Kostas Karpouzis, is an assistant professor at the Department of Communication, Media and Culture, Panteion University of Social and Political Sciences in Athens, Greece. In the past, Kostas served for almost 20 years as a Research Director at the Institute of Communication and Computer Systems, National Technical University of Athens. Since 1998 and participated in more than twenty research projects funded by Greek and European bodies; most notably the Humaine Network of Excellence, within which he completed his post-doc in the field of mapping signals to signs of emotion, leading research efforts in emotion modelling and recognition, the FP6 IP CALLAS project, where I served as Area Leader of Affective applications, the FP7 TeL Siren project (Technical Manager), which was voted Best Learning Game in Europe for 2013 by the Games and Learning Alliance Network of Excellence, and the H2020 iRead project, which produced Navigo, the winner of the GALA Serious Games competition for 2018.
Kostas is currently serving as a member of the IEEE Explainable AI Working Group, which aims to produce a Standard for XAI – eXplainable Artificial Intelligence – for Achieving Clarity and Interoperability of AI Systems Design. Kostas is also the Student Activities and Computer Chapter Chair for IEEE Greece and a Member of the National Bioethics and Technoethics committee and the Ethics Advisory Board for ICCS-NTUA.
Currently the General Chair of the Foundations of Digital Games 2022 Conference, the Chief Editor for Frontiers in CS/Human-Media Interaction section, a member of the Editorial Board for Springer’s Journal on Multi-modal User Interfaces, Personal and Ubiquitous Computing, and International Journal of Applied Intelligence; and MDPI’s Big Data and Cognitive Computing and Virtual Worlds journals. In 2016, I co-edited a book on “Emotion in Games: Theory and Practice”, published by Springer.
Besides this, Kostas is involved in a number of science communication activities, most notably Famelab Greece and openscience.gr. I’m also an advocate for technology and CS in primary schools, participating in the Girls Go Coding initiative and serving as an Ambassador of EU Code Week in Greece (until 2018). I have participated as a speaker in 3 TEDx events, including TEDxAthens in 2019, while in 2016, I authored a lesson on the TED-ed platform titled “Can machines read your emotions?”; the lesson surpassed 300.000 views in its first week.
DVP term expires December 2025
Emotion, games and interfaces
The advent of ubiquitous and wearable sensors and computing power and, especially, natural interfaces in the form of speech-based commands or hand-held devices enables users to interact with computers, gaming consoles, and portable devices in a human-like fashion, surpassing the conventional paradigm of keyboards, mice and hand-held controllers. This emerging paradigm opens up new means of non-verbal communication: users can shrug their shoulders to indicate indifference to the options presented to them, nod when agreeing or shout when angry, thus producing feedback that computing systems can take advantage of to provide a truly natural and personalized experience. In addition to this, both seasoned gamers and casual users can interact with computer and console games in the same manner as they would when playing a conventional physical or mental game. This talk aims to introduce games not as a leisure or entertainment activity, but as a means to educate children and adults. Natural interaction and expressivity, personalization (starting from the user interface, all the way down to producing individual content based on what players enjoy), along with accessible computing and aesthetic emotions constitute concepts which can benefit from studying user behaviour and expressivity when playing games.
AI/ML for games for AI/ML
Digital games have recently emerged as a very powerful research instrument for a number of reasons: they involve a wide variety of computing disciplines, from databases and networking to hardware and devices, and they are very attractive to users regardless of age or cultural background, making them popular and easy to evaluate with actual players. In the fields of Artificial Intelligence and Machine Learning, games are used in a two-fold manner: to collect information about the players’ individual characteristics (player modelling), expressivity (affective computing) and playing style (adaptivity) and also to develop AI-based player bots to assist and face the human players and as a test-bed for contemporary AI algorithms. In this talk, we will discuss both approaches that relate AI/ML to games: starting from a theoretical review of user/player modelling concepts, we will discuss how we can collect data from the users during gameplay and use them to adapt the player experience or model the players themselves. Following that, we will discuss AI/ML algorithms used to train computer-based players and how these can be used in contexts outside gaming. Finally, we will discuss player modelling in contexts related to serious gaming, such as health and education. Intended audience: researchers in the fields of Machine Learning and Human-Computer Interaction, game developers and designers, and health and education practitioners. Outline: – Collecting behavioural and preference data from gameplay – Player modelling from emerging information – Player modelling from game behaviour – Clustering/classification from game behaviour – Estimating and maximising player experience – Adapting to player experience – Machine learning for non-player characters – Imitating player personalities – Learning from player behaviour – Game agents
Performance-based content generation for language learning
Generation of appropriate content for serious/educational games is an extremely important concept since it can make all the difference between adoption and retainment of the game, which increase the possibility to achieve its learning objectives and attrition. In the iRead project, we are creating a serious game and supporting applications for entry-level language learning. Content for the game is generated based on language models for each school year, learning models (sequencing of features that need to be mastered) and student performance with respect to each language feature. Several teaching and learning objectives are enforced through rules which govern content generation, such as fostering motivation and efficacy, promoting accuracy before automaticity, and allowing revision of already mastered language features. The main takeaway of the proposed lecture will be a flexible way to model language, teaching priorities and student mastery, allowing for personalized learning and student analytics. This modelling approach can be extended to different languages, as long as each of them can be modelled as a set of features, taught sequentially; based on that, personalized content selection may be used to select words, sentences or passages of text, suited for each student based on their performance and teacher-selected learning objectives.
- Emotion, games and interfaces
- AI/ML for games for AI/ML
- Performance-based content generation for language learning