Gamma belief network (GBN), widely viewed as deep probabilistic topic models, has demonstrated its potential for uncovering multi-layer interpretable latent representations from text corpora. Its notable performance in document modeling largely arises from the expressive nature of gamma-distributed latent variables, which naturally capture sparsity, nonnegativity, skewness, heavy-tailed pattens, and from their seamless extension to multi-layer hierarchical structures. However, existing GBN and its variations are constrained by linear generative model, thereby limiting their expressiveness and applicability. To address this limitation, we introduce Generalized Gamma Belief Network (Generalized GBN), which extends original linear generative model to a more expressive non-linear generative model. Since parameters of Generalized GBN no longer possess an analytic conditional posterior, we further propose an upward-downward Weibull inference network to approximate posterior distribution of latent variables. The parameters of both generative model and inference network are jointly trained within variational inference framework. In addition, we provide theoretical analyses that demonstrate the effectiveness of Generalized GBN in modeling data variability and achieving disentangled representations. The former benefit arises from its hierarchical latent-variable structure, while the latter stems from its inherent ability to model sparsity. Finally, we conduct comprehensive experiments on both expressivity and disentangled representation learning tasks to evaluate the performance of Generalized GBN against Gaussian variational autoencoders serving as strong baseline models.

Multimodal remote sensing combines optical and synthetic aperture radar (SAR) imagery to improve perception, yet real deployments face spatially varying degradations (e.g., clouds, low light, sensor interference) that can corrupt fusion. To make robustness measurable, we introduce a controlled mixed-severity setting in which only the optical stream is synthetically cloud-degraded while SAR remains intact, providing a standardized testbed for evaluating multimodal detection under modality imbalance. We further present CAIR-Net, a reliability–aware information routing network that follows adenoise-then-fuseprinciple: a Local Reliability Modulation (LRM) module learns soft, spatial reliability maps to suppress degraded regionsbeforecross-modal interaction, and a Global Information Selection Mechanism (GISM) performs confidence-aware expert routing across optical, fused, and SAR experts. On the mixed-severity benchmark, CAIR-Net consistently outperforms strong unimodal and fusion baselines and exhibits a substantially smaller performance drop under severe clouds (only a 7.3% AP reduction versus drops exceeding 25% for representative alternatives). These results indicate that explicit reliability modeling and quality-guided routing provide a practical path toward robust multimodal detection when one modality is partially or nearly completely occluded.

Concept-based models can map black-box representations to human-understandable concepts, which makes the decision-making process more transparent and then allows users to understand the reason behind predictions. However, domain-specific concepts often impact the final predictions, which subsequently undermine the model generalization capabilities, and prevent the model from being used in high-stake applications. In this paper, we propose a novel Language-guided Concept-Erasing (LanCE) framework. In particular, we empirically demonstrate that pre-trained vision-language models (VLMs) can approximate distinct visual domain shifts via domain descriptors while prompting large Language Models (LLMs) can easily simulate a wide range of descriptors of unseen visual domains. Then, we introduce a novel plug-in domain descriptor orthogonality (DDO) regularizer to mitigate the impact of these domain-specific concepts on the final predictions. Notably, the DDO regularizer is agnostic to the design of concept-based models and we integrate it into several prevailing models. Through evaluation of domain generalization on four standard benchmarks and three newly introduced benchmarks, we demonstrate that DDO can significantly improve the out-of-distribution (OOD) generalization over the previous state-of-the-art concept-based models. Our code is available at https://github.com/joeyz0z/LanCE.

Many large-scale systems rely on high-quality deep representations (embeddings) to facilitate tasks like retrieval, search, and generative modeling. Matryoshka Representation Learning (MRL) recently emerged as a solution for adaptive embedding lengths, but it requires full model retraining and suffers from noticeable performance degradations at short lengths. In this paper, we show that sparse coding offers a compelling alternative for achieving adaptive representation with minimal overhead and higher fidelity. We propose Contrastive Sparse Representation (CSR), a method that specifies pre-trained embeddings into a high-dimensional but selectively activated feature space. By leveraging lightweight autoencoding and task-aware contrastive objectives, CSR preserves semantic quality while allowing flexible, cost-effective inference at different sparsity levels. Extensive experiments on image, text, and multimodal benchmarks demonstrate that CSR consistently outperforms MRL in terms of both accuracy and retrieval speed—often by large margins—while also cutting training time to a fraction of that required by MRL. Our results establish sparse coding as a powerful paradigm for adaptive representation learning in real-world applications where efficiency and fidelity are both paramount. Code is available at this URL.

Image captioning evaluation metrics can be divided into two categories, reference-based metrics and reference-free metrics. However, reference-based approaches may struggle to evaluate descriptive captions with abundant visual details produced by advanced multimodal large language models, due to their heavy reliance on limited human-annotated references. In contrast, previous reference-free metrics have been proven effective via CLIP cross-modality similarity. Nonetheless, CLIP-based metrics, constrained by their solution of global image-text compatibility, often have a deficiency in detecting local textual hallucinations and are insensitive to small visual objects. Besides, their single-scale designs are unable to provide an interpretable evaluation process such as pinpointing the position of caption mistakes and identifying visual regions that have not been described. To move forward, we propose a novel reference-free metric for image captioning evaluation, dubbed Hierarchical Image Captioning Evaluation Score (HICE-S). By detecting local visual regions and textual phrases, HICE-S builds an interpretable hierarchical scoring mechanism, breaking through the barriers of the single-scale structure of existing reference-free metrics. Comprehensive experiments indicate that our proposed metric achieves the SOTA performance on several benchmarks, outperforming existing reference-free metrics like CLIP-S and PAC-S, and reference-based metrics like METEOR and CIDEr. Moreover, several case studies reveal that the assessment process of HICE-S on detailed captions closely resembles interpretable human judgments.Our code is available at https://github.com/joeyz0z/HICE.