Workshop at ACM MM’2017: Multimodal Understanding of Social, Affective and Subjective Attributes
Xavier Alameda-Pineda, Miriam Redi, Mohammad Soleymani, Nicu Sebe, Shih-Fu Chang and Samuel D. Gosling
Traditionally, the recognition of tangible properties of data, such as objects and scenes, have overwhelmingly covered the spectra of applications in multimedia, computer vision and signal processing. In the recent past and partly fostered by social media, the understanding of social, affective and subjective attributes of data has attracted the attention of many research teams at the crossroads of computer vision, multimedia, and social sciences. These attributes include the ones assessed by individuals (e.g. safety, interestingness, evoked emotions, memorability) as well as aggregated emergent properties (such as popularity or virality).
The ACM MM’17 MUSA2 workshop aims to gather high-quality contributions on the latest methodologies for understanding and recognizing intangible properties of multimodal data.
In a nutshell, the focus of the workshop is on computational and experimental methods to learn, infer, or retrieve SA from multimodal data and their applications (e.g. SA-based advertising, retrieval and search), as well as to understand how and why humans perceive SA. More specifically, the topics of the special session include:
- Data collection/annotation and evaluation methods for SA studies, including active learning and crowdsourcing.
- Learning and inference techniques for individual SA recognition in multimedia data, including beauty, sentiment, interestingness, memorability, creativity, ambiance.
- Learning and inference techniques for aggregated SA detection in multimedia data, including virality, popularity, engagement.
- User diversity-aware models for individual and collective SA detection and retrieval. Systematic studies regarding the impact of user’s demographics (e.g., gender, age, race) and psychological characteristics (e.g., personality, emotional state) in relation to subjective preferences.
- Systematic analysis on the effect of the individual’s social network on the perception of SA.
- Methods using SA to structure data, for instance for visualization or media selection
- Experimental Psychology and Computational Social Science studies using multimedia data to understand human behavior.