VideoAgent: A Memory-augmented Multimodal Agent for Video Understanding

We explore how reconciling several foundation models (large language models and vision-language models) with a novel unified memory mechanism could tackle the challenging video understanding problem, especially capturing the long-term temporal relations in lengthy videos. In particular, the proposed multimodal agent VideoAgent: 1) constructs a structured memory to store both the generic temporal event descriptions and object-centric tracking states of the video; 2) given an input task query, it employs tools including video segment localization and object memory querying along with other visual foundation models to interactively solve the task, utilizing the zero-shot tool-use ability of LLMs. VideoAgent demonstrates impressive performances on several long-horizon video understanding benchmarks, an average increase of 6.6% on NExT-QA and 26.0% on EgoSchema over baselines, closing the gap between open-sourced models and private counterparts including Gemini 1.5 Pro.

Towards Principled Assessment of Tabular Data Synthesis Algorithms

Data synthesis has been advocated as an important approach for utilizing data while protecting data privacy. A large number of tabular data synthesis algorithms (which we call synthesizers) have been proposed. Some synthesizers satisfy Differential Privacy, while others aim to provide privacy in a heuristic fashion. A comprehensive understanding of the strengths and weaknesses of these synthesizers remains elusive due to lacking principled evaluation metrics and missing head-to-head comparisons of newly developed synthesizers that take advantage of diffusion models and large language models with state-of-the-art marginal-based synthesizers. In this paper, we present a principled and systematic evaluation framework for assessing tabular data synthesis algorithms. Specifically, we examine and critique existing evaluation metrics, and introduce a set of new metrics in terms of fidelity, privacy, and utility to address their limitations. Based on the proposed metrics, we also devise a unified objective for tuning, which can consistently improve the quality of synthetic data for all methods. We conducted extensive evaluations of 8 different types of synthesizers on 12 datasets and identified some interesting findings, which offer new directions for privacy-preserving data synthesis.

Parameter-Efficient Fine-Tuning for Pre-Trained Vision Models: A Survey

Large-scale pre-trained vision models (PVMs) have shown great potential for adaptability across various downstream vision tasks. However, with state-of-the-art PVMs growing to billions or even trillions of parameters, the standard full fine-tuning paradigm is becoming unsustainable due to high computational and storage demands. In response, researchers are exploring parameter-efficient fine-tuning (PEFT), which seeks to exceed the performance of full fine-tuning with minimal parameter modifications. This survey provides a comprehensive overview and future directions for visual PEFT, offering a systematic review of the latest advancements. First, we provide a formal definition of PEFT and discuss model pre-training methods. We then categorize existing methods into three categories: addition-based, partial-based, and unified-based. Finally, we introduce the commonly used datasets and applications and suggest potential future research challenges. A comprehensive collection of resources is available at https://github.com/synbol/Awesome-Parameter-Efficient-Transfer-Learning.

CLOVA: A Closed-Loop Visual Assistant with Tool Usage and Update

Leveraging large language models (LLMs) to integrate off-the-shelf tools (e.g., visual models and image processing functions) is a promising research direction to build powerful visual assistants for solving diverse visual tasks. However, the learning capability is rarely explored in existing methods, as they freeze the used tools after deployment, thereby limiting the generalization to new environments requiring specific knowledge. In this paper, we propose CLOVA, a Closed-LOop Visual Assistant to address this limitation, which encompasses inference, reflection, and learning phases in a closed-loop framework. During inference, LLMs generate programs and execute corresponding tools to accomplish given tasks. The reflection phase introduces a multimodal global-local reflection scheme to analyze whether and which tool needs to be updated based on environmental feedback. Lastly, the learning phase uses three flexible manners to collect training data in real-time and introduces a novel prompt tuning scheme to update the tools, enabling CLOVA to efficiently learn specific knowledge for new environments without human involvement. Experiments show that CLOVA outperforms tool-usage methods by 5% in visual question answering and multiple-image reasoning tasks, by 10% in knowledge tagging tasks, and by 20% in image editing tasks, highlighting the significance of the learning capability for general visual assistants.

LaplaceConfidence: a Graph-based Approach for Learning with Noisy Labels

In real-world applications, perfect labels are rarely available, making it challenging to develop robust machine learning algorithms that can handle noisy labels. Recent methods have focused on filtering noise based on the discrepancy between model predictions and given noisy labels, assuming that samples with small classification losses are clean. This work takes a different approach by leveraging the consistency between the learned model and the entire noisy dataset using the rich representational and topological information in the data. We introduce LaplaceConfidence, a method that to obtain label confidence (i.e., clean probabilities) utilizing the Laplacian energy. Specifically, it first constructs graphs based on the feature representations of all noisy samples and minimizes the Laplacian energy to produce a low-energy graph. Clean labels should fit well into the low-energy graph while noisy ones should not, allowing our method to determine data's clean probabilities. Furthermore, LaplaceConfidence is embedded into a holistic method for robust training, where co-training technique generates unbiased label confidence and label refurbishment technique better utilizes it. We also explore the dimensionality reduction technique to accommodate our method on large-scale noisy datasets. Our experiments demonstrate that LaplaceConfidence outperforms state-of-the-art methods on benchmark datasets under both synthetic and real-world noise.

Knowledge-refined Denoising Network for Robust Recommendation

Knowledge graph (KG), which contains rich side information, becomes an essential part to boost the recommendation performance and improve its explainability. However, existing knowledge-aware recommendation methods directly perform information propagation on KG and user-item bipartite graph, ignoring the impacts of \textit{task-irrelevant knowledge propagation} and \textit{vulnerability to interaction noise}, which limits their performance. To solve these issues, we propose a robust knowledge-aware recommendation framework, called \textit{Knowledge-refined Denoising Network} (KRDN), to prune the task-irrelevant knowledge associations and noisy implicit feedback simultaneously. KRDN consists of an adaptive knowledge refining strategy and a contrastive denoising mechanism, which are able to automatically distill high-quality KG triplets for aggregation and prune noisy implicit feedback respectively. Besides, we also design the self-adapted loss function and the gradient estimator for model optimization. The experimental results on three benchmark datasets demonstrate the effectiveness and robustness of KRDN over the state-of-the-art knowledge-aware methods like KGIN, MCCLK, and KGCL, and also outperform robust recommendation models like SGL and SimGCL.

Towards Explainable Collaborative Filtering with Taste Clusters Learning

Collaborative Filtering (CF) is a widely used and effective technique for recommender systems. In recent decades, there have been significant advancements in latent embedding-based CF methods for improved accuracy, such as matrix factorization, neural collaborative filtering, and LightGCN. However, the explainability of these models has not been fully explored. Adding explainability to recommendation models can not only increase trust in the decisionmaking process, but also have multiple benefits such as providing persuasive explanations for item recommendations, creating explicit profiles for users and items, and assisting item producers in design improvements. In this paper, we propose a neat and effective Explainable Collaborative Filtering (ECF) model that leverages interpretable cluster learning to achieve the two most demanding objectives: (1) Precise - the model should not compromise accuracy in the pursuit of explainability; and (2) Self-explainable - the model's explanations should truly reflect its decision-making process, not generated from post-hoc methods. The core of ECF is mining taste clusters from user-item interactions and item profiles.We map each user and item to a sparse set of taste clusters, and taste clusters are distinguished by a few representative tags. The user-item preference, users/items' cluster affiliations, and the generation of taste clusters are jointly optimized in an end-to-end manner. Additionally, we introduce a forest mechanism to ensure the model's accuracy, explainability, and diversity. To comprehensively evaluate the explainability quality of taste clusters, we design several quantitative metrics, including in-cluster item coverage, tag utilization, silhouette, and informativeness. Our model's effectiveness is demonstrated through extensive experiments on three real-world datasets.

Spatial-Temporal Graph Convolutional Gated Recurrent Network for Traffic Forecasting

As an important part of intelligent transportation systems, traffic forecasting has attracted tremendous attention from academia and industry. Despite a lot of methods being proposed for traffic forecasting, it is still difficult to model complex spatial-temporal dependency. Temporal dependency includes short-term dependency and long-term dependency, and the latter is often overlooked. Spatial dependency can be divided into two parts: distance-based spatial dependency and hidden spatial dependency. To model complex spatial-temporal dependency, we propose a novel framework for traffic forecasting, named Spatial-Temporal Graph Convolutional Gated Recurrent Network (STGCGRN). We design an attention module to capture long-term dependency by mining periodic information in traffic data. We propose a Double Graph Convolution Gated Recurrent Unit (DGCGRU) to capture spatial dependency, which integrates graph convolutional network and GRU. The graph convolution part models distance-based spatial dependency with the distance-based predefined adjacency matrix and hidden spatial dependency with the self-adaptive adjacency matrix, respectively. Specially, we employ the multi-head mechanism to capture multiple hidden dependencies. In addition, the periodic pattern of each prediction node may be different, which is often ignored, resulting in mutual interference of periodic information among nodes when modeling spatial dependency. For this, we explore the architecture of model and improve the performance. Experiments on four datasets demonstrate the superior performance of our model.

Self-Guided Learning to Denoise for Robust Recommendation

The ubiquity of implicit feedback makes them the default choice to build modern recommender systems. Generally speaking, observed interactions are considered as positive samples, while unobserved interactions are considered as negative ones. However, implicit feedback is inherently noisy because of the ubiquitous presence of noisy-positive and noisy-negative interactions. Recently, some studies have noticed the importance of denoising implicit feedback for recommendations, and enhanced the robustness of recommendation models to some extent. Nonetheless, they typically fail to (1) capture the hard yet clean interactions for learning comprehensive user preference, and (2) provide a universal denoising solution that can be applied to various kinds of recommendation models. In this paper, we thoroughly investigate the memorization effect of recommendation models, and propose a new denoising paradigm, i.e., Self-Guided Denoising Learning (SGDL), which is able to collect memorized interactions at the early stage of the training (i.e., "noise-resistant" period), and leverage those data as denoising signals to guide the following training (i.e., "noise-sensitive" period) of the model in a meta-learning manner. Besides, our method can automatically switch its learning phase at the memorization point from memorization to self-guided learning, and select clean and informative memorized data via a novel adaptive denoising scheduler to improve the robustness. We incorporate SGDL with four representative recommendation models (i.e., NeuMF, CDAE, NGCF and LightGCN) and different loss functions (i.e., binary cross-entropy and BPR loss). The experimental results on three benchmark datasets demonstrate the effectiveness of SGDL over the state-of-the-art denoising methods like T-CE, IR, DeCA, and even state-of-the-art robust graph-based methods like SGCN and SGL.

HAKG: Hierarchy-Aware Knowledge Gated Network for Recommendation

Knowledge graph (KG) plays an increasingly important role to improve the recommendation performance and interpretability. A recent technical trend is to design end-to-end models based on information propagation schemes. However, existing propagation-based methods fail to (1) model the underlying hierarchical structures and relations, and (2) capture the high-order collaborative signals of items for learning high-quality user and item representations. In this paper, we propose a new model, called Hierarchy-Aware Knowledge Gated Network (HAKG), to tackle the aforementioned problems. Technically, we model users and items (that are captured by a user-item graph), as well as entities and relations (that are captured in a KG) in hyperbolic space, and design a hyperbolic aggregation scheme to gather relational contexts over KG. Meanwhile, we introduce a novel angle constraint to preserve characteristics of items in the embedding space. Furthermore, we propose a dual item embeddings design to represent and propagate collaborative signals and knowledge associations separately, and leverage the gated aggregation to distill discriminative information for better capturing user behavior patterns. Experimental results on three benchmark datasets show that, HAKG achieves significant improvement over the state-of-the-art methods like CKAN, Hyper-Know, and KGIN. Further analyses on the learned hyperbolic embeddings confirm that HAKG offers meaningful insights into the hierarchies of data.