Title : MONEE Talks
| Presenter | Evert Haasdijk |
| Abstract | In our vision, evolution serves two purposes: on the one hand to allow robots to adapt to the environment and to behave so that they can operate at all. On the other hand, evolution is a force to promote task-performance, where we interpret `task' in a broad sense: it is any user-defined preference with a measurable level of compliance. It can be a direct task, like collecting rubbish (measured by the amount of rubbish cleared), but also more indirect, like energy efficiency (measured by battery lifetime). Combining these two (seemingly) contradictory roles of evolution is a generic, fundamental challenge that to our knowledge has to date not been tackled successfully. I will discuss our ideas and some early experimental research with our novel MONEE (Multi-Objective aNd open-Ended Evolution) algorithm that combines the open-ended and task-driven aspects of evolution. |
Title : Improving Video Search through Serious-game Data
| Presenter | Riste Gligorov |
| Abstract | Serious games are increasingly used in audio-visual collections as a mechanism for annotating videos through tagging. This trend is driven by the assumption that user tags will improve video search. In this paper we study whether this is indeed the case. To this end, we create an evaluation dataset that consists of: (i) a set of videos tagged by users via video labelling game, (ii) a set of queries derived from real-life query logs, and (iii) relevance judgements. Besides user tags from the labelling game, we exploit the existing metadata associated with the videos (textual descriptions and curated in-house tags) and closed captions. Our findings show that search based on user tags alone outperforms search based on all other metadata types. Combining user tags with the other types of metadatayields an increase in search performance of 33%. We also find that search performance of user tags steadily increases as more tags are collected. |