The paper introduces a divergence-based similarity function (DSF) for multi-view contrastive learning, representing each set of augmented views as a distribution and measuring similarity via distribution divergence. DSF captures joint structure across all views, outperforming prior pairwise methods on tasks such as kNN classification, linear evaluation, transfer learning, and distribution shift. It also offers greater efficiency and eliminates the need for a temperature hyperparameter, unlike cosine similarity.
arXiv:2512. 02653v2 Announce Type: replace Abstract: Multi-view learning integrates diverse representations of the same instances and can improve performance when interactions across views are effectively exploited.
arXiv:2609.24125v1 Announce Type: new Abstract: Contrastive self-supervised learning has achieved strong performance by learning representations from multiple augmented views of the same image. Howev...
The paper introduces Anchor Divergences, a technique that leverages contrastive learning, exponential families, and information geometry to define context‑specific semantic geometries on fixed vector representations. By establishing a correspondence between probability distributions over anchors and Bregman geometries, the method allows the geometry itself to be modeled through the anchor distribution. Experiments on retrieval tasks demonstrate that Anchor Divergences effectively and efficiently capture context‑dependent semantic similarity.
arXiv:2605.07786v4 Announce Type: replace-cross Abstract: As generative models achieve unprecedented visual quality, the gold standard for image evaluation remains traditional feature-distribution me...
The paper investigates whether supervised contrastive learning objectives produce a regular simplex geometry in representation space when trained on long‑tailed data. It shows that existing objectives fail to achieve this geometry because they depend on class sample counts. The authors propose a fully balanced contrastive objective that provably forms a regular simplex, leading to a more balanced representation space and improved performance on long‑tailed recognition and transfer learning tasks.
The paper introduces a new framework for unsupervised visible‑infrared person re‑identification that leverages modality‑unified prototypes. By contrasting with prototypes that unify both modalities, the method jointly optimizes similarity within and across modalities, improving modality invariance. A self‑distillation step refines instance‑prototype relationships using a steady teacher, resulting in a simple yet effective model validated on standard VI‑ReID benchmarks.
arXiv:2504. 18455v2 Announce Type: replace-cross Abstract: We study distributed multiview representation learning, a problem in which $K$ clients each observe a distinct but possibly statistically correlated view.
arXiv:2608. 18339v1 Announce Type: cross Abstract: Vision-language models (VLMs) have demonstrated remarkable zero-shot capabilities yet remain sensitive to real-world distribution shifts during inference.
The paper introduces a method for selecting the best vision‑language model for a downstream task by analyzing the internal dynamics of the visual encoder. It represents each task with layer‑wise conductance and uses an entropy‑regularized alignment to derive a target‑conditioned block importance distribution. The proposed Directional Conductance Divergence (DCD) metric captures asymmetric transferability, enabling accurate prediction of model rankings without direct inference, and achieves a 14.7% NDCG@5 improvement over SWAB on 48 VLMs across 21 datasets.
arXiv:2607. 23149v1 Announce Type: new Abstract: Random Vector Functional Link (RVFL) networks provide an efficient randomized learning framework for classification.
arXiv:2603.02767v4 Announce Type: replace-cross Abstract: Image--text contrastive pretraining has become a dominant paradigm for visual representation learning, yet existing methods often yield repre...
arXiv:2606. 22220v2 Announce Type: replace-cross Abstract: Memorization in machine learning models enables high performance on rare in-distribution samples by capturing their atypical patterns.