contrastive-learning

This paper introduces a geometry-aware adversarial attack framework that targets relational structure in contrastive embedding manifolds, showing that verification systems like Markmatch can be severely degraded by distorting pairwise similarities rather than decision boundaries.

InsightEmb is a contrastive embedding framework for agentic insight retrieval that learns progress-oriented retrieval geometry from mathematical reasoning data alone, improving retrieval for LLM agents without environment-specific training.

This paper explores three LLM-based approaches for translating puns from English to French, combining contrastive learning and phonetic-semantic embeddings. Their multi-agent and guided chain-of-thought systems ranked first and second in the CLEF JOKER 2025 Task 2 competition under expert human evaluation.

This paper introduces Bayesian Data Reweighting, a probabilistic framework that adaptively weights query-document pairs to mitigate false negatives in contrastive training, improving multimodal retrieval across multiple knowledge-based VQA benchmarks.

This paper introduces CoCoS, a contrastive pretraining framework that learns whole-cell representations from complementary transcriptomic views, addressing limitations of masked gene reconstruction in single-cell foundation models. Experiments on cell-type annotation and gene regulatory network inference show competitive transfer performance.

The paper proposes QQ, a framework that leverages the intrinsic duality between multi-hop question generation and question answering via bidirectional alignment constraints and contrastive learning, improving question quality on HotpotQA and MuSiQue.

Proposes FedTCR, the first systematic federated multimodal graph learning algorithm that handles task, modality, and topology heterogeneity via topology-aware cross-modal routing and tri-level contrastive learning, outperforming baselines across 7 domains.

SKILL-KD is a contrastive skill distillation framework that improves LLM agents by distilling actionable discrepancies between teacher and student trajectories into textual skill patches, with drift-aware consolidation to iteratively refine skills.

This paper describes a two-stage vision-language adaptation system for Nepali meme classification, using Qwen3-VL-8B-Instruct with LoRA fine-tuning and contrastive learning. The system achieved 2nd place in hate speech detection and 4th in sentiment analysis at the CHiPSAL 2026 shared task.

This technical report introduces Douyin Multimodal Embedding (DME), a two-stage trained model that combines contrastive pre-training with evidence-grounded latent reasoning and cross-conditional reconstruction, achieving state-of-the-art results on MMEB-v2 and deployment in Douyin search.

Proposes FATE, a frame-level audio-visual temporal embedding method that aligns frame sequences on a physical timeline, enabling joint semantic and temporal understanding. It outperforms baselines on temporal retrieval, event localization, and generation evaluation metrics.

This paper analyzes contrastive critics used as value-like objectives in reinforcement learning, showing that good ranking accuracy does not make them safe to maximize due to off-support norm inflation and misranking, and demonstrates that value-calibrated scalar critics like TD-Q succeed where contrastive critics fail.

This paper proposes modifications to encoder-decoder transformers for unconstrained molecular structure elucidation from IR spectra, using a Mixture-of-Experts decoder and contrastive alignment loss, achieving over 10 percentage points improvement in Top-K accuracy.

The author revisits influential open-source works in representation learning, listing key papers from MoCo v1 to LeJEPA that advanced vision foundation models and self-supervised learning.

OmniVAE is a jointly trained audio-video VAE that uses segment-level contrastive learning and feature distillation to align latent spaces, improving joint generation quality and synchronization in text-to-audio-video generation.

VCSD removes the need for external teachers, privileged answers, or visual evidence in on-policy self-distillation by using content-erased control images to produce contrastive signals, consistently outperforming existing methods on vision-language benchmarks.

This paper proposes UMMT, a token-level cross-modal transformer with contrastive multi-task learning for breast cancer subtype classification and survival prediction, achieving state-of-the-art results on METABRIC and TCGA-BRCA datasets.

Proposes PEACE, a knowledge-guided framework for transferring adult ECG interpretation to pediatric populations using label-conditioned contrastive alignment, achieving significant improvements under limited supervision.

This paper investigates the failure of boundary-seeking knowledge distillation (CAKE) when applied to bottlenecked generative autoencoders, showing that the shared latent manifold creates gradient conflicts that prevent effective synthesis of contrastive samples. A simple noise forward pass baseline is proposed instead.

LLM4EHR proposes a clinical foundation model that temporally aligns Electronic Health Record time series with medical event sequences using a domain-adapted large language model and a regularized contrastive objective, improving downstream prediction tasks.