Data-Driven Decoding of Russell's Circumplex Model of Affect
Two new arXiv papers explore the geometric properties of emotion representation in AI models. The first paper demonstrates that multimodal Transformers can perfectly align with Russell's circumplex model of affect, suggesting that the model's structure is intrinsically encoded in embeddings. The second paper argues that failures in rare-class emotion recognition are due to the geometric degeneracy of these classes on the circumplex, rather than simple class imbalance, proposing that new representations are needed to distinguish these emotions. AI
IMPACT These papers suggest that while AI models can encode complex emotional structures, achieving robust recognition of rare emotions may require new representational approaches beyond current geometric or imbalance-based methods.
