AI Wiki — Knowledge Base

Google AI 인프라의 미래: 에이전틱 시대를 위한 확장

* 본 아티클의 원문은 2026년 4월 23일 Google Cloud 블로그(영문)에 게재되었습니다. AI는 질문에 답하는 수준을 넘어 추론하고 행동하는 단계로 진화하고 있습니다. 오늘날의 에이전틱 시대(agentic era)를 선도하고자 하는 기업에는 이러한 새로운 요구사항에 맞춰 설계되고 최적화된 컴퓨팅 인프라가 필요합니다. 오늘 Google Cloud

model-releasesgoogle-cloud-ai

28 Apr 2026

Search: “models”

Avionic Main Fuel Pump Simulation and Fault-Diagnosis Benchmark

BERT-APC: A Reference-free Framework for Automatic Pitch Correction via Musical Context Inference

Beyond the Attention Stability Boundary: Agentic Self-Synthesizing Reasoning Protocols

Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training

BRIEF-Pro: Universal Context Compression with Short-to-Long Synthesis for Fast and Accurate Multi-Hop Reasoning

Causal Representation Learning from General Environments under Nonparametric Mixing

Chat-Scene++: Exploiting Context-Rich Object Identification for 3D LLM

ContextWeaver: Selective and Dependency-Structured Memory Construction for LLM Agents

DARC-CLIP: Dynamic Adaptive Refinement with Cross-Attention for Meme Understanding

DeepTaxon: An Interpretable Retrieval-Augmented Multimodal Framework for Unified Species Identification and Discovery

Delta-XAI: A Unified Framework for Explaining Prediction Changes in Online Time Series Monitoring

Dialysis Risk Prediction and Treatment Effect Estimation for AKI patients using Longitudinal Electronic Health Records

Discovering Agentic Safety Specifications from 1-Bit Danger Signals

Distilling Self-Consistency into Verbal Confidence: A Pre-Registered Negative Result and Post-Hoc Rescue on Gemma 3 4B

Do Synthetic Trajectories Reflect Real Reward Hacking? A Systematic Study on Monitoring In-the-Wild Hacking in Code Generation

DyABD: The Abdominal Muscle Segmentation in Dynamic MRI Benchmark

DynProto: Dynamic Prototype Evolution for Out-of-Distribution Detection

Efficient learning by implicit exploration in bandit problems with side observations

EMCompress: Video-LLMs with Endomorphic Multimodal Compression

Empirical Ablation and Ensemble Optimization of a Convolutional Neural Network for CIFAR-10 Classification

Energy-Arena: A Dynamic Benchmark for Operational Energy Forecasting

Evaluating Remote Sensing Image Captions Beyond Metric Biases

EX-FIQA: Leveraging Intermediate Early eXit Representations from Vision Transformers for Face Image Quality Assessment

Explaining Temporal Graph Predictions With Shapley Values

Exploring Hierarchical Consistency and Unbiased Objectness for Open-Vocabulary Object Detection

Factual and Edit-Sensitive Graph-to-Sequence Generation via Graph-Aware Adaptive Noising

FastAT Benchmark: A Comprehensive Framework for Fair Evaluation of Fast Adversarial Training Methods

FastOMOP: A Foundational Architecture for Reliable Agentic Real-World Evidence Generation on OMOP CDM data

FOCUS: Fused Observation of Channels for Unveiling Spectra

FreqFormer: Hierarchical Frequency-Domain Attention with Adaptive Spectral Routing for Long-Sequence Video Diffusion Transformers

From Similarity to Structure: Training-free LLM Context Compression with Hybrid Graph Priors

From Stateless Queries to Autonomous Actions: A Layered Security Framework for Agentic AI Systems

Gated Tree Cross-Attention for Checkpoint-Compatible Syntax Injection in Decoder-Only LLMs

GCP: Guarded Collaborative Perception with Spatial-Temporal Aware Malicious Agent Detection

Generalising maximum mean discrepancy: kernelised functional Bregman divergences

GeoCert: Certified Geometric AI for Reliable Forecasting

GeoFunFlow-3D: A Physics-Guided Generative Flow Matching Framework for High-Fidelity 3D Aerodynamic Inference over Complex Geometries

Geometry-Aware Offline-to-Online Learning in Linear Contextual Bandits

Google AI 인프라의 미래: 에이전틱 시대를 위한 확장

GradMAP: Gradient-Based Multi-Agent Proximal Learning for Grid-Edge Flexibility

Graph-augmented Segmentation of Complex Shapes in Laser Powder bed Fusion for Enhanced In Situ Inspection

GSAR: Typed Grounding for Hallucination Detection and Recovery in Multi-Agent LLMs

Identity-Decoupled Anonymization for Visual Evidence in Multi-modal Retrieval-Augmented Generation

IMPA-Net: Meteorology-Aware Multi-Scale Attention and Dynamic Loss for Extreme Convective Radar Nowcasting

Implicit Framing in Obstetric Counseling Notes: A Grounded LLM Pipeline on a VBAC-Eligible Cohort

Inference of Online Newton Methods with Nesterov's Accelerated Sketching

Interpretable Physics-Informed Load Forecasting for U.S. Grid Resilience: SHAP-Guided Ensemble Validation in Hybrid Deep Learning Under Extreme Weather

It also has a tendency to just take on obligations to you without noting them anywhere.