arXiv data · independent discovery, no YC companies

FRONTIER · INDEPENDENT DISCOVERY

Signal, before it’s YC

Everything else on this site starts from a YC label. This doesn’t: arXiv paper abstracts across sixteen CS-adjacent categories are embedded and clustered independently, with no YC-derived keyword steering which papers get pulled in. Each cluster is tested against its own history, not a flat comparison — new means too young to have a trustworthy baseline yet, surging means its latest month is a real surprise (3× or more) against its own historical average, and sustained means it stayed elevated into a later check. Watching is the lowest-confidence tier on the whole site: real above-baseline activity (1.5×–3×) that just doesn’t clear the surging bar — shown rather than thrown away, but read it as barely worth a glance. There is no YC company data underneath any of it, and no backtest, since there is no YC-confirmed formation this could ever be checked against. Read it as a raw pulse from the literature, nothing more.

Signal, before it’s YC

Knowledge Distillation

Vla Model Visionlanguageaction Vla

Demonstration Robotic Manipulation

Video Understanding

Gaussian Splatting

Humanoid Robots

Failure LLM Agent

Policy Robotic Manipulation

Visual Visionlanguage Model

GPU LLM Inference

Depth Estimation

Model Vlms Visionlanguage Model

Tactile Sensing

Runtime LLM Agent

Frame Video Generation

Model Vlms Visionlanguage Model

Vulnerability Detection

Postquantum Quantum Computing

Lidar Autonomous Driving

Navigation Vln

Machine Translation

Verifiable Reward Reinforcement Learning

Human Pose

Memory LLM Agent

Vehicle Autonomous Driving

Attack LLM Agent

Reward Reinforcement Learning

Diffusion Model Video Generation

Multiarmed Bandit

Imitation Learning

Pose Estimation

Soft Robot

Protocol LLM Agent

Anomaly Detection

Retrievalaugmented Generation

Humanrobot Interaction

Evaluation Model Llms

Modality Multimodal Large

Prompt Injection

Diffusion Reinforcement Learning

Flow Matching

Code Coding Agent

Segmentation Medical Image

Attention Mechanism

Agentic Multiagent System

Recommender System

Speculative Decoding

LLM Agent Multiagent System

Task Planning

Semantic Segmentation

Gui Agent

Posttraining Quantization

Dexterous Manipulation

Automatic Speech Speech Recognition

Quadrotor Aerial Vehicle

Catastrophic Forgetting

Braincomputer Interface

LLM Serving

Keyvalue Cache

Multiobject Tracking

Model Merging

Policy Reinforcement Learning

Engineering Coding Agent

Code Generation

Bayesian Optimization

Report Generation

Skill LLM Agent

Human Model Llms

Dialogue System

Texttoimage Diffusion Model

Environment Reinforcement Learning

Cloud Registration Point Cloud

Bias Model Llms

Stochastic Gradient

Physicsinformed Neural

Denoising Diffusion Model

Generative Models

Quantum Machine

Question Answering