talk about scaling of Large Language Models (LLMs).

On July 6, Hinrich Schütze will talk about "Glot500: Scaling Multilingual Corpora and Language Models to 500 Languages" The talk will be virtual, you can find the link to the Zoom lecture below, we are very much looking forward to it!
Abstract:
Large language models (LLMs) are currently the most active area of research in NLP. Most work has focused on what we call "vertical" scaling: making LLMs even better for a relatively small number of high-resource languages. We address "horizontal" scaling instead: extending LLMs to a large subset of the world's languages, focusing on low-resource languages. Our Glot500-m model is trained on 500 languages, many of which are not covered by any other language model. I will talk about the major challenges we faced in creating Glot500: (i) finding, validating and cleaning training data for that many languages; (ii) evaluating performance of Glot500-m on languages for which native speakers and labeled datasets were not available to us; and (iii) determining the factors that ultimately make training on a language successful. We find that trying to reduce such factors to the so-called curse of multilinguality is naive and there is in fact also a "boon of multilinguality". We are in the process of making Glot500-c, our training corpus covering 500 languages, publicly available.

We announce a new talk in the JAII Lecture Series. The talk is part of a new thematic series of talks, "Co-Constructing Intelligence", which is jointly supported by the universities of Bremen, Paderborn and Bielefeld. This thematic lecture series focuses on the human-machine interface and on the question how humans and machines can learn together, or acquire new knowledge and skills together. Our presenter, Joanna Bryson (Hertie School, Berlin) is an expert in artificial and natural intelligence and is working to improve the governance and ethics of digital technology.

On April 20, Joanna Bryson will talk about "Principle Agents of AI – With Whom Do We CoCreate?" The talk will be a hybrid event taking place online via Zoom and in the CITEC building at Bielefeld University, Room 1.204. You can find the link to the Zoom lecture below, we are very much looking forward to it!
Abstract:
Artificial Intelligence is a set of techniques facilitating our capacity to navigate the information spaces afforded by substantial improvements in digital technologies and infrastructures. But whose is this “our” – who is gaining in capacities and at what costs? In this talk I will review a sampling of AI impacts in the individual, national, and global spheres. I will present my recent research in transparency for and through AI, governance of those that produce AI, and the transnational dynamics that may be obscuring and even compromising our agency. I use this evidence to suggest that ultimately ethics and responsibility are only sensible framings for relationships between peers, and artefacts are never true peers with organisms. I will conclude by discussing ways to meet our obligations for AI transparency and human centredness, as agreed globally through the UNESCO AI Ethics Recommendation.

We announce a new talk in the JAII Lecture Series. The talk is part of a new thematic series of talks, "Co-Constructing Intelligence", which is jointly supported by the universities of Bremen, Paderborn and Bielefeld. This thematic lecture series focuses on the human-machine interface and on the question how humans and machines can learn together, or acquire new knowledge and skills together. Our presenter, Enrico Motta, leads the Intelligent Systems and Data Science research group at Open University.

On the 23nd of March, Enrico Motta will talk about "A Hybrid Approach to Enabling Visual Intelligence in Robots"
Abstract:
Robots operating in the real world require sophisticated Visual Intelligence (VI), to be able to make sense of the variety of situations they may encounter. Computationally, the basic capability here has to do with Object Recognition – as a minimum, a visually intelligent robot needs to be able to recognise the content of its observations. Object recognition is typically carried out by means of Deep Learning (DL) methods, which provide the de facto standard for several Artificial Intelligence (AI) tasks, including image and speech recognition. However, despite the major successes in these and other benchmark tasks, from a cognitive point of view DL architectures still fare poorly when compared against human abilities, in terms of both efficiency and epistemology. From an efficiency perspective, DL methods are notoriously very data hungry, while humans are able to learn and generalise even from a single example. Even more importantly, from an epistemological point of view, a key aspect of human learning is that it goes well beyond pattern recognition. Humans learn concepts (not just patterns) and therefore are able to recognise instances of these concepts even when key features are missing (e.g., a car from which all the wheels have been removed or a cartoon pink elephant dressed in a tutu), thus avoiding the brittleness that characterises not just DL methods but also other types of AI systems. For these reasons, there has been much interest in recent years in hybrid computational architectures, integrating DL methods with other AI reasoning components, to try and address these issues. Our research is situated in the context of this paradigm and in this talk I will present the work that my team and I are currently carrying out, which has developed a hybrid architecture augmenting a Deep Learning approach with a variety of reasoning components drawn from Cognitive Science, with the aim of producing a new class of visually intelligent robots. In particular, I will illustrate the methodological process we have followed for the design of the architecture, which includes both a top-down analysis of the cognitive science literature as well as bottom-up experimental work. I will also report on our initial implementation of the reasoning components of the architecture, which include a size reasoner and a common-sense spatial reasoner, and on the experiments that we have carried out, which indeed demonstrate the improvement in performance afforded by the hybrid architecture.

You can find the link to the Zoom lecture below, we are very much looking forward to it!

We announce a new talk in the JAII Lecture Series. The talk is part of a new thematic series of talks, "Co-Constructing Intelligence", which is jointly supported by the universities of Bremen, Paderborn and Bielefeld. This thematic lecture series focuses on the human-machine interface and on the question how humans and machines can learn together, or acquire new knowledge and skills together. Our presenter, Ken Forbus (Northwestern University) is one of the early pioneers of AI, working on qualitative methods of reasoning as well as analog reasoning.

On the 19th of January, Ken Forbus will talk about "Qualitative Representations and Analogical Learning for Human-like AI Systems"
Abstract:
While there has been substantial progress in AI, we are still far away from systems that can learn incrementally from small amounts of data while producing results that are understandable by human partners. Our hypothesis is that qualitative representations and analogical learning are central in human cognition, and that these ideas provide the basis for new technologies that will help us create more human-like AI systems. We illustrate using examples from vision, language, and reasoning. These advances should support building software social organisms, that interact with people as collaborators rather than tools, which ultimately could revolutionize how AI systems are built and used.

You can find the link to the Zoom lecture below, we are very much looking forward to it!

For this December, we are happy to announce a public talk by Kary Främing (Umeå University) that is part of the ongoing activities of the TRR 318 "Constructing Explainability". Kary Främing is a professor of data science and an expert on explainable AI.
Abstract:
Current state-of-the-art Explainable AI (XAI) methods tend to produce a static "explanation" that shows how influential different features were for the outcome of an AI model, without offering any means of interaction to the user. One reason for that is that those methods are not capable of providing explanations in more than one way. The Contextual Importance and Utility (CIU) method differs from those methods by natively providing counterfactual "what-if" explanations in addition to the typical feature influence explanations. More importantly, CIU's so called intermediate concepts give the possibility to provide explanations on different levels of abstraction and different vocabularies, depending on the user. This flexibility doesn't automatically make explanations interactive but it does open true possibilities for creating interaction and adapting explanations to the needs of individual users.
You can find the Zoom lecture below, we are very much looking forward to it!

Since September, two new AI junior research groups at Paderborn University have been examining how to optimise and more effectively deploy machine learning. One is headed by Dr.-Ing. Oliver Wallscheid, the other by Jun.-Prof. Dr. Sebastian Peitz. The young researchers’ aim is firstly to use the combination of machine learning and expert knowledge to improve the quality of dynamic system models, and secondly to make training deep neural networks using ‘multi-objective optimisation’ models a more robust, efficient and interactive process. The Federal Ministry of Education and Research (BMBF) is funding both groups for three years with a total of approximately 1.8 million euros.

There will be a plenary talk, which will take place in the frame of the AI-initiatives of NRW including its graduate school DataNinja, and which will be streamed as zoom-webinar:

Who: Prof.Dr.Thomas Martinetz, Director Centre of AI (ZKIL), University of Lübeck

When: Wednesday, 12. Oct, 11:45 - 12:30, CITEC lecture hall and online.

Title: Artificial Intelligence for Medicine

Abstract: In Germany, up to 2030 the number of people older than 65 will increase by 50%. At the same time, the working population will shrink by 20%. This is an enormous challenge for society in general, but for the public healthcare system in particular. To keep the public healthcare system functioning, productivity in hospitals, for example, must be increased with the help of technological innovations but without jeopardizing human dignity. AI and digitization will play a key role, which is one of the research priorities of the University of Lübeck with the second largest university hospital in Germany and its well-developed computer science department. I will present projects and challenges for AI in the medical field.

Scientists from Paderborn University, in a research project with professional handball club SG Flensburg-Handewitt, have studied how the use of algorithms can minimize the risk of injury and improve athlete performance. The data comes from wearables, devices that produce vast quantities of data and provide insight into fitness and physical strain.

The know-how of four regional universities (Bielefeld University, Paderborn University, OWL University of Applied Sciences, Bielefeld University of Applied Sciences) flows together in the newly launched Digital Railway Systems degree program. Study location is the RailCampus OWL on the premises of DB Systemtechnik GmbH in Minden. Students are supported in projects by engineers from the companies involved in RailCampus OWL: Deutsche Bahn, Harting and Wago - and there will be even more companies in the future. Click on the links below to learn more.

We continue our JAII lecture series with a talk by Dr. Steffen Eger, currently substitute professor in Bielefeld, and also independent research group leader “NLP for the Humanities” at TU Darmstadt. Steffen will talk about "Evaluation Metrics, Natural Language Generation and Humanities Applications".
Abstract:
In this talk, I will present our ongoing works in the context of (i) evaluation metrics for natural language generation (especially machine translation) and (ii) natural language generation (NLG) itself. Regarding (i), I will speak about efficiency, explainability, reproducibility, among others, of recent BERT-based evaluation metrics. Regarding (ii), I will speak about abstract-to-title generation, poetry and argument generation as well as cross-lingual cross-temporal summarization. I will (iii) finally talk about humanities and social science applications including social solidarity in times of crises and change of biases over time.
You can register for the Zoom lecture below, we are very much looking forward to it!

On Friday, 29 April, Paderborn University celebrated the official opening of its new high-performance computing centre which houses the “Noctua 2” supercomputer. The roughly 340-square-metre computer hall, home to the HPC (High Performance Computing) system and its possible expansion, is at the heart of the research building.
"With more than 140,000 latest generation AMD Milan CPU processor cores and particularly powerful Nvidia A100 graphics card accelerators, we are increasing our existing computing power tenfold. This takes us into a completely new order of magnitude, enabling even the most demanding computer simulations in the natural, material and engineering sciences. In computer systems research, too, we are setting a focus on research and infrastructure that is unique in Europe by further expanding the computer partition with particularly energy-efficient, programmable FPGA-based hardware accelerators," explained Professor Christian Plessl, Chairman of the Board of PC2.

Together with the Faculty of Technology of Bielefeld University, the JAII will co-host the inauguration lecture by Prof. Yaochu Jin. The lecture will be a hybrid event, taking place in the CITEC lecture hall (located on the ground floor of the CITEC) and online via Zoom.
Prof Jin is an Alexander von Humboldt Professor for Artificial Intelligence (endowed by the German Federal Ministery of Education and Research) at the Faculty of Technology, Bielefeld University. His main research interests include data-driven evolutionary optimization, multi-objective optimization, evolutionary learning, secure and privacy-preserving machine learning, and evolutionary developmental systems. Prof Jin is presently the Editor-in-Chief of the IEEE TRANSACTIONS ON COGNITIVE AND DEVELOPMENTAL SYSTEMS and the Editor-in-Chief of Complex & Intelligent Systems. He was named by the Web of Science as “a Highly Cited Researcher” in 2019 and 2020.
Abstract:
Considering the limitations of the deep learning approach to artificial intelligence, this talk proposes to understand and emulate human intelligence from an evolutionary developmental perspective. We first provide a brief introduction to the biological findings about evolution and development of human brain and nervous systems. Then preliminary computational models of neural and morphological evolution and development are presented. Experimental results reveal that energy minimization is a main principle behind the organization of nervous systems and there is a close coupling between body and brain in evolution and development. Finally, we describe some recent advances in computational modeling of neural plasticity embedded in the reservoir computing and discuss their influences on the learning performance of echo state networks and spiking neural networks. The talks is concluded by an outline of future research.
You can register for the Zoom lecture below or join us in person in Bielefeld; we are very much looking forward to it!

North Rhine-Westfalia funds five research networks with a total of 81 Million Euros. One of those networks is SAIL, "Sustainable Life-cycle of Intelligent Socio-Technical Systems". SAIL builds on existing collaborations between Bielefeld and Paderborn University, the University of Applied Sciences Bielefeld, and the OWL University of Applied Sciences and Arts. The aim of SAIL is to strengthen interdisciplinary research ties in core AI, engineering, and the social sciences and humanities, as well as establishing a strong transfer chain from basic research to industry applications. As a central node of the research network, the JAII will play an important role in SAIL, and we are looking very much forward to the begin of the project in Summer.

For those who are interested in how AI can make water supply more sustainable, we recommend a blog entry from Bielefeld University that discusses the ERC Synergy Grant Project "Water-Futures", that JAII director Barbara Hammer is involved in.

The DataNinja Spring School is aimed at PhD students as well as master students or interested researchers from the broad area of Artificial Intelligence and Machine Learning. The goal is to provide in-depth tutorials on current hot topics which are spanning the spectrum of current Machine Learning approaches with a specific focus on explainable models that allow for inspection. Furthermore, the tutorials are geared towards providing hands-on experiences and empowering the participants to directly apply or transfer methods onto their own tasks or problems.

VEDLIoT – “Very Efficient Deep Learning in IoT” is an EU H2020 ICT-56-2020 funded research project running for 36 months, driven by challenging use cases in key sectors like automotive, automation, and smart home. The main objective of VEDLIoT is to develop the next generation of connected IoT devices utilising distributed deep learning.

You can become part of the project by applying for an Open Call project. Around ten research experiments are expected to be funded incorporating additional use cases in the project utilising the developed technologies. The call is open now and has a deadline on 8th of May 2022 at 23:59 CEST; each proposal can have a budget of up to 120.000 €. It is expected that open call projects leverage VEDLIoT technologies for their own AI-related IoT use case, thereby broadening the VEDLIoT use-case basis and help making the overall concept more robust. More information can be found on the project website.

We continue the JAII Lecture Series with a talk by Prof. Dr. Michael Beetz (University of Bremen, head of the Institute for Artificial Intelligence). He will talk about "Knowledge representation & reasoning in CRAM - a cognitive architecture for robot agents accomplishing everyday manipulation tasks."

Abstract: Robotic agents that can accomplish manipulation tasks with the competence of humans have been one of the grand research challenges for artificial intelligence (AI) and robotics research for more than 50 years. However, while the fields made huge progress, this ultimate goal is still out of reach. I believe that this is the case because the knowledge representation and reasoning methods that have been proposed in AI so far are necessary but too abstract. I propose to address this problem by endowing robots with the capability to internally emulate and simulate theirperception-action loops based on realistic images and faithful physics simulations, which are made machine-understandable by casting them as virtual symbolic knowledge bases. These capabilities allow robots to generate huge collections of machine-understandable manipulation experiences, which robotic agents can generalize into commonsense and intuitive physics knowledge applicable to open varieties of manipulation tasks. This will equip robots with an understanding of the relation between their motions and the physical effects they cause at an unprecedented level of realism, depth, and breadth, and enable them to master human-scale manipulation tasks.
You can register for the lecture below; we are very much looking forward to it!

We continue the JAII Lecture Series with a talk by Prof. Dr. Max Welling. He will talk about "The Impact of Deep Learning on the Natural Sciences".

Abstract: Deep learning has significantly changed the fields of speech recognition, computer vision and natural language processing, to name a few. What’s next? We are currently witnessing fast progress in the application of deep learning to scientific computation, such as protein folding, weather prediction, and molecular simulation. In this talk I will discuss a number of important tools and concepts that are pushing the frontier of this new disruption. In particular I will discuss graph neural networks for the prediction of molecular properties and the numerical integration of PDEs, and flows for molecular synthesis. We also discuss the important concept of incorporating symmetries in these models through equivariant layers. We will end with some thoughts on the impact on sustainability and health that these new breakthroughs might enable.
Max Welling is a research chair in Machine Learning at the University of Amsterdam and has various other positions, such as serving on the founding board of ELLIS. You can register for the lecture below; we are very much looking forward to it!

The Lecture by Prof. Kristian Kersting (AIML lab @ TU Darmstadt) in January had to be postponed. We will try to reschedule the lecture.

Abstract: Deep neural networks have shown excellent performances in many real-world applications. Unfortunately, they may show “Clever Hans”-like behavior—making use of confounding factors within datasets—to achieve high performance. In this talk, I shall touch upon explanatory interactive learning (XIL). XIL adds the expert into the training loop such that she interactively revises the original model via providing feedback on its explanations. Since “visual” explanations may not be sufficient to grapple with the model’s true concept, I shall also touch upon revising a model on the semantic level, e.g. “never focus on the color to make your decision”. Our experimental results demonstrate that XIL can help avoiding Clever Hans moments in machine learning. Overall, this all illustrates the benefits of a Hybrid AI, the combination of neural and symbolic AI.
The talk will be a virtual event. Please register here to receive the Zoom link:

The inauguration lecture by Humboldt professor Yaochu Jin, which was planned for December as a hybrid event, has been postponed due to the current developments in the Corona pandemic. The lecture will be rescheduled as soon as conditions allow (possibly in March 2022) and announced on this website.

First talk in JAII colloquium series: We are looking forward to a talk by Prof. Dr. Marius Lindauer, professor of machine learning at the Leibniz University of Hannover, on November 11, 16.15-17.45h.
Abstract: Automated machine learning (AutoML) supports developers by determining appropriate machine learning pipelines (incl. feature pre-processing, predictive model selection and hyperparameter optimization) and thus contributes towards achieving peak performance on a given machine learning task. However, we cannot ignore that humans are involved in the development of new machine learning applications. So, we have to strive for both, efficient AutoML and explainable AutoML.
This talk will first cover state-of-the-art techniques for AutoML, proposing a new technique for local Bayesian optimization, and how AutoML leads to new state-of-the-art performance on tabular datasets by using well regularized deep neural networks. In the second part of the talk, I will argue that full automation of machine learning design without any humans in the loop is not what developers are looking for, and propose ideas along the lines of explainable AutoML. Both together will not only lower the entrance barrier and increase performance of machine learning, but also of AutoML.
The talk will be a hybrid event, either join us in person in lecture hall H4, Bielefeld University main building or online via Zoom. Registration is necessary to either receive the Zoom link or to attend in person; the "3G" rule applies for the in-person event.
Please register here:

The “Intelligent Data Science Chatbot” project has developed a chatbot to simplify data analysis for users. The aim of the intelligent chatbot, developed within the scope of a research project by Prof. Dr. Axel Ngonga’s DICE group at Paderborn University, is to allow users to evaluate large data volumes quickly and easily. As a rule, more expensive tools, programming skills, and experts are required for this purpose, which is a cost factor that should not be underestimated, particularly for small and medium-sized enterprises. Created as a web-based platform, the chatbot helps users perform independent analysis of their data.

September 15th 2021 marked the kickoff of the Joint Artificial Intelligence Institute of the Bielefeld and Paderborn universities. Participants included the president of Paderborn and the rector of Bielefeld University, the founding members of the institute, as well as national and international guests.
Prof. Dr. Gerhard Sagerer, rector of Bielefeld University, and Prof. Dr. Birgitt Riegraf, president of Paderborn University, both stressed the importance of AI research for their university and the complementary expertise of both universities, as evidenced by several existing collaborations, also including the universities of applied sciences FH Bielefeld and TH OWL. Prof. Dr. Philipp Cimiano, co-director of the institute, sketched the core mission of JAII: He argued that the goal to pursue human-centered AI is about more than designing AI systems for humans, it is about establishing long-term partnerships where humans can dynamically shape the interface, process and outcomes of AI systems. The invited speaker, Prof. Dr. Frank van Harmelen from the VU Amsterdam, presented strengths and weaknesses of the traditions of symbolic and statistical AI, as well as new approaches to integrate both traditions, which was followed by a lively discussion of the potential of such approaches.

We are happy to announce the kick off event for the JAII, which will feature a greeting by the president of Paderborn University and rector of Bielefeld University, the director of the JAII and a talk by Prof. Frank van Harmelen, Vrije Universiteit Amsterdam.
Title: "How to build AI systems that both learn and reason?".
Abstract: Truly intelligent systems will have to both learn from experience and reason with acquired knowledge. Progress in the two fields of knowledge representation and machine learning has lead to an explosion of publications on how to build hybrid, "neuro-symbolic" systems that combine the techniques from this hitherto separate fields, with hundreds of new proposals appearing in the literature in recent years. In this talk I will describe a modular approach to such neuro-symbolic systems that allows us to re-interpret and systematize this rapidly growing literature on "3rd wave AI" system.

Bielefeld University and Paderborn University, together with the Bielefeld University of Applied Sciences and OWL University of Applied Sciences and Arts have participated in the "Netzwerke 2021" call of the MKW NRW. With their application, the universities plan to further strengthen their exisiting research ties by developing machine learning techniques for the entire life cycle of intelligent technical systems. This shifts attention from technological requirements, which are particularly prominent in the introduction phase, to societal and human needs, which are impacted by the long-term operation of intelligent technical systems. Insofar as the application is successful, the JAII would become the organizational hub of the project, facilitating the dissemination of best practices across all four collaborating universities.

Bielefeld University has been awarded an Alexander von Humboldt Professorship. It goes to the computer scientist Professor Dr.-Ing. Yaochu Jin. Jin is moving from the University of Surrey (Great Britain) to Bielefeld University in autumn 2021, complementing the research profile of the Faculty of technology through his focus on the multi-objective optimization of AI systems. The Humboldt Professorship enables researchers who have previously worked abroad to take up a professorship at a German university in order to carry out innovative research there. It is the most highly endowed international research award in Germany.

The German Research Foundation (DFG) has announced the formation of a new Collaborative Research Centre (TRR) “Constructing Explainability” at the universities of Paderborn and Bielefeld. Over the next four years, the foundation will provide about 14 million euros in support funding for this project. The highly interdisciplinary research program develops new approaches towards the explainability of AI that requires people to participate actively in socio-technical systems. The goal is to improve human-machine interaction, to focus on an understanding of algorithms, and to study this as the product of a multi-modal explanation process.

We are happy to announce the official inauguration of the research training group Dataninja on trustworthy AI. We will celebrate this with a kickoff meeting on May 3, (4 to 6 PM) – information and schedule can be found here. The training group is hosted at the CoR-Lab in Bielefeld and funded by the state of Norh Rhine-Westfalia.

For transportation to be C02-neutral in the future, more people and goods need to be transported by rail. For rail transport to fully exploit its potential, innovative technologies are required. This is what the RailCampus OWL project is all about. Bielefeld University, the University of Applied Sciences Bielefeld, Paderborn University, and and OWL University of Applied Sciences and Arts, Deutsche Bahn and business partners are developing a unique innovation network for intelligent and efficient rail technology at the Minden location.