Public talks by Paul Lukowicz Humane AI Net coordinator

Our PI Paul Lukowicz gave several talks in the last couple of weeks in leading AI-venues disseminating and reflecting on the research done with the network. For example, he gave a talk in an event about “Legal Protection Debt in the ML Pipeline” organized by FARI – AI for the Common Good Institute about how to curate data in the age of AI. He also gave a keynote in the AISafety & SafeRL workshop in the 32th conference: IJCAI International Joint Conferences on Artificial Intelligence Organization 2023. It targets the need to rethink the safety of AI systems beyond mere adherence to formal specs & and regulations, especially when it comes to interactions on societal levels.

Know more about IJCAI workshop? https://www.aisafetyw.org

DFKI Researchers Win Big with 20-Year Impact and Best Paper Awards in AI and Wearable Technology

Congratulations to the Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI) and its collaborators for winning two prestigious awards, highlighting groundbreaking contributions to AI and wearable technology.

🎉 20-Year Impact Award
Principal Investigator Paul Lukowicz, alongside co-authors Helene Brashear, PhD, Thad Starner, and Holger Junker, received the 20-Year Impact Award for their seminal paper, “Using Multiple Sensors for Mobile Sign Language Recognition.” This pioneering work set a strong foundation for sign language recognition using multi-sensor technology, establishing frameworks that continue to influence mobile and AI-assisted accessibility.

📰 Read the paper: https://lnkd.in/edJBVkJD
📽 Announcement video: https://lnkd.in/e_4dmM4S

🎉 Best Paper Award
The DFKI and RPTU Kaiserslautern-Landau team, including Lala Ray, Dr. Bo Zhou, Sungho Suh, and Paul Lukowicz, won the Best Paper Award for their innovative research titled “Selecting the Motion Ground Truth for Loose-fitting Wearables: Benchmarking Optical MoCap Methods.” This paper addresses a key challenge in wearable technology, offering a robust benchmarking approach for motion tracking in loose-fitting wearables, which promises to enhance accuracy and usability in real-world applications.

📰 Read the paper: https://lnkd.in/efYqvPW4
📽 Announcement video: https://lnkd.in/ec4Mrw7e

Kudos to all the winners for their outstanding work in advancing AI and wearable tech research!

 

Tutorial on Experimental Methodologies for Cognitive Human Augmentation

Check one of our senior researchers at Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI)), Agnes Gruenerbl joining forces with Prof. Dr. Thomas Lachmann and @jan spilski from the Cognitive Science department in RPTU Kaiserslautern-Landau and Giulia Barbareschi from the Media Fesign group in Keio University organizing a tutorial in the top tier conference UbiComp/ISWC 2023 in #cancun, mexico.

The tutorial is titled (Experimental methodologies for cognitive human augmentation) and it was widely attended and well perceived by the conference attendees.

Know more about the tutorial from here: https://lnkd.in/ec9h5yPj

 

HumaneAI project passed second review by European Commisssion

The HumaneAI project, recently underwent a comprehensive review by the European Commission. We are elated to announce that the review culminated in high praise for our initiative, highlighting the efforts of our collaborating European research centers, universities, and leading industrial partners.

Our EU-funded HumanE-AI-Net project stands as a beacon of collaborative excellence, uniting forefront AI laboratories with critical stakeholders in areas like human-computer interaction, cognitive sciences, social studies, and complexity sciences. This convergence of knowledge and innovation is pivotal in driving researchers to venture beyond the confines of their specialized domains and engage with the broader expanse of AI studies.

The central challenge of HumaneAI is not merely technical but also philosophical: How can we cultivate AI systems that are not only technically robust but also profoundly humane? Our mission is to develop AI platforms that can resonate with human emotions, adapt seamlessly to multifaceted real-world environments, and interact meaningfully within intricate social contexts.

This recent commendation by the European Commission serves as a testament to our commitment and the strides we have made so far. It propels us further in our pursuit to establish the bedrock principles of a novel scientific approach that envisions AI systems aligned with European values, bringing them closer to the aspirations and needs of European citizens.

As we bask in this moment of recognition, we remain deeply cognizant of the journey ahead. We are more determined than ever to ensure that the future of AI not only resonates with cutting-edge technology but also with the very essence of human values, aspirations, and well-being.

We express our heartfelt gratitude to all our collaborators, partners, and the entire HumaneAI community for their relentless dedication and contributions. Together, we’re not just pioneering technological advancements, but sculpting the very future of how humans and AI coalesce.

Stay tuned for more updates as we continue our journey in shaping the AI of tomorrow, today.

TrueLearn presented at HumaneAI review with European Commission

In informational recommenders, many challenges arise from the need to handle the semantic and hierarchical structure between knowledge areas. This work aims to advance towards building a state-aware educational recommendation system that incorporates semantic relatedness between knowledge topics, propagating latent information across semantically related topics.

We introduce a novel learner model that exploits this semantic relatedness between knowledge components in learning resources using the Wikipedia link graph, with the aim to better predict learner engagement and latent knowledge in a lifelong learning scenario.

In this sense, Semantic TrueLearn builds a humanly intuitive knowledge representation while leveraging Bayesian machine learning to improve the predictive performance of the educational engagement. Our experiments with a large dataset demonstrate that this new semantic version of TrueLearn algorithm achieves statistically significant improvements in terms of predictive performance with a simple extension that adds semantic awareness to the model.

Check the HumaneAI paper on Semantic TrueLearn: Using Semantic Knowledge Graphs in Recommendation Systems, S. Bulathwela, M. Pérez-Ortiz, E. Yilmaz, J. Shawe-Taylor, KGSWC’21 Workshop on Joint Use of Probabilistic Graphical Models and Ontologies, 2021, 10.48550/arxiv.2112.04368, (download)

Humane AI Net supports 2023 AI Beyond Borders Awards (AIBBA)

🏆The moment you’ve all been waiting for is here! We are proud to announce the 3 winners of the 2023 AI Beyond Borders Awards (AIBBA), recognizing the top German startups pushing innovative and international borders of AI technologies.

Coming at you live from the Startup Nation Stage at Bits & Pretzels in #Munich, the winners of this year’s AIBBAs are:

🥇PreciTaste in 1st place! PreciTaste provides food AI management solutions to support the food service industry in reducing waste and maximizing efficiency.
🥈vialytics in 2nd place! vialytics employs AI to track road conditions and detect pavement defects, such as cracks or potholes.
🥉ACCURE Battery Intelligence in 3rd place! ACCURE leverages cloud computing to improve battery performance, safety, and reliability.

Each of the winners will receive a cash prize amounting to 25,000€, 15,000€, and 10,000€, respectively. They will also receive a variety of additional benefits, which you can learn more about here: https://lnkd.in/e_e7bvQY

👏 A special thanks to our partners Humane AI Net, Bird & Bird, and KI Park e.V. for making our 2023 AIBBA season a success!

Looking for other ways to connect with us at Bits & Pretzels? Don’t forget to stop by our booth with German Accelerator and EXIST – Existenzgründungen aus der Wissenschaft!

Winners of the 2023 AI Beyond Borders Awards (AIBBA)
Winners of the 2023 AI Beyond Borders Awards (AIBBA)

HumaneAI releases AI research agenda for the telecommunications industry

In collaboration with the Humane AI net project, the German Research Centre for Artificial Intelligence (DFKI), Telefónica, GSMA, The European Telecommunications Network Operators’ Association (ETNO) and German Entrepreneurship organised a closed-door industry workshop which took place on 29 November 2022 in Munich, Germany to discuss promising artificial intelligence (AI) research areas for the telecommunications industry.

AI Research agenda for the telecommunications industry
AI Research agenda for the telecommunications industry

The objective of the workshop was to define the telecommunications industry research agenda in AI. More than 10 mobile network operators (MNOs) participated from Europe and from the rest of the world (Australia, Middle East) including Axiata, O2 Germany, Orange, stc, Telefónica, Telenor, Telia, Telstra, TIM, Turkcell and Vodafone, and they came together for two days to discuss what is needed from AI in the future.

During the workshop, the Humane AI net project explained the major trends in AI research. Then, several companies shared how they are using AI currently in their operations and business practises. Finally, the respective companies explained what AI capabilities they would be interested in for the future.

This last part has been the input for the telecommunications industry AI research agenda. The agenda is available in English and Spanish.

Humane AI Net support and host the ABC 2023

We were happy to have one of the partners (RPTU Kaiserslautern-Landau) in the Humane AI Net support and host the ABC 2023: The 5th conference on Activity and Behaviour Computing organized by Carl-Zeiss-Stiftung and Deutsches Forschungszentrum für Künstliche Intelligenz (DFKI). Our PI @Paul Lukowicz gave the keynote about the impact of Large language models on creating common understanding between sensor data.

 

Towards machines that understand people

The essay “Towards machines that understand people” is a project at HumaneAI Net and is out in the AI Magazine. Open access article here: https://onlinelibrary.wiley.com/doi/10.1002/aaai.12116

Meeting at the European Commission to announce the European Strategic Research Agenda (SRA) for AI

Today, there was a meeting at the European Commission to announce the European Strategic Research Agenda (SRA) for AI led by @Tailor in collaboration with the rest of the ICT-48 networks of excellence: Humane AI Net , @Elise, AI4Media, euROBIN project, and @ELSA. Paul Lukowicz, Marko Grobelnik, Dino Pedreschi, Andreas Keilhacker, & Passant Elagroudy represented Humane AI Net.

** Summarized SRA: https://lnkd.in/ewiPQD3D

DISCLAIMER: these are only some quick notes from some attendees highlighting discussion points

MAIN OBJECTIVES
1) Introducing the SRA to the community
2) Discuss what to do about large language models in the EU => interesting as a response to a recent call for action by the NoEs (https://lnkd.in/eFTZWg6e)


SOME DISCUSSION POINTS
1/ There’s a need to encourage European investment in AI to match other countries like china & USA
2/ There’s a need for larger visionary “change the world” projects to retain talents in Europe.
3/ There is a necessity to have a European alternative to chatGPT with European values ingrained (ActGPT capitalizing on industrial IoT).
4/ Request to create a separate AI track in the ERC applications from the rest of computer science.
5/ There’s a problem in joining forces & sharing the data across organizations => focus on building pipelines & each orga. trains on it’s own.
6/ ADRA has money & connections. Reach out for industrial collaboration & contribute to EU policies! (https://lnkd.in/eJ4aVXjQ)
7/ There’s a new research agenda being built tailored around industry called SRIDA. It is lead by ADRA. The SRA and SRIDA will drive a lot of funding calls in the upcoming period.
8/ Can we have a European startup accelerator focused on AI rather than only national initiatives?
9/ Current infrastructure is not the optimal one for LLMs (use GPUs that’ll be obsolete later or optimize algo. now?)
10/ Can the commission provide flexibility to the project’s outputs to readjust direction to the current changes?
11/ Add your content to AI on Demand platform (https://www.ai4europe.eu/)

Funding opportunities
1/ For startups / SMEs: [BB france] to train models – [German Entrepreneurship] generic AI.
2/ THE GRAND CHALLENGE
** Grand challenge like DARPA ones upcoming next September.
** Commission is currently collecting feedback on the process. There will be a group recommendation by NoEs on the process.
** All NoEs should now look: who can compete in the teams + recommendations for process + how to bring industry sponsors

** The winner will have access to HPC resources + (maybe man power). The commission is currently collecting exact needs.
** It will also be available for rising stars (aka. PhD level)

Upcoming events
1/ ELISE foundational model workshop in July with the big open source models.
2/ 26-sep here: better discuss the infrastructure situation.

HHAI23 conference hackathon

Before restarting the AI hackathon organized by Pietro Michelucci, time for a coffee break at #HHAI23 conference

Matteo Busso on “A context model for collecting diversity-aware data”

Interesting presentation by Matteo Busso on “A context model for collecting diversity-aware data”, co-authored with Xiaoyue Li, at the DHHAI workshop #HHAI2023 #DiversityAwareData Hybrid Human-Artificial Intelligence Conference

 Matteo Busso on "A context model for collecting diversity-aware data"
Matteo Busso on “A context model for collecting diversity-aware data”