Transactions on Affective Computing
IEEE Transactions on Affective Computing (TAC) is intended to be a cross disciplinary and international archive journal aimed at disseminating results of research on the design of systems that can recognize, interpret, and simulate human emotions and related affective phenomena. Read the full scope of TAC
Expand your horizons with Colloquium, a monthly survey of abstracts from all CS transactions!
TAC Seeks Editor-in-Chief for 2019-2020 Term
TAC seeks Editor-in-Chief applicants for a two-year term starting 1 January 2019, renewable for two years. Prospective candidates are asked to provide a complete curriculum vitae, a brief plan for the publication's future, and a letter of support from their institution or employer to Jennifer Carruth, email@example.com, by 1 March 2018. Please click here for additional information.
From the October-December 2017 issue
Audio-Facial Laughter Detection in Naturalistic Dyadic Conversations
By Bekir Berker Turker, Yucel Yemez, T. Metin Sezgin, and Engin Erzin
We address the problem of continuous laughter detection over audio-facial input streams obtained from naturalistic dyadic conversations. We first present meticulous annotation of laughters, cross-talks and environmental noise in an audio-facial database with explicit 3D facial mocap data. Using this annotated database, we rigorously investigate the utility of facial information, head movement and audio features for laughter detection. We identify a set of discriminative features using mutual information-based criteria, and show how they can be used with classifiers based on support vector machines (SVMs) and time delay neural networks (TDNNs). Informed by the analysis of the individual modalities, we propose a multimodal fusion setup for laughter detection using different classifier-feature combinations. We also effectively incorporate bagging into our classification pipeline to address the class imbalance problem caused by the scarcity of positive laughter instances. Our results indicate that a combination of TDNNs and SVMs lead to superior detection performance, and bagging effectively addresses data imbalance. Our experiments show that our multimodal approach supported by bagging compares favorably to the state of the art in presence of detrimental factors such as cross-talk, environmental noise, and data imbalance.
Editorials and Announcements
- According to Clarivate Analytics' 2016 Journal Citation Report, TAC has an impact factor of 3.149.
- Heartfelt congratulations are offered to Georgios N. Yannakakis and Julian Togelius, authors of "Experience-Driven Procedural Content Generation," who were presented with TAC's Most Influential Paper Award by Editor-in-Chief Björn W. Schuller at the 2015 6th AAAC Affective Computing and Intelligent Interaction Conference in Xi'An, P.R. China on 22 September 2015.
- Editorial: IEEE Transactions on Affective Computing – Challenges and Chances (Jan-March 2017)
- Editorial: Transactions on Affective Computing – Changes and Continuance (Jan-March 2016)
- Editorial: State of the Journal (July-Sept 2014)
- Introduction to TAC by J. Gratch
- Towards Machines Able to Deal with Laughter (Oct-Dec 2017)
- Toward Commercial Applications of Affective Computing (April-June 2017)
- Best of Bodynets 2014: Editorial (July-Sept 2016)
- Challenges and Perspectives for Affective Analysis in Multimedia (July-Sept 2015)
- Introduction to the "Best of ACII 2013" Special Section (April-June 2015)
- Affect and Wellbeing: Introduction to Special Section (July-Sept 2014)
- Editorial for the Special Section on Ethics and Affective Computing
- Introduction to the Affect-Based Human Behavior Understanding Special Issue
- Affective Computing: From Laughter to IEEE by R.W. Picard
Call for Papers
Special Issue on Computational Modelling of Emotion: Theory and Applications
Submission deadline: June 9, 2018. View PDF.
In the early decades of cognitive science research, emotion was either absent or side-lined from most computational models of human behaviour. Since then interest in attempts to computationally model emotions has grown, with many projects now either attempting to understand natural emotions or to implement synthetic emotions in chatbots, virtual agents or robots, for practical uses of many sorts from entertainment to caring. Whilst there are now numerous models of affective phenomena in the literature, they differ in important respects. They differ in how they describe and explain a range of phenomena, including the nature and order of perceptual, cognitive and emotional mental processes and behavioural responses in emotional episodes. They also differ in their target level of granularity: from fine-grained neural to coarse-grained psychological. Different models simulate emotions (and other mental states) with different ontological status and with a different focus on whether they model external behaviour or internal states. This diversity provides a challenge, but also an opportunity.
This special issue aims to facilitate movement towards a mature integrated field with a deeper and richer understanding of biological minds and also design functionalities of applied models by more clearly setting out interrelationships between models and present attempts to provide formal or standard models of particular approaches within emotion modelling. For example, Marsella, Gratch and Petta (2010) focus on appraisal and dimensional models and Scherer (2010) sets out a broader taxonomic analysis including radically different kinds of emotion models, including: appraisal; adaptational; dimensional; motivational; circuit; discrete; lexical and social constructivist models. Whilst Broekens, DeGroot and Kosters (2008) provide a deeper yet narrower analysis by formalising the structure of emotional appraisal structures with a notation for the declarative semantics of these kinds of emotional states. Hudlicka (2011) shows how a broader organising approach can progress by highlighting the generation and effect of emotions as fundamental processes with associated 'generic tasks' that can lead to broad categorisations useful in creating guidelines for model development and more systematic comparison of existing models. The project for standardisation and formalisation for emotion models is taken further by Reisenzein, Hudlicka, Dastani, Gratch, Hindriks, Lorini, and Meyer (2013), who propose further standardisation; formalisation; and in addition, integration of emotion models with existing prominent and widely used cognitive architectures. Standardisation can involve benchmark scenarios and replication of results. However, benchmarks can have a negative influence on progress if they become narrow targets for model development. This kind of narrow development can be minimised by clarity regarding how the modelling is done and what theoretical or applied goals are to be achieved for a given model.
Contributions that move this debate in the literature forward by further identifying and attempting to remedy gaps in current research on affective phenomena are particularly welcome. For example, some emotion models fail to acknowledge that emotions are just a subcategory of "affect". Richer theories and models should include motives, attachments, preferences, values, standards, attitudes, moods, ambitions, obsessions, humour, grief, various kinds of pride, and various other social, complex and secondary emotions as well as moral and aesthetic phenomena. The narrow focus may not matter much for narrowly focused applications of AI, such as toys or entertainment, but it can lead to serious omissions and distortions in attempts to advance the science of mind through computational modelling.
Therefore the aims of this special issue include: presenting the state of the art in emotion modelling and considering how existing research in modelling of emotions, motivation and other varieties of affect can be integrated, validated and compared with each other as well as with possible 'standard models' of emotion. The special edition also aims to explaining how technological applications based on this broader, more standardised and formalised approach can be used to make contributions to psychological theory.
Access recently published TAC articles
Subscribe to the RSS feed of recently published TAC content
Sign up for e-mail notifications through IEEE Xplore Content Alerts
View TAC preprints in the Computer Society Digital Library
TAC is indexed in ISI