Human emotion can be presented in different modes i.e., audio, video, and text. However, the contribution of each mode in exhibiting each emotion is not uniform. Furthermore, the availability of complete mode-specific details may not always be guaranteed in the test time. In this work, we propose AM^2-EmoJE, a model for Adaptive Missing-Modality Emotion Recognition in Conversation via Joint Embedding Learning model that is grounded on two-fold contributions: First, a query adaptive fusion that can automatically learn the relative importance of its mode-specific representations in a query-specific manner. By this the model aims to prioritize the mode-invariant spatial query details of the emotion patterns, while also retaining its mode-exclusive aspects within the learned multimodal query descriptor. Second the multimodal joint embedding learning module that explicitly addresses various missing modality scenarios in test-time. By this, the model learns to emphasize on the correlated patterns across modalities, which may help align the cross-attended mode-specific descriptors pairwise within a joint-embedding space and thereby compensate for missing modalities during inference. By leveraging the spatio-temporal details at the dialogue level, the proposed AM^2-EmoJE not only demonstrates superior performance compared to the best-performing state-of-the-art multimodal methods, by effectively leveraging body language in place of face expression, it also exhibits an enhanced privacy feature. By reporting around 2-5% improvement in the weighted-F1 score, the proposed multimodal joint embedding module facilitates an impressive performance gain in a variety of missing-modality query scenarios during test time.

Evaluating speaker emotion in conversations is crucial for various applications requiring human-computer interaction. However, co-occurrences of multiple emotional states (e.g. 'anger' and 'frustration' may occur together or one may influence the occurrence of the other) and their dynamic evolution may vary dramatically due to the speaker's internal (e.g., influence of their personalized socio-cultural-educational and demographic backgrounds) and external contexts. Thus far, the previous focus has been on evaluating only the dominant emotion observed in a speaker at a given time, which is susceptible to producing misleading classification decisions for difficult multi-labels during testing. In this work, we present Self-supervised Multi- Label Peer Collaborative Distillation (SeMuL-PCD) Learning via an efficient Multimodal Transformer Network, in which complementary feedback from multiple mode-specific peer networks (e.g.transcript, audio, visual) are distilled into a single mode-ensembled fusion network for estimating multiple emotions simultaneously. The proposed Multimodal Distillation Loss calibrates the fusion network by minimizing the Kullback-Leibler divergence with the peer networks. Additionally, each peer network is conditioned using a self-supervised contrastive objective to improve the generalization across diverse socio-demographic speaker backgrounds. By enabling peer collaborative learning that allows each network to independently learn their mode-specific discriminative patterns,SeMUL-PCD is effective across different conversation environments. In particular, the model not only outperforms the current state-of-the-art models on several large-scale public datasets (e.g., MOSEI, EmoReact and ElderReact), but with around 17% improved weighted F1-score in the cross-dataset experimental settings. The model also demonstrates an impressive generalization ability across age and demography-diverse populations.

Analyzing individual emotions during group conversation is crucial in developing intelligent agents capable of natural human-machine interaction. While reliable emotion recognition techniques depend on different modalities (text, audio, video), the inherent heterogeneity between these modalities and the dynamic cross-modal interactions influenced by an individual’s unique behavioral patterns make the task of emotion recognition very challenging. This difficulty is compounded in group settings, where the emotion and its temporal evolution are not only influenced by the individual but also by external contexts like audience reaction and context of the ongoing conversation. To meet this challenge, we propose a Multimodal Attention Network (MAN) that captures cross-modal interactions at various levels of spatial abstraction by jointly learning its interactive bunch of mode-specific Peripheral and Central networks. The proposed MAN "injects" cross-modal attention via its Peripheral keyvalue pairs within each layer of a mode-specific Central query network. The resulting cross-attended mode-specific descriptors are then combined using an Adaptive Fusion (AF) technique that enables the model to integrate the discriminative and complementary mode-specific data patterns within an instance-specific multimodal descriptor. Given a dialogue represented by a sequence of utterances, the proposed AMuSE (Adaptive Multimodal Analysis for Speaker Emotion) model condenses both spatial (within-mode and within-utterance) and temporal (across-mode and across-utterances in the sequence) features into two dense descriptors: speaker-level and utterance-level. This helps not only in delivering better classification performance (3−5% improvement in Weighted-F1 and 5−7% improvement in Accuracy) in large-scale public datasets (MELD and IEMOCAP) but also helps the users in understanding the reasoning behind each emotion prediction made by the model via its Multimodal Explainability Visualization module.