Tagged articles
Representation Learning
48 articles · Page 1 of 1
Yann LeCun argues intelligence is rapid adaptation to the unknown, not knowledge accumulation. He presents JEPA and hierarchical H-JEPA as world models that predict action consequences in abstract representation spaces, and urges researchers to drop generative modeling, probabilistic models, contrastive learning, and reinforcement learning in favor of energy-based models, regularization, and model-predictive control.
Energy-Based ModelsH-JEPAJEPA
0 likes · 25 min read

LeCun and Balestriero prove image and text models trained on different views of the same scene align via a single rotation; a 109M text JEPA outperforms 72B Qwen on image-text alignment by pairing captions of the same image instead of next-token prediction.
JEPAPlatonic Representation HypothesisRepresentation Learning
0 likes · 7 min read

VLAct introduces a representation-centric continued pre-training framework for Vision-Language-Action models, achieving 92.5% on RoboTwin 2.0 and surpassing all World Action Models on RoboDojo using only 16 GPUs and open data; with 20% downstream data it outperforms GR00T N1.6 on unseen GR-1 robot.
Representation LearningRoboDojoRoboTwin
0 likes · 10 min read

As high‑quality text and image datasets near exhaustion, researchers explore "weight‑space learning"—treating trained neural‑network parameters as a learnable data modality—to extract knowledge from existing models, evaluate its feasibility, and identify challenges such as generation quality, geometry, and data collection.
Neural Network WeightsRepresentation LearningSelf-Supervised Learning
0 likes · 7 min read

The article explains that large‑model pre‑training is fundamentally a next‑word prediction task that forces the model to compress massive text corpora, discover statistical regularities, build semantic representations, and, at sufficient scale, exhibit emergent abilities, with practical implications for model selection and AI system design.
AI AlignmentPretrainingRAG
0 likes · 12 min read

This survey reviews continual self‑supervised learning (CSSL) for vision models, explaining why self‑supervised objectives better resist forgetting, outlining evaluation protocols, categorizing six CSSL method families, and discussing challenges such as unified benchmarks, large‑scale models, and multimodal drift.
CSSLContinual LearningRepresentation Learning
0 likes · 16 min read

This review systematically surveys 3D Scene Graph research from 2019‑2026, defining their structure, construction pipelines, applications, evaluation protocols, and highlighting open challenges such as unified definitions, dynamic modeling, functional affordances, and fragmented benchmarks that hinder real‑world deployment.
3D Scene GraphsDynamic ModelingRepresentation Learning
0 likes · 15 min read

The article argues that pixel‑level world models are too low‑level and costly for robotics, proposes the μ₀ representation—compact 3D interaction traces that capture object, tool and contact dynamics—demonstrates its training pipeline, experimental speed and success rates, and suggests it as a scalable, interpretable physical language for embodied agents.
3D interaction tracesEmbodied AIRepresentation Learning
0 likes · 11 min read

The article analyzes how Aether AI moves beyond statistical prediction toward causal world models, arguing that true physical‑world AI must identify the variables that actually drive outcomes, simulate interventions, and reason about changes, illustrated with robot manipulation examples and recent research results.
Physical AIRepresentation LearningRobotics
0 likes · 18 min read

The paper revisits Koopman representation learning through the lens of the information bottleneck, identifies the exact information needed for stable, long‑term dynamics propagation, proposes an information‑shaped objective combining mutual information and von Neumann entropy, and demonstrates superior prediction accuracy and stability across physical, visual, and graph‑based dynamical systems.
KoopmanRepresentation Learningmutual information
0 likes · 11 min read

From early 2024 explorations to a 2026 research surge, this review explains how large‑model latent space replaces explicit token‑based processing, outlines its five analytical lenses—foundation, evolution, mechanism, ability, outlook—compares representational properties, details architectural and computational strategies, enumerates new capabilities, and discusses remaining challenges and future directions.
Representation Learningartificial intelligencelarge models
0 likes · 20 min read

This article reviews two recent papers—LeWorldModel, which uses a minimal JEPA framework to train an end‑to‑end world model from pixels with only two loss terms, and Fast‑WAM, which questions the necessity of test‑time future imagination and achieves comparable performance with a faster inference pipeline.
JEPAModel Predictive ControlRepresentation Learning
0 likes · 9 min read

Bi-TEAM, a cross‑scale representation learning framework that injects local chemical variations into a global protein context, outperforms state‑of‑the‑art baselines on ten biochemical datasets, achieving a 350% boost in hemolysis prediction accuracy and a 66% MCC increase under strict scaffold splits.
Bi-TEAMRepresentation Learningchemical modification
0 likes · 15 min read

The article examines the chaotic rise of “world models” in AI, tracing their origins from early mental‑model theory to modern representation‑ and generation‑based approaches, and argues that the current hype reflects a broader shift away from large language models toward embodied, physics‑grounded intelligence.
AI researchRepresentation LearningWorld Models
0 likes · 13 min read

FreRA introduces a lightweight, plug‑in frequency‑refined augmentation that adaptively refines spectral components to preserve global semantics while injecting variance, dramatically improving contrastive learning performance on time‑series classification, anomaly detection, and transfer learning across multiple benchmark datasets.
Contrastive LearningData AugmentationRepresentation Learning
0 likes · 13 min read

This article summarizes four recent time‑series forecasting papers, covering a universal delay‑embedding foundation model, a dual causal network that leverages exogenous variables, a distribution‑aware alignment plug‑in called TimeAlign, and a shapelet‑based framework for interpretable directional forecasting in noisy financial markets.
Representation LearningTime Seriescausal network
0 likes · 9 min read

Tencent's LFM4Ads introduces a full‑representation, multi‑granular knowledge transfer framework that moves user, item, and cross representations from a large foundation model to downstream tasks, achieving up to 2.45% platform GMV uplift across more than ten advertising scenarios.
Representation Learningads recommendationfoundation model
0 likes · 12 min read

Tencent's LFM4Ads foundation model introduces a full-representation, multi-granular knowledge transfer framework that moves user, item, and cross representations to downstream tasks, dramatically improving ad recommendation metrics across dozens of business scenarios.
Representation Learningad recommendationfoundation model
0 likes · 10 min read


Amap TechJul 14, 2025 · Artificial Intelligence
The UPRE paper, presented at ICCV, introduces a multi‑view domain prompt and a unified representation enhancement to enable zero‑shot domain adaptation for object detection, achieving state‑of‑the‑art performance across diverse weather, geographic, and synthetic‑to‑real scenarios.
Computer VisionObject DetectionRepresentation Learning
0 likes · 10 min read

This article explains contrastive learning for visual language models, covering its definition, four‑step workflow, how to choose positive and negative pairs, the difference between supervised and self‑supervised variants, and why the technique is essential for zero‑shot and cross‑modal capabilities.
Contrastive LearningData AugmentationRepresentation Learning
0 likes · 6 min read

This article reviews cutting‑edge research on integrating large language models with graph‑based recommendation systems, detailing four key strategies—LLM node embeddings, deep graph‑LLM fusion, model‑driven graph data training, and text‑modal enhancements—while analyzing representation learning, InfoNCE optimization, explainable recommendations, and extensive experimental validation.
InfoNCELLMRepresentation Learning
0 likes · 18 min read

This article provides a comprehensive overview of user profiling algorithms, covering the evolution from ontology‑based traditional methods to modern deep‑learning approaches, including structured label prediction, representation learning, active learning, and large‑model integration, while discussing challenges, practical applications, and future research directions.
Active LearningRepresentation Learningdeep learning
0 likes · 26 min read

This article provides a comprehensive overview of user portrait algorithms, covering their historical development, ontology‑based traditional approaches, deep‑learning enhancements, representation‑learning techniques such as lookalike, active‑learning driven iteration, and the integration of large‑model world knowledge, while also discussing current challenges and future research directions.
Active LearningRepresentation Learningdeep learning
0 likes · 26 min read

This article presents a comprehensive overview of recent graph neural network research, covering adversarial representation learning for network embedding, block‑model guided GCN, enhanced class‑discriminative GNNs, self‑supervised contrastive GNNs, experimental results, and conclusions, highlighting their significance in real‑world applications.
GCNGraph Neural NetworksRepresentation Learning
0 likes · 13 min read

This article introduces the fundamentals of knowledge graphs, explains how graph neural networks can be adapted for knowledge graph reasoning, presents specialized GNN designs such as CompGCN and RED‑GNN, and discusses experimental results, interpretability, efficiency improvements, and future research directions.
Graph Neural NetworkKG reasoningRED-GNN
0 likes · 11 min read

This article reviews the importance and challenges of time‑series data, then summarizes recent contrastive‑learning approaches (TS2Vec, TF‑C) and mask‑reconstruction methods (PatchTST), highlighting their designs, key mechanisms, and future research directions.
Contrastive LearningPatchTSTRepresentation Learning
0 likes · 7 min read

This article reviews recent developments in causal representation learning, explaining why causal reasoning is essential, describing methods for i.i.d. data, time‑series, and multi‑distribution scenarios, and illustrating applications such as domain adaptation, video analysis, and financial data with numerous examples and visualizations.
Causal InferenceRepresentation Learningcausal discovery
0 likes · 22 min read

A recent paper by Prof. Ma Yi and Turing‑Award winner Yann LeCun introduces a simple, interpretable unsupervised learning approach that combines sparse coding, manifold learning, and slow feature analysis, achieving near‑state‑of‑the‑art performance on MNIST, CIFAR‑10, and CIFAR‑100 without data augmentation or extensive hyper‑parameter tuning.
AIRepresentation LearningUnsupervised Learning
0 likes · 8 min read

This article presents several mature knowledge‑graph application cases, including Meituan’s large‑scale “Meituan Brain” for lifestyle services, the Sage Knowledge Base platform by Fourth Paradigm, and additional examples in recommendation, medical, QA, and power‑industry domains, highlighting methods, challenges, and model designs.
AIGraph Neural NetworkRepresentation Learning
0 likes · 11 min read

This article reviews several mature knowledge‑graph applications, describing Meituan’s large‑scale “Meituan Brain” for lifestyle services, the Fourth Paradigm’s Sage Knowledge Base platform with various representation‑learning models, and additional use cases in recommendation, QA, drug discovery, and power‑grid domains.
AI applicationsGraph Neural NetworkIndustry Case Study
0 likes · 11 min read

This talk introduces automated knowledge graph representation learning, covering background, key techniques such as triple‑based, path‑based and subgraph‑based models, AutoML‑driven model search (AutoSF, Interstellar, RED‑GNN), evaluation metrics, and future research directions in AI.
AutoMLEmbeddingGraph Neural Networks
0 likes · 21 min read

The 4th International Workshop on Deep Learning Practice for High‑Dimensional Sparse and Imbalanced Data (DLP‑KDD 2022) invites submissions on deep‑learning systems, data representation, and user modeling for large‑scale sparse data, with a deadline of May 26, 2022 and acceptance notifications by June 20, 2022.
AIRepresentation LearningSparse Data
0 likes · 5 min read

This talk introduces the background, key directions, and model designs for automated knowledge‑graph representation learning, covering triple‑based, path‑based, and subgraph‑based approaches, the role of AutoML in searching optimal bilinear scoring functions, and future research challenges such as scalability, inductive inference, and domain‑specific applications.
AutoMLEmbeddingRepresentation Learning
0 likes · 20 min read

The AURL framework, presented by Pu Shi, introduces alignment‑uniformity aware representation learning for zero‑shot video classification, achieving up to 28 % top‑1 accuracy gains on UCF101 and HMDB51, and has already boosted business metrics in Tencent’s advertising, search, and video‑channel recommendation systems.
Computer VisionRepresentation Learningalignment
0 likes · 19 min read

The DataFun Summit 2022 Knowledge Graph Forum, held on March 12, presents cutting‑edge research on knowledge graph representation learning, multi‑hop reasoning, temporal KG question answering, and their applications in finance and retail, featuring talks by leading experts from JD, Fourth Paradigm, Stanford, and Meituan.
AI applicationsGraph Neural NetworksMulti-hop Reasoning
0 likes · 9 min read

On Jan 7, 2024, BIT’s “Hundred Lectures” will feature Assistant Professor Hanwang Zhang presenting his group‑theoretic self‑supervised representation learning work, including the IP‑IRM method that iteratively partitions data and applies invariant risk minimization to achieve fully disentangled visual features, with the session streamed via Tencent Meeting.
AIRepresentation LearningSelf-Supervised Learning
0 likes · 4 min read

This article explains the fundamentals of contrastive learning, reviews typical models such as SimCLR, MoCo, SwAV, BYOL, SimSiam and Barlow Twins, and demonstrates how these methods are applied to Weibo text and multimodal (text‑image) representation tasks like hashtag generation and image‑text matching.
Contrastive LearningNLPRepresentation Learning
0 likes · 18 min read

This article presents an overview of Xiaomi's knowledge graph platform, introduces text‑augmented knowledge representation learning methods such as Jointly and DKRL, and details their practical applications in entity linking, entity recommendation, and knowledge graph completion within AI‑driven services.
Representation Learningartificial intelligenceentity linking
0 likes · 20 min read

This article explains the concept of contrastive learning, its relationship to self‑supervised and metric learning, describes key system components and loss functions, reviews major image, NLP and multimodal models such as SimCLR, MoCo, SwAV, BYOL, and demonstrates how contrastive learning is applied to Weibo hashtag generation, similar‑post retrieval, and text‑image matching using CD‑TOM and W‑CLIP models.
AIContrastive LearningRepresentation Learning
0 likes · 19 min read

This article introduces Xiaomi's knowledge graph architecture, presents text‑graph joint representation learning methods, and demonstrates their practical use in entity linking, entity recommendation, and knowledge completion, highlighting model designs, feature engineering, and experimental results.
AIRepresentation Learningentity linking
0 likes · 16 min read

This article presents a comprehensive study on using graph representation learning, particularly GraphSAGE and its optimizations, to identify and mitigate black‑market accounts in virtual networks, covering background, algorithm design, handling isolated nodes and heterogeneity, and evaluation results.
Graph AlgorithmsGraphSAGERepresentation Learning
0 likes · 13 min read

This article reviews the rapid progress of knowledge graph completion, covering its background, formal problem definition, major technical approaches—including representation learning, path‑based search, reinforcement learning, logical reasoning, and meta‑learning—while discussing their challenges, recent improvements, and promising future research directions.
CompletionMeta-learningRepresentation Learning
0 likes · 14 min read


Didi TechMay 15, 2020 · Artificial Intelligence
The article outlines DiDi Food’s search relevance challenge, defines semantic matching versus traditional keyword methods, describes the recall‑ranking pipeline, and reviews three families of deep matching models—representation‑based (e.g., DSSM), interaction‑based (e.g., DRMM) and hybrid (e.g., DUET)—including experimental results and a recruitment notice.
DiDi FoodRepresentation Learningdeep matching
0 likes · 16 min read

This article presents a comprehensive overview of the challenges, requirements, and technical solutions for multilingual content understanding in UC's international information‑flow recommendation system, covering structured signal construction, low‑resource NLP techniques, transfer learning, quality modeling, and image‑based signal integration.
NLPRepresentation Learningcontent understanding
0 likes · 14 min read

This article provides a comprehensive overview of deep learning, covering essential mathematics and machine learning fundamentals, modern deep network architectures and regularization techniques, advanced research topics such as structured probabilistic models and generative methods, and a curated reading list for practitioners.
AI FundamentalsRepresentation Learningmachine learning
0 likes · 4 min read

The article explains how gcForest, an ensemble‑of‑decision‑tree model that mimics deep neural network hierarchies, uses cascade forests and multi‑grained sliding‑window scanning to achieve effective representation learning with fewer hyper‑parameters, especially on small datasets.
Ensemble MethodsRepresentation Learningcascade forest
0 likes · 11 min read

This article examines the construction, reasoning, and large‑scale applications of knowledge graphs, discussing graph building techniques, storage solutions, deep‑learning‑based entity extraction, inference models such as TransR and RESCAL, and how these graphs enhance search, recommendation, and other AI systems.
Representation Learningdeep learningentity recognition
0 likes · 13 min read

The talk reviews the history and concepts of knowledge graphs, explains logical and statistical reasoning methods—including rule‑based and representation‑learning approaches—and demonstrates how these techniques can be applied to build an intelligent robot that assists students in solving high‑school exam problems.
Representation Learningexam robotknowledge graph
0 likes · 5 min read
