KOINEU

PerfDojo: Automated ML Library Generation for Heterogeneous Architectures

1. 연구 배경 및 동기 이기종 환경의 난제 : 현재 ML 워크로드는 CPU·GPU·TPU·FPGA 등 다양한 가속기에 배포된다. 각 아키텍처마다 명령어 집합, 메모리 계층, 연산 특성이 달라 동일한 커널을 최적화하기가 어렵다. 기존 자동화 한계 : 기존 AutoML/AutoTVM 등은 하드웨어‑특화 규칙 기반 혹은 탐색 기반 방법을 사용한다. 규칙이 복잡하고, 중간 표현이 비가시적이라 인간이 이해·수정하기 힘들다. 2. 핵심 아이디어 LLM 기반 코드 생성 : 대형 언어 모델을 활용해 고수준 연산을 인간이 이해 가능한 수학적 DS

PerfDojo: Automated ML Library Generation for Heterogeneous Architectures

POrTAL: Plan-Orchestrated Tree Assembly for Lookahead

Reinforcement Learning for Pollution Detection in a Randomized, Sparse and Nonstationary Environment with an Autonomous Underwater Vehicle

Remarks on the inverse Littlewood conjecture

Retrieval Augmented Generation (RAG) for Fintech: Agentic Design and Evaluation

Reward-Guided Discrete Diffusion via Clean-Sample Markov Chain for Molecule and Biological Sequence Design

SA-SSL-MOS: Self-supervised Learning MOS Prediction with Spectral Augmentation for Generalized Multi-Rate Speech Assessment

Scaling Laws for Economic Productivity: Experimental Evidence in LLM-Assisted Consulting, Data Analyst, and Management Tasks

Scenario Approach with Post-Design Certification of User-Specified Properties

ScRPO: From Errors to Insights

Solution Space Topology Guides CMTS Search

Sparse Additive Model Pruning for Order-Based Causal Structure Learning

Structured Uncertainty guided Clarification for LLM Agents

Surrogate-Based Prevalence Measurement for Large-Scale A/B Testing

Temporal Latent Variable Structural Causal Model for Causal Discovery under External Interferences

Test-Time Tuned Language Models Enable End-to-end De Novo Molecular Structure Generation from MS/MS Spectra

Time-to-Move: Training-Free Motion Controlled Video Generation via Dual-Clock Denoising

TiS-TSL: Image-Label Supervised Surgical Video Stereo Matching via Time-Switchable Teacher-Student Learning

Towards a Standard, Enterprise-Relevant Agentic AI Benchmark: Lessons from 5.5 billion tokens' worth of agentic AI evaluations

Towards a topological view of blood pressure regulation

Towards Reliable Human Evaluations in Gesture Generation: Insights from a Community-Driven State-of-the-Art Benchmark

Trajectory Design for UAV-Based Low-Altitude Wireless Networks in Unknown Environments: A Digital Twin-Assisted TD3 Approach

Tunable Ferroelectric Acoustic Resonators in Monolithic Thin-Film Barium Titanate

Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture

URDF-Anything: Constructing Articulated Objects with 3D Multimodal Language Model

Variational Polya Tree

Vector Symbolic Algebras for the Abstraction and Reasoning Corpus

Vibe Learning: Education in the age of AI

ViPRA: Video Prediction for Robot Actions

Weighted Contrastive Learning for Anomaly-Aware Time-Series Forecasting

What Are the Facts? Automated Extraction of Court-Established Facts from Criminal-Court Opinions

Zero Reinforcement Learning Towards General Domains

ZeroSyl: Simple Zero-Resource Syllable Tokenization for Spoken Language Modeling

A golden-ratio partition of information and the balance between prediction and surprise: a neuro-cognitive route to antifragility

A Lorentzian Equivariant Index Theorem

A Topology-Aware Graph Convolutional Network for Human Pose Similarity and Action Quality Assessment

A type theory for invertibility in weak $ω$-categories

A Universal Neural Receiver that Learns at the Speed of Wireless

A.E. Convergence vs Boundedness

ALDEN: Reinforcement Learning for Active Navigation and Evidence Gathering in Long Documents

An Interdisciplinary and Cross-Task Review on Missing Data Imputation

Anticoncentration of Random Sums in $mathbb{Z}_p$

Approximation Theory for Lipschitz Continuous Transformers

Bayesian Quadrature: Gaussian Processes for Integration

Beyond Shortest Path: Agentic Vehicular Routing with Semantic Context

Characterization of an MPPC-Based Scintillator Telescope and Measurement of Cosmic Muon Angular Distribution

CL API: Real-Time Closed-Loop Interactions with Biological Neural Networks

CLIN-LLM: A Safety-Constrained Hybrid Framework for Clinical Diagnosis and Treatment Generation

Combined dynamic-kinematic validation of droplet-wall impact modeling

Constraint-Informed Active Learning for End-to-End ACOPF Optimization Proxies

< Category Statistics (Total: 5051) >

Start searching

No results found