Few-Shot Video Recognition via Hierarchical Metric Learning
本文针对少样本视频识别问题，提出了一种层次度量学习方法HML-FSAR，通过增强空间模块和多层次度量策略来优化特征紧凑性和类别区分性。
0 citationsRead paper
Selected work
Representative Papers
DiEC: Diffusion Embedded Clustering
Existing deep clustering methods rely on a single encoder to produce static embeddings, overlooking the discriminative clustering potential embedded in the dynamic representation trajectories of pretrained diffusion models—specifically, their layer-wise and noise-step-wise internal activations. This paper proposes DiEC, the first method to directly leverage the two-dimensional (layer–time) internal activations of diffusion models for unsupervised clustering. DiEC employs weakly coupled decomposition and a two-stage search (CML + OTS) to identify clustering-favorable bottleneck layers and optimal denoising timesteps. To enhance representation separability and structural robustness, it introduces denoising consistency regularization, adaptive graph regularization, and entropy regularization. Evaluated on multiple benchmarks, DiEC achieves state-of-the-art or leading performance, significantly outperforming conventional clustering paradigms based on fixed embeddings.
0 citationsRead paper
DCSCR: A Class-Specific Collaborative Representation based Network for Image Set Classification
Image set classification (ISC) faces two key challenges: insufficient feature representation learning and non-adaptive inter-set similarity measurement—particularly limiting performance in few-shot settings. To address these, we propose a novel few-shot ISC framework that synergistically integrates deep learning with classical collaborative representation. Our approach is the first to embed class-specific collaborative representation (CSCR) into an end-to-end deep network, enabling adaptive concept-level feature refinement and joint optimization of inter-set distances. The framework comprises three core components: a fully convolutional feature extractor, a global feature aggregator, and a CSCR-based metric learning module, augmented by a newly designed CSCR contrastive loss. Extensive experiments on multiple mainstream few-shot ISC benchmarks demonstrate significant improvements over state-of-the-art methods, validating both effectiveness and robustness.
0 citationsRead paper
Conformal Sets in Multiple-Choice Question Answering under Black-Box Settings with Provable Coverage Guarantees
Large language models (LLMs) frequently exhibit hallucination and overconfidence in multiple-choice question answering (MCQA), undermining their reliability. Method: We propose a black-box, model-agnostic uncertainty quantification method grounded in answer-frequency statistics—replacing logit-based probabilities with empirical frequencies from multiple independent samplings to construct a distribution-free conformal prediction framework. The prediction set is derived from the entropy of the most frequent answer, ensuring theoretically guaranteed coverage. Results: Evaluated across six mainstream LLMs and four MCQA benchmarks, our approach significantly outperforms logit-based baselines, achieving substantial AUROC improvements while empirically satisfying the prescribed risk level for miscoverage. This provides a verifiable, theoretically sound solution for trustworthy reasoning in high-stakes applications.
0 citationsRead paper
Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences
In disentangled representation learning, statistical independence does not guarantee semantic irrelevance, rendering conventional independence-based methods incapable of ensuring semantic disentanglement. This work identifies this fundamental inconsistency and proposes a novel “difference-driven disentanglement” paradigm: it abandons the latent-variable independence assumption and instead explicitly models the intrinsic semantic distinctions among factors. Specifically, we design a difference encoder to capture discriminative semantic features across factors and introduce a cross-dimensional contrastive loss to achieve explicit, semantic-level disentanglement in a fully unsupervised manner. Evaluated on dSprites and 3DShapes, our method consistently outperforms state-of-the-art disentanglement models across multiple standard metrics—including DCI, SAP, and MIG—demonstrating both the effectiveness and generalizability of semantic difference modeling for disentanglement.
0 citationsRead paper
Recent publications
Latest Papers
Few-Shot Video Recognition via Hierarchical Metric Learning
本文针对少样本视频识别问题，提出了一种层次度量学习方法HML-FSAR，通过增强空间模块和多层次度量策略来优化特征紧凑性和类别区分性。
0 citationsRead paper
DiEC: Diffusion Embedded Clustering
Existing deep clustering methods rely on a single encoder to produce static embeddings, overlooking the discriminative clustering potential embedded in the dynamic representation trajectories of pretrained diffusion models—specifically, their layer-wise and noise-step-wise internal activations. This paper proposes DiEC, the first method to directly leverage the two-dimensional (layer–time) internal activations of diffusion models for unsupervised clustering. DiEC employs weakly coupled decomposition and a two-stage search (CML + OTS) to identify clustering-favorable bottleneck layers and optimal denoising timesteps. To enhance representation separability and structural robustness, it introduces denoising consistency regularization, adaptive graph regularization, and entropy regularization. Evaluated on multiple benchmarks, DiEC achieves state-of-the-art or leading performance, significantly outperforming conventional clustering paradigms based on fixed embeddings.
0 citationsRead paper
DCSCR: A Class-Specific Collaborative Representation based Network for Image Set Classification
Image set classification (ISC) faces two key challenges: insufficient feature representation learning and non-adaptive inter-set similarity measurement—particularly limiting performance in few-shot settings. To address these, we propose a novel few-shot ISC framework that synergistically integrates deep learning with classical collaborative representation. Our approach is the first to embed class-specific collaborative representation (CSCR) into an end-to-end deep network, enabling adaptive concept-level feature refinement and joint optimization of inter-set distances. The framework comprises three core components: a fully convolutional feature extractor, a global feature aggregator, and a CSCR-based metric learning module, augmented by a newly designed CSCR contrastive loss. Extensive experiments on multiple mainstream few-shot ISC benchmarks demonstrate significant improvements over state-of-the-art methods, validating both effectiveness and robustness.
0 citationsRead paper
Conformal Sets in Multiple-Choice Question Answering under Black-Box Settings with Provable Coverage Guarantees
Large language models (LLMs) frequently exhibit hallucination and overconfidence in multiple-choice question answering (MCQA), undermining their reliability. Method: We propose a black-box, model-agnostic uncertainty quantification method grounded in answer-frequency statistics—replacing logit-based probabilities with empirical frequencies from multiple independent samplings to construct a distribution-free conformal prediction framework. The prediction set is derived from the entropy of the most frequent answer, ensuring theoretically guaranteed coverage. Results: Evaluated across six mainstream LLMs and four MCQA benchmarks, our approach significantly outperforms logit-based baselines, achieving substantial AUROC improvements while empirically satisfying the prescribed risk level for miscoverage. This provides a verifiable, theoretically sound solution for trustworthy reasoning in high-stakes applications.
0 citationsRead paper
Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences
In disentangled representation learning, statistical independence does not guarantee semantic irrelevance, rendering conventional independence-based methods incapable of ensuring semantic disentanglement. This work identifies this fundamental inconsistency and proposes a novel “difference-driven disentanglement” paradigm: it abandons the latent-variable independence assumption and instead explicitly models the intrinsic semantic distinctions among factors. Specifically, we design a difference encoder to capture discriminative semantic features across factors and introduce a cross-dimensional contrastive loss to achieve explicit, semantic-level disentanglement in a fully unsupervised manner. Evaluated on dSprites and 3DShapes, our method consistently outperforms state-of-the-art disentanglement models across multiple standard metrics—including DCI, SAP, and MIG—demonstrating both the effectiveness and generalizability of semantic difference modeling for disentanglement.
0 citationsRead paper