Maosongcao Maosongcao, Taolin Zhang, Mo Li, Chuyu Zhang, Yunxin Liu, Conghui He, Haodong Duan, Songyang Zhang, Kai Chen. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

In this paper, we address the problem of novel class discovery (NCD), which aims to cluster novel classes by leveraging knowledge from disjoint known classes. While recent advances have made significant progress in this area, existing NCD methods face two major limitations. First, they primarily focus on single-view data (e.g., images), overlooking the increasingly common multi-view data, such as multi-omics datasets used in disease diagnosis. Second, their reliance on pseudo-labels to supervise novel class clustering often results in unstable performance, as pseudo-label quality is highly sensitive to factors such as data noise and feature dimensionality. To address these challenges, we propose a novel framework named Intra-view and Inter-view Correlation Guided Multi-view Novel Class Discovery (IICMVNCD), which is the first attempt to explore NCD in multi-view setting so far. Specifically, at the intra-view level, leveraging the distributional similarity between known and novel classes, we employ matrix factorization to decompose features into view-specific shared base matrices and factor matrices. The base matrices capture distributional consistency among the two datasets, while the factor matrices model pairwise relationships between samples. At the inter-view level, we utilize view relationships among known classes to guide the clustering of novel classes. This includes generating predicted labels through the weighted fusion of factor matrices and dynamically adjusting view weights of known classes based on the supervision loss, which are then transferred to novel class learning. Experimental results validate the effectiveness of our proposed approach.

Deep clustering, which learns representation and semantic clustering without labels information, poses a great challenge for deep learning-based approaches. Despite significant progress in recent years, most existing methods focus on uniformly distributed datasets, significantly limiting the practical applicability of their methods. In this paper, we first introduce a more practical problem setting named deep imbalanced clustering, where the underlying classes exhibit an imbalance distribution. To tackle this problem, we propose a novel pseudo-labeling-based learning framework. Our framework formulates pseudo-label generation as a progressive partial optimal transport problem, which progressively transports each sample to imbalanced clusters under prior distribution constraints, thus generating imbalance-aware pseudo-labels and learning from high-confident samples. In addition, we transform the initial formulation into an unbalanced optimal transport problem with augmented constraints, which can be solved efficiently by a fast matrix scaling algorithm. Experiments on various datasets, including a human-curated long-tailed CIFAR100, challenging ImageNet-R, and large-scale subsets of fine-grained iNaturalist2018 datasets, demonstrate the superiority of our method.

We tackle the problem of novel class discovery, which aims to learn novel classes without supervision based on labeled data from known classes. A key challenge lies in transferring the knowledge in the known-class data to the learning of novel classes. Previous methods mainly focus on building a shared representation space for knowledge transfer and often ignore modeling class relations. To address this, we introduce a class relation representation for the novel classes based on the predicted class distribution of a model trained on known classes. Empirically, we find that such class relation becomes less informative during typical discovery training. To prevent such information loss, we propose a novel knowledge distillation framework, which utilizes our class-relation representation to regularize the learning of novel classes. In addition, to enable a flexible knowledge distillation scheme for each data point in novel classes, we develop a learnable weighting function for the regularization, which adaptively promotes knowledge transfer based on the semantic similarity between the novel and known classes. To validate the effectiveness and generalization of our method, we conduct extensive experiments on multiple benchmarks, including CIFAR100, Stanford Cars, CUB, and FGVC-Aircraft datasets. Our results demonstrate that the proposed method outperforms the previous state-of-the-art methods by a significant margin on almost all benchmarks. Code is available at here.

MOTIVATION: Narrative visualization for scientific data explorations can help users better understand the domain knowledge, because narrative visualizations often present a sequence of facts and observations linked together by a unifying theme or argument. Narrative visualization in immersive environments can provide users with an intuitive experience to interactively explore the scientific data, because immersive environments provide a brand new strategy for interactive scientific data visualization and exploration. However, it is challenging to develop narrative scientific visualization in immersive environments. In this paper, we propose an immersive narrative visualization tool to create and customize scientific data explorations for ordinary users with little knowledge about programming on scientific visualization, They are allowed to define POIs (point of interests) conveniently by the handler of an immersive device. RESULTS: Automatic exploration animations with narrative annotations can be generated by the gradual transitions between consecutive POI pairs. Besides, interactive slicing can be also controlled by device handler. Evaluations including user study and case study are designed and conducted to show the usability and effectiveness of the proposed tool. AVAILABILITY: Related information can be accessed at: https://dabigtou.github.io/richenliu/.

Abstract Multiuser can collaboratively complete complex visualization tasks that cannot be completed by a single user. Although multiuser collaboration system has made great progress, there are many challenges in the collaborative visualization of 3D volumetric scientific data due to the difficulties in multiuser collaboration, and collaborative slice analysis. This article proposes a client‐server based collaborative visualization system, which consists of a 3D volume explorer and a 2D slice analyzer, to help domain experts to fully utilize their background domain knowledge to illustrate different parts of the data. For example, the brain surgeon expert, pulmonologist, and cardiologist can visualize and analyze the different subsets of the volumetric scientific data, that is, the corresponding subvolumes of the brain, heart, lungs, and blood vessels. It also allows taking full advantage of the hardware resources, because all the computation intensive tasks especially for the whole data rendering can be allocated to the powerful server while the light‐weight tasks can be allocated to the portable clients. Besides, we design a seed point tracing algorithm based on flood fill algorithm to illustrate the slice more efficiently. We evaluate the system by collecting the feedback from domain experts and the people who are unfamiliar with data computation or visualization. The evaluation shows that the 3D volume explorer and the 2D slice analyzer are capable of supporting peer‐expert discussion and medical case teaching, respectively.