1
Training nGPT
把 Transformer 参数与激活约束到超球面，nGPT 带来表示学习新范式，一文看懂归一化训练的关键设计。
arXiv:2608.01284v1 Announce Type: cross Abstract: The normalized Transformer (nGPT) realizes hyperspherical representation learning by constraining mo…
2
Douyin Multimodal Embedding Model Technical Report
抖音多模态嵌入模型技术报告，揭秘工业级搜索推荐背后的向量表示学习。
arXiv:2608.02148v1 Announce Type: cross Abstract: Multimodal representation learning is a cornerstone of modern AI. By encoding multimodal queries and…
3
IRIS: Reusable Identity Representations from Frozen LLMs for Entity Alignment
利用冻结大语言模型生成可复用身份表示，实现精准且高效的实体对齐，为知识图谱融合提供新思路
arXiv:2607.25579v1 Announce Type: cross Abstract: Entity alignment (EA) identifies entities across knowledge graphs (KGs) that refer to the same real-…
4
Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs
新型越狱攻击方法，通过混合有害和无害的隐藏状态绕过安全对齐，揭示大模型防御漏洞。
arXiv:2508.10029v3 Announce Type: replace-cross Abstract: Safety-aligned large language models can still be manipulated through white-box intervention…
5
Graph Neural Networks are Heuristics
颠覆认知：GNN本质是高级启发式算法而非特征学习器，理论推导+实验验证给出新视角
arXiv:2601.13465v4 Announce Type: replace Abstract: Graph neural networks are usually treated as auxiliaries for combinatorial optimization: they imit…
6
Teacher Supervision over Representation Equivalence Classes
颠覆知识蒸馏常规认知：预训练表征只有等价类意义，匹配坐标是伪命题
arXiv:2607.03572v1 Announce Type: cross Abstract: Knowledge distillation is usually framed as a choice of what to match in the teacher - its logits, h…
7
Graph Representation Learning Augmented Model Manipulation on Federated Fine-Tuning of LLMs
联邦微调藏风险！图表示学习竟能放大模型操纵攻击，LLM安全防线告急。前沿研究不可错过。
arXiv:2605.07961v2 Announce Type: replace Abstract: Federated fine-tuning (FFT) has emerged as a privacy-preserving paradigm for collaboratively adapt…
8
Graph Representation Learning of Longitudinal Medical Imaging Trajectories for Treatment Response Prediction
把纵向医学影像建模成图，用 AI 预判治疗反应，为个性化医疗带来全新可能。
arXiv:2607.04912v1 Announce Type: cross Abstract: In patients with breast cancer, pathological complete response (pCR) has been established as a clini…
9
Fusion is not one-size-fits-all: Cross-Modal Representation Alignment for Time-to-Event Modeling
告别一刀切融合，提出针对时间事件建模的跨模态表示对齐新范式，精准提升多模态生存预测性能。
arXiv:2606.15038v1 Announce Type: new Abstract: Accurate time-to-event (TTE) prediction from multimodal clinical data remains challenging due to modal…
10
S$^2$COPE: Self-Supervised Concept Discovery via Preference Learning
自我监督概念发现新方法，用偏好学习破解可解释性与可扩展性的两难困境。
arXiv:2606.14586v1 Announce Type: new Abstract: Current representation learning paradigms force a fundamental compromise: self-supervised methods scal…
11
Temporal Straightening for Latent Planning
ICML 2026论文提出潜空间规划的时间直化方法，有效提升长期规划的连贯性与效率。
arXiv:2603.12231v2 Announce Type: replace Abstract: Learning good representations is essential for latent planning with world models. While pretrained…
12
Equivariant Representation Learning via Class-Pose Decomposition
提出通过类别-姿态分解实现等变表示学习，在变换不变性与泛化性上取得新突破。
arXiv:2207.03116v4 Announce Type: replace Abstract: We introduce a general method for learning representations that are equivariant to symmetries of d…
13
ReverseEOL: Improving Training-free Text Embeddings via Text Reversal in Decoder-only LLMs
无需训练，仅通过反转输入文本，就能显著提升解码器LLM的文本嵌入质量。
arXiv:2606.05858v1 Announce Type: new Abstract: Recent advances in Large Language Models (LLMs) have opened new avenues for generating training-free t…
14
Noise-Aware Visual Representation Learning for Medical Visual Question Answering
医学视觉问答新突破：噪声感知学习提升视觉表示鲁棒性
arXiv:2606.05535v1 Announce Type: cross Abstract: Medical visual question answering (Med-VQA) has strong potential for clinical decision support by en…
15
CoralBay: A Self-Supervised CT Foundation Model
自监督学习新突破：CoralBay开创CT基础模型，无需标注数据即可高效训练医学影像AI
arXiv:2606.03888v1 Announce Type: new Abstract: Self-supervised learning has enabled large-scale pre-training on 2D natural images, producing general-…
16
CURP: Codebook-based Continuous User Representation for Personalized Generation with LLMs
基于码本的连续用户表示方法，让LLM实现高效个性化生成，无需微调模型参数。
arXiv:2602.00742v2 Announce Type: replace Abstract: User modeling characterizes individuals through their preferences and behavioral patterns to enabl…
17
View Space: Learning Representation across Arbitrary Graphs
针对任意图结构学习表示的通用框架，原创性高，已被ICML 2026接收。
arXiv:2512.11561v2 Announce Type: replace Abstract: Generalizing pretrained models to unseen datasets without retraining is a central challenge toward…
18
Domain-Specific Data Synthesis for LLMs via Minimal Sufficient Representation Learning
KDD 2026接收，提出用最小充分表示学习为LLM高效合成领域数据，降低人工成本。
arXiv:2605.30039v1 Announce Type: new Abstract: Large Language Models have demonstrated remarkable progress in general-purpose capabilities and can ac…
19
Beyond Chain-of-Thought: Rewrite as a Universal Interface for Generative Multimodal Embeddings
超越思维链？新研究提出"重写"范式，将生成式多模态嵌入统一为通用接口，重塑AI推理与表示方式。
arXiv:2604.22280v2 Announce Type: replace Abstract: Multimodal Large Language Models (MLLMs) have emerged as a promising foundation for universal mult…
20
Training-Free Vector Quantization via Gaussian VAEs
提出一种无需训练的向量量化新方法，利用高斯VAE实现高效表示学习，为量化领域开辟新路径。
arXiv:2512.06609v3 Announce Type: replace Abstract: Vector-quantized variational autoencoders (VQ-VAEs) are discrete autoencoders that compress images…