Im Rahmen dieses Forschungskolloquiums werden aktuelle Forschungsarbeiten im Bereich Interaktion vorgestellt.
Fragen zum Kolloquium richten Sie bitte an Julia Gordalla.
Mittwoch, den 19.07.2017 (15:30 Uhr in Raum 301, G29)
Breaking the Billion Variable Barrier in Optimization
Prof. Kalyanmoy Deb, Koenig Endowed Chair Professor, Department of Electrical and Computer Engineering, Michigan State University, East Lansing, USA
Optimization methods and practices are around for more than 50 years, but they are still criticized for their "curse of dimensionality". In this talk, we shall look at a specific large-dimensional integer-valued resource allocation problem from practice and review the performance of well-known softwares, such as IBM's CPLEX, on the problem. Thereafter, we shall present a population-based heuristic search algorithm that has the ability to recombine short-sized building blocks, despite having overlapping variable linkage, to form larger-sized building blocks. The process is eventually able to solve a billion-variable version of the problem to near-optimality in a polynomial computational time, making the application one of the largest size optimization problems ever solved.
Bio-sketch of the Speaker:
Kalyanmoy Deb is Koenig Endowed Chair Professor at Department of Electrical and Computer Engineering in Michigan State University, USA. Prof. Deb's research interests are in evolutionary optimization and their application in multi-criterion optimization, modeling, and machine learning. He has worked at various universities across the world including IITs in India, University of Dortmund and Karlsruhe Institute of Technology in Germany, Aalto University in Finland, University of Skovde in Sweden, Nanyang Technological University in Singapore. He was awarded Infosys Prize, TWAS Prize in Engineering Sciences, CajAstur Mamdani Prize, Distinguished Alumni Award from IIT Kharagpur, Edgeworth-Pareto award, Bhatnagar Prize in Engineering Sciences, and Bessel Research award from Germany. He has been just awarded IEEE CIS's "EC Pioneer Award". He is fellow of IEEE, ASME, and three Indian science and engineering academies. He has published over 475 research papers with Google Scholar citation of over 100,000 with h-index 102. He is in the editorial board on 20 major international journals. More information about his research contribution can be found from http://www.egr.msu.edu/~kdeb.
31.01.2017 (17:00 Uhr in Raum 301)
Rolling Horizon Evolutionary Methods for General Video Game Playing
Dr. Diego Perez Liebana, University of Essex, UK
Games have been used as benchmarks for research in Computational Intelligence for several decades, typically applying state of the art techniques to specific games in order to achieve a high quality of play. Breakthroughs like combining Monte Carlo Tree Search and Deep Learning in the game of Go last year demonstrate the efficiency and popularity of this methodology. However, there's another trend in recent years that tries to push our knowledge and understanding of Artificial General Intelligence, and one of the means to achieve this is tackling the problem of General (Video) Game Playing. MCTS techniques have generally dominated this problem, but recent research has started looking at Evolutionary Algorithms and their potential at matching tree search level of play or even outperforming these methods. This talk will first introduce the problem of General Video Game Playing, via our General Video Game AI framework and competition, to later describe Rolling Horizon Evolution, a technique that has been proven useful for creating agents that are able to play any game is given to it. It will also describe recent works and improvements on this technique, to finally discuss promising ideas for improving these methods in the future.
Diego is a Lecturer in Computer Games and Artificial Intelligence at the University of Essex (Colchester, UK). He achieved a PhD in Computer Science from the same institution (2015) and a MSc degree in Computer Science from University Carlos III (Madrid, Spain; 2007). His research is centred in the application of Artificial Intelligence to games, Reinforcement Learning and Evolutionary Computation. At the moment, Diego is especially interested on General Video Game Playing, which involves the creation of content and agents that play any real-time game that is given to it. He has published in the field of Game AI, in the main conferences and journals of the field of Computational Intelligence in Games and Evolutionary Computation and has also organised international competitions for the Game AI research community, such as the Physical Travelling Salesman Competition, and the General Video Game AI Competition, held in IEEE (WCCI, CIG) and ACM (GECCO) International Conferences. Diego has experience in the videogames industry as a game programmer (Revistronic; Madrid, Spain), with titles published for both PC and consoles and worked as a software engineer (Game Brains; Dublin, Ireland), where he was in charge of developing AI tools that can be applied to the latest industry videogames.
23.11.2016 (14:00 Uhr in Raum 301)
Software-Defined Multicast for Over-the-Top and Overlay-based Live Streaming in ISP Networks
Jun.-Prof. Dr. David Hausheer, TU Darmstadt
The scalable and efficient support of over-the-top (OTT) applications such as video streaming poses a variety of challenges to today's Internet Service Providers (ISPs). Software-defined networking (SDN) is a novel concept that can help ISPs to deal with these challenges. In this talk, we will present Software-Defined Multicast (SDM), an SDN-based cross-layer approach enabling ISPs to offer network layer multicast support for OTT and overlay-based live streaming as a service. SDM is specifically tailored towards the needs of P2P-based video stream delivery originating from outside the ISP network and can easily be integrated with existing streaming systems. Prototypical evaluations show significantly improved network layer transmission efficiencies when compared to other overlay streaming mechanisms, down to a level as low as for IP multicast, at linearly bounded costs.
15.09.2016 (17:00 Uhr in Raum 307)
Ground Truth Bias in External Cluster Validity Indices
Prof. Jim Bezdeck
This talk begins with a short review of clustering that emphasizes external cluster validity indices (CVIs). A method for generalizing external pair-based CVIS (e.g., the crisp Rand and Jacard indices) to evaluate soft partitions is described and illustrated. Three types of validation experiments conducted with synthetic and real world labeled data are discussed: "best c" (internal validation with labeled data), and "best I/E" (agreement between an internal and external CVI pair). As is always the case in cluster validity, conclusions based on empirical evidence are at the mercy of the data, so the reported results might be invalid for different data sets and/or clustering models and algorithms. But much more importantly, we discovered during these tests that some external cluster validity indices are also at the mercy of the distribution of the ground truth itself. We believe that our study of this surprising fact is the first systematic analysis of a largely unknown but very important problem ~ bias due to the distribution of the ground truth partition. Specifically, in addition to the well known bias in many external CVIs caused by monotonic dependency on c, the number of clusters in candidate partitions, there are two additional kinds of bias that can be caused by an unusual distribution of the clusters in the ground truth partition provided with labeled data. The most important ground truth bias is caused by imbalance (unequally sized labeled subsets). We demonstrate these effects with randomized experiments on 25 pair-based external CVIs. Then we provide a theoretical analysis of bias due to ground truth for several CV is by relating Rand's index to the Havrda-Charvat quadratic entropy.
Jim received the PhD in Applied Mathematics from Cornell University in 1973. Jim is past president of NAFIPS (North American Fuzzy Information Processing Society), IFSA (International Fuzzy Systems Association) and the IEEE CIS (Computational Intelligence Society as the NNC): founding editor the Int'l. Jo. Approximate Reasoning and the IEEE Transactions on Fuzzy Systems: Life fellow of the IEEE and IFSA; and a recipient of the IEEE 3rd Millennium, CIS Fuzzy Systems Pioneer, and technical field award Rosenblatt medals, and the IPMU Kempe de Feret Award. Jim retired in 2007, and will be coming to a university near you soon. Jim's interests: woodworking, optimization, motorcycles, pattern recognition, cigars, clustering in big data, fishing, co-clustering, blues music, wireless sensor networks, gardening, cluster validity, poker and visual clustering.
28.06.2016 (17:00 Uhr in Raum 301)
flora robotica - wie Roboter und Pflanzen zum Bio-Hybrid verschmelzen
Prof. Dr. Heiko Hamann, Universität Paderborn
Im EU Horizon 2020 Projekt “flora robotica” untersuchen wir die Vorteile der Kombination eines verteilten Robotersystems und Pflanzen. Potentielle Anwendungsmöglichkeiten reichen vom modernen Gewächshaus bis zur mechanischen Unkrautbekämpfung. Wir konzentrieren uns aber darauf, architektonische Artefakte in Form eines bio-hybriden Systems wachsen zu lassen. Roboter und Pflanzen sollen eine Einheit bilden, Synergien nutzen und tendenziell miteinander verschmelzen. Dieses System zeichnet sich durch gleichberechtigte Rollen zwischen Pflanze, Roboter und Mensch aus. Zeitweise sollen die Roboter Wachstum und Bewegung der Pflanzen steuern, es soll aber auch einen Informationsfluss von den Pflanzen an die Roboter geben. Zusätzlich interagiert der Mensch mit dem flora robotica System. Ich präsentiere unsere ersten Ergebnisse zur Steuerung von Pflanzen durch Roboter, die dafür entwickelte Hardware und unsere Vision.
07.06.2016 (16:00 in Raum 301)
Schwarmintelligenz - von Bienen und Algorithmen
Prof. Dr. habil. Martin Middendorf (Universität Leipzig)
Die Schwarmintelligenz hat sich in den letzten Jahren zu einem sehr vielfältigen und interdisziplinären Forschungsgebiet entwickelt. Im Vortrag stellen wir ausgewählte neuere Arbeiten aus unserer Forschungsgruppe zu diesem Thema vor. Insbesondere geht es um grundlegende algorithmische Ansätze zur Optimierung mit Schwarmverfahren, um Methoden zur Visualisierung des Verhaltens von Schwarmverfahren sowie um Kommunikationsnetzwerke bei Honigbienen.