Metric Learning with Progressive Self-Distillation for Audio-Visual Embedding Learning

Metric learning projects samples into an embedded space, where similarities and dissimilarities are quantified based on their learned representations. However, existing methods often rely on label-guided representation learning, where representations of different modalities, such as audio and visual data, are aligned based on annotated labels. This approach tends to underutilize latent complex features and potential relationships inherent in the distributions of audio and visual data that are not directly tied to the labels, resulting in suboptimal performance in audio-visual embedding learning. To address this issue, we propose a novel architecture that integrates cross-modal triplet loss with progressive self-distillation. Our method enhances representation learning by leveraging inherent distributions and dynamically refining soft audio-visual alignments -- probabilistic alignments between audio and visual data that capture the inherent relationships beyond explicit labels. Specifically, the model distills audio-visual distribution-based knowledge from annotated labels in a subset of each batch. This self-distilled knowledge is used t

Top-down Activity Representation Learning for Video Question Answering

Capturing complex hierarchical human activities, from atomic actions (e.g., picking up one present, moving to the sofa, unwrapping the present) to contextual events (e.g., celebrating Christmas) is crucial for achieving high-performance video question answering (VideoQA). Recent works have expanded multimodal models (e.g., CLIP, LLaVA) to process continuous video sequences, enhancing the model's temporal reasoning capabilities. However, these approaches often fail to capture contextual events that can be decomposed into multiple atomic actions non-continuously distributed over relatively long-term sequences. In this paper, to leverage the spatial visual context representation capability of the CLIP model for obtaining non-continuous visual representations in terms of contextual events in videos, we convert long-term video sequences into a spatial image domain and finetune the multimodal model LLaVA for the VideoQA task. Our approach achieves competitive performance on the STAR task, in particular, with a 78.4% accuracy score, exceeding the current state-of-the-art score by 2.8 points on the NExTQA task.

Multi-object event graph representation learning for Video Question Answering

Video question answering (VideoQA) is a task to predict the correct answer to questions posed about a given video. The system must comprehend spatial and temporal relationships among objects extracted from videos to perform causal and temporal reasoning. While prior works have focused on modeling individual object movements using transformer-based methods, they falter when capturing complex scenarios involving multiple objects (e.g., "a boy is throwing a ball in a hoop"). We propose a contrastive language event graph representation learning method called CLanG to address this limitation. Aiming to capture event representations associated with multiple objects, our method employs a multi-layer GNN-cluster module for adversarial graph representation learning, enabling contrastive learning between the question text and its relevant multi-object event graph. Our method outperforms a strong baseline, achieving up to 2.2% higher accuracy on two challenging VideoQA datasets, NExT-QA and TGIF-QA-R. In particular, it is 2.8% better than baselines in handling causal and temporal questions, highlighting its strength in reasoning multiple object-based events.

Counterfactual Reasoning Using Predicted Latent Personality Dimensions for Optimizing Persuasion Outcome

Customizing persuasive conversations related to the outcome of interest for specific users achieves better persuasion results. However, existing persuasive conversation systems rely on persuasive strategies and encounter challenges in dynamically adjusting dialogues to suit the evolving states of individual users during interactions. This limitation restricts the system's ability to deliver flexible or dynamic conversations and achieve suboptimal persuasion outcomes. In this paper, we present a novel approach that tracks a user's latent personality dimensions (LPDs) during ongoing persuasion conversation and generates tailored counterfactual utterances based on these LPDs to optimize the overall persuasion outcome. In particular, our proposed method leverages a Bi-directional Generative Adversarial Network (BiCoGAN) in tandem with a Dialogue-based Personality Prediction Regression (DPPR) model to generate counterfactual data. This enables the system to formulate alternative persuasive utterances that are more suited to the user. Subsequently, we utilize the D3QN model to learn policies for optimized selection of system utterances on counterfactual data. Experimental results we obtained from using the PersuasionForGood dataset demonstrate the superiority of our approach over the existing method, BiCoGAN. The cumulative rewards and Q-values produced by our method surpass ground truth benchmarks, showcasing the efficacy of employing counterfactual reasoning and LPDs to optimize reinforcement learning policy in online interactions.

Anchor-aware Deep Metric Learning for Audio-visual Retrieval

Metric learning minimizes the gap between similar (positive) pairs of data points and increases the separation of dissimilar (negative) pairs, aiming at capturing the underlying data structure and enhancing the performance of tasks like audio-visual cross-modal retrieval (AV-CMR). Recent works employ sampling methods to select impactful data points from the embedding space during training. However, the model training fails to fully explore the space due to the scarcity of training data points, resulting in an incomplete representation of the overall positive and negative distributions. In this paper, we propose an innovative Anchor-aware Deep Metric Learning (AADML) method to address this challenge by uncovering the underlying correlations among existing data points, which enhances the quality of the shared embedding space. Specifically, our method establishes a correlation graph-based manifold structure by considering the dependencies between each sample as the anchor and its semantically similar samples. Through dynamic weighting of the correlations within this underlying manifold structure using an attention-driven mechanism, Anchor Awareness (AA) scores are obtained for each anchor. These AA scores serve as data proxies to compute relative distances in metric learning approaches. Extensive experiments conducted on two audio-visual benchmark datasets demonstrate the effectiveness of our proposed AADML method, significantly surpassing state-of-the-art models. Furthermore, we investigate the integration of AA proxies with various metric learning methods, further highlighting the efficacy of our approach.

Two-Stage Triplet Loss Training with Curriculum Augmentation for Audio-Visual Retrieval

The cross-modal retrieval model leverages the potential of triple loss optimization to learn robust embedding spaces. However, existing methods often train these models in a singular pass, overlooking the distinction between semi-hard and hard triples in the optimization process. The oversight of not distinguishing between semi-hard and hard triples leads to suboptimal model performance. In this paper, we introduce a novel approach rooted in curriculum learning to address this problem. We propose a two-stage training paradigm that guides the model's learning process from semi-hard to hard triplets. In the first stage, the model is trained with a set of semi-hard triplets, starting from a low-loss base. Subsequently, in the second stage, we augment the embeddings using an interpolation technique. This process identifies potential hard negatives, alleviating issues arising from high-loss functions due to a scarcity of hard triples. Our approach then applies hard triplet mining in the augmented embedding space to further optimize the model. Extensive experimental results conducted on two audio-visual datasets show a significant improvement of approximately 9.8% in terms of average Mean Average Precision (MAP) over the current state-of-the-art method, MSNSCA, for the Audio-Visual Cross-Modal Retrieval (AV-CMR) task on the AVE dataset, indicating the effectiveness of our proposed method.

VideoAdviser: Video Knowledge Distillation for Multimodal Transfer Learning

Multimodal transfer learning aims to transform pretrained representations of diverse modalities into a common domain space for effective multimodal fusion. However, conventional systems are typically built on the assumption that all modalities exist, and the lack of modalities always leads to poor inference performance. Furthermore, extracting pretrained embeddings for all modalities is computationally inefficient for inference. In this work, to achieve high efficiency-performance multimodal transfer learning, we propose VideoAdviser, a video knowledge distillation method to transfer multimodal knowledge of video-enhanced prompts from a multimodal fundamental model (teacher) to a specific modal fundamental model (student). With an intuition that the best learning performance comes with professional advisers and smart students, we use a CLIP-based teacher model to provide expressive multimodal knowledge supervision signals to a RoBERTa-based student model via optimizing a step-distillation objective loss -- first step: the teacher distills multimodal knowledge of video-enhanced prompts from classification logits to a regression logit -- second step: the multimodal knowledge is distilled from the regression logit of the teacher to the student. We evaluate our method in two challenging multimodal tasks: video-level sentiment analysis (MOSI and MOSEI datasets) and audio-visual retrieval (VEGAS dataset). The student (requiring only the text modality as input) achieves an MAE score improvement of up to 12.3% for MOSI and MOSEI. Our method further enhances the state-of-the-art method by 3.4% mAP score for VEGAS without additional computations for inference. These results suggest the strengths of our method for achieving high efficiency-performance multimodal transfer learning.

Topic-switch adapted Japanese Dialogue System based on PLATO-2

Large-scale open-domain dialogue systems such as PLATO-2 have achieved state-of-the-art scores in both English and Chinese. However, little work explores whether such dialogue systems also work well in the Japanese language. In this work, we create a large-scale Japanese dialogue dataset, Dialogue-Graph, which contains 1.656 million dialogue data in a tree structure from News, TV subtitles, and Wikipedia corpus. Then, we train PLATO-2 using Dialogue-Graph to build a large-scale Japanese dialogue system, PLATO-JDS. In addition, to improve the PLATO-JDS in the topic switch issue, we introduce a topic-switch algorithm composed of a topic discriminator to switch to a new topic when user input differs from the previous topic. We evaluate the user experience by using our model with respect to four metrics, namely, coherence, informativeness, engagingness, and humanness. As a result, our proposed PLATO-JDS achieves an average score of 1.500 for the human evaluation with human-bot chat strategy, which is close to the maximum score of 2.000 and suggests the high-quality dialogue generation capability of PLATO-2 in Japanese. Furthermore, our proposed topic-switch algorithm achieves an average score of 1.767 and outperforms PLATO-JDS by 0.267, indicating its effectiveness in improving the user experience of our system.

Do I Have Your Attention: A Large Scale Engagement Prediction Dataset and Baselines

The degree of concentration, enthusiasm, optimism, and passion displayed by individual(s) while interacting with a machine is referred to as `user engagement'. Engagement comprises of behavioural, cognitive, and affect related cues. To create engagement predictions systems, which can work in real-world conditions it is quintessential to learn from rich diverse datasets. To this end, a large scale multi-faceted engagement in the wild dataset is proposed. 31 hours duration data of 127 participants representing different illumination conditions is recorded. Thorough experiments are performed exploring applicability of different features action units, eye gaze and head pose and transformers. To further validate the rich nature of the dataset, evaluation is also performed on the EngageWild dataset. The experiments show the usefulness of the proposed dataset. The code, models and dataset will be made publicly available.

Complete Cross-triplet Loss in Label Space for Audio-visual Cross-modal Retrieval

The heterogeneity gap problem is the main challenge in cross-modal retrieval. Because cross-modal data (e.g. audiovisual) have different distributions and representations that cannot be directly compared. To bridge the gap between audiovisual modalities, we learn a common subspace for them by utilizing the intrinsic correlation in the natural synchronization of audio-visual data with the aid of annotated labels. TNN-CCCA is the best audio-visual cross-modal retrieval (AV-CMR) model so far, but the model training is sensitive to hard negative samples when learning common subspace by applying triplet loss to predict the relative distance between inputs. In this paper, to reduce the interference of hard negative samples in representation learning, we propose a new AV-CMR model to optimize semantic features by directly predicting labels and then measuring the intrinsic correlation between audio-visual data using complete cross-triple loss. In particular, our model projects audio-visual features into label space by minimizing the distance between predicted label features after feature projection and ground label representations. Moreover, we adopt complete cross-triplet loss to optimize the predicted label features by leveraging the relationship between all possible similarity and dissimilarity semantic information across modalities. The extensive experimental results on two audio-visual double-checked datasets have shown an improvement of approximately 2.1% in terms of average MAP over the current state-of-the-art method TNN-CCCA for the AV-CMR task, which indicates the effectiveness of our proposed model.