KOINEU Logo
No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์ด์‚ฐ ํ™•์‚ฐ ๋ชจ๋ธ ์€ ์ˆœ์ฐจ์ (autoregressive) ์ ‘๊ทผ๋ฒ•๊ณผ ๋‹ฌ๋ฆฌ ๋ฐ์ดํ„ฐ์— ๊ณ ์ •๋œ ์ˆœ์„œ๊ฐ€ ์—†์„ ๋•Œ๋„ ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ์ ์šฉ ๊ฐ€๋Šฅํ•˜๋‹ค๋Š” ์žฅ์ ์ด ์žˆ๋‹ค. ํ™”ํ•™ยท์ƒ๋ฌผ ๋ถ„์•ผ์—์„œ๋Š” ๋ณด์ƒ ํ•จ์ˆ˜ (drugโ€‘likeness, enhancer activity ๋“ฑ)๊ฐ€ ๊ทน๋„๋กœ ๋น„์—ฐ์† ์ด๋ฉฐ, ์ž‘์€ ํ† ํฐ ๋ณ€ํ˜•์ด ์ „์ฒด ๊ตฌ์กฐ๋ฅผ ๋ฌดํšจํ™”ํ•˜๊ฑฐ๋‚˜ ๋ณด์ƒ์„ 0์œผ๋กœ ๋งŒ๋“ ๋‹ค. ๊ธฐ์กด์˜ intermediateโ€‘reward ๊ธฐ๋ฐ˜ ๊ฐ€์ด๋“œ (SMC, SVDD, particleโ€‘based ๋“ฑ)๋Š” ๋…ธ์ด์ฆˆ๊ฐ€ ์„ž์ธ ์ค‘๊ฐ„ ์ƒ˜ํ”Œ ์— ๋Œ€ํ•ด ๋ณด์ƒ์„ ๊ณ„์‚ฐํ•˜๊ณ , ์ด๋ฅผ ๊ธฐ๋ฐ˜์œผ

Computer Science Machine Learning
No Image

RF-GPT: Teaching AI to See the Wireless World

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋ฉ€ํ‹ฐ๋ชจ๋‹ฌ LLM์˜ ํ•œ๊ณ„ : ํ˜„์žฌ GPTโ€‘4o, Gemini ๋“ฑ์€ ์ด๋ฏธ์ง€ยทํ…์ŠคํŠธยท์Œ์„ฑ๊นŒ์ง€ ํฌ๊ด„ํ•˜์ง€๋งŒ, ๊ณ ์ฃผํŒŒ ๋ณต์†Œ์ˆ˜ ์‹ ํ˜ธ์ธ RF๋Š” ์ „ํ˜€ ๋‹ค๋ฃจ์ง€ ์•Š๋Š”๋‹ค. ๊ธฐ์กด RF ๋”ฅ๋Ÿฌ๋‹์˜ ๋ฌธ์ œ์  : ์ž‘์—…โ€‘๋ณ„ ๋ชจ๋ธยท์†Œ๊ทœ๋ชจ ๋ผ๋ฒจ๋งยทํ•˜๋“œ์›จ์–ดยท์ฑ„๋„ ์˜์กด์„ฑ ๋“ฑ์œผ๋กœ ์žฌ์‚ฌ์šฉ์„ฑ์ด ๋‚ฎ๊ณ , ์„ค๋ช…ยท๋Œ€ํ™” ์ธํ„ฐํŽ˜์ด์Šค๊ฐ€ ๋ถ€์žฌํ•˜๋‹ค. ํ†ตํ•ฉ ์ธํ„ฐํŽ˜์ด์Šค ํ•„์š”์„ฑ : 6G ๋กœ๋“œ๋งต์ด ์ œ์‹œํ•˜๋Š” โ€œAIโ€‘native ๋„คํŠธ์›Œํฌโ€๋Š” ๋ฌผ๋ฆฌ์ธต๊นŒ์ง€ ์ดํ•ดํ•˜๊ณ  ์„ค๋ช…ํ•  ์ˆ˜ ์žˆ๋Š” ๋ชจ๋ธ์„ ์š”๊ตฌํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด RF ํ† ํฐํ™” : IQ ์‹œํ€€์Šค๋ฅผ STFT โ†’ ํŒŒ์›Œ ์ŠคํŽ™ํŠธ๋กœ๊ทธ๋žจ โ†’ (

Electrical Engineering and Systems Science
No Image

RobotArena $infty$: Scalable Robot Benchmarking via Real-to-Sim Translation

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์‹ค์„ธ๊ณ„ ํ…Œ์ŠคํŠธ์˜ ํ•œ๊ณ„ : ๋กœ๋ด‡ ์‹คํ—˜์€ ๋ฌผ๋ฆฌ์  ์žฅ๋น„ยท์ธ๋ ฅยท์‹œ๊ฐ„์ด ๋งŽ์ด ์†Œ์š”๋˜๊ณ , ์‚ฌ๊ณ  ์œ„ํ—˜์ด ์กด์žฌํ•œ๋‹ค. ํŠนํžˆ ๋Œ€๊ทœ๋ชจ ์ •์ฑ…(์˜ˆ: ๋Œ€ํ˜• ์–ธ์–ด ๋ชจ๋ธ ๊ธฐ๋ฐ˜ VLA) ํ‰๊ฐ€ ์‹œ ๋ฐ˜๋ณต ์‹คํ—˜์ด ๊ฑฐ์˜ ๋ถˆ๊ฐ€๋Šฅํ•˜๋‹ค. ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ๋ฒค์น˜๋งˆํฌ์˜ ๋ฌธ์ œ์  : ๊ธฐ์กด ์‹œ๋ฎฌ์€ โ€œ์‹œ๋ฎฌโ€‘์‹œ๋ฎฌโ€ ๋ฃจํ”„์— ๋จธ๋ฌผ๋Ÿฌ, ์‹ค์ œ ๋ฐ์ดํ„ฐ์—์„œ ํ•™์Šต๋œ ์ •์ฑ…์ด๋‚˜ ๋‹ค๋ฅธ ์‹œ๋ฎฌ๋ ˆ์ดํ„ฐ์™€์˜ ๊ต์ฐจ ๊ฒ€์ฆ์ด ์–ด๋ ค์›€. ์ด๋Š” โ€œ์‹œ๋ฎฌโ€‘ํ˜„์‹ค ๊ฒฉ์ฐจโ€(simโ€‘toโ€‘real gap)๋ฅผ ์€ํํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด | ์š”์†Œ | ๊ธฐ์กด ์ ‘๊ทผ | ์ œ์•ˆ ๋ฐฉ์‹ | | | | | | ๋ฐ์ดํ„ฐ ๋ณ€ํ™˜ | ์‹ค์„ธ

No Image

Robust and Diverse Multi-Agent Learning via Rational Policy Gradient

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๋Œ€๋ฆฝ์  ์ตœ์ ํ™”(Adversarial Optimization) ๋Š” ์ฃผ๋กœ ์ œ๋กœ์„ฌ ๊ฒŒ์ž„์—์„œ ์—์ด์ „ํŠธ์˜ ์•ฝ์ ์„ ์ฐพ์•„๋‚ด์–ด ๊ฐ•์ธํ•œ ์ •์ฑ…์„ ํ•™์Šตํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋ผ ์™”์Œ. ํ˜‘๋ ฅยท์ผ๋ฐ˜ํ•ฉ ํ™˜๊ฒฝ ์—์„œ๋Š” ๋™์ผํ•œ ์ ‘๊ทผ๋ฒ•์„ ์ ์šฉํ•˜๋ฉด ์—์ด์ „ํŠธ๊ฐ€ ์„œ๋กœ๋ฅผ ๋ฐฉํ•ดํ•˜๋„๋ก ํ•™์Šต๋˜๋Š” ์ž๊ธฐ ํŒŒ๊ดด ํ˜„์ƒ์ด ๋ฐœ์ƒ, ์ด๋Š” ์‹ค์ œ ํ˜‘์—… ์‹œ์Šคํ…œ์—์„œ ์น˜๋ช…์ ์ธ ์˜ค๋ฅ˜. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ด ๋ฌธ์ œ๋ฅผ ๋ณด์ƒ ์„ค๊ณ„ ํ˜น์€ ์ œ์•ฝ ์กฐ๊ฑด ์„ ์ถ”๊ฐ€ํ•˜๋Š” ์ˆ˜์ค€์— ๋จธ๋ฌผ๋ €์œผ๋ฉฐ, ๊ทผ๋ณธ์ ์ธ ํ•ฉ๋ฆฌ์„ฑ ๋ณด์žฅ ๋ฉ”์ปค๋‹ˆ์ฆ˜์€ ๋ถ€์žฌํ•จ. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด: RPO์™€ RPG | ์š”์†Œ | ์„ค๋ช… | ์—ญํ•  | | |

Learning
No Image

Robust Stochastic Gradient Posterior Sampling with Lattice Based Discretisation

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋ฒ ์ด์ฆˆ ํ•™์Šต ์€ ํŒŒ๋ผ๋ฏธํ„ฐ์— ๋Œ€ํ•œ ๋ถˆํ™•์‹ค์„ฑ์„ ์ •๋Ÿ‰ํ™”ํ•˜๊ณ  ๊ณผ์ ํ•ฉ์„ ๋ฐฉ์ง€ํ•˜๋Š” ๊ฐ•๋ ฅํ•œ ํ”„๋ ˆ์ž„์›Œํฌ์ด์ง€๋งŒ, MCMC ๋Š” ์ „์ฒด ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ๊ทธ๋ž˜๋””์–ธํŠธ ๊ณ„์‚ฐ์ด ํ•„์š”ํ•ด ๋Œ€๊ทœ๋ชจ ํ™˜๊ฒฝ์—์„œ ๋น„ํ˜„์‹ค์ ์ด๋‹ค. Stochasticโ€‘gradient MCMC(SGโ€‘MCMC) , ํŠนํžˆ SGLD๋Š” ๋ฏธ๋‹ˆ๋ฐฐ์น˜๋ฅผ ์‚ฌ์šฉํ•ด ๋น„์šฉ์„ ๋‚ฎ์ถ”์ง€๋งŒ ๋ฏธ๋‹ˆ๋ฐฐ์น˜ ํฌ๊ธฐ์— ๋ฏผ๊ฐ ํ•˜๊ณ , ์ž‘์€ ๋ฐฐ์น˜์—์„œ๋Š” ๋ฐœ์‚ฐํ•˜๊ฑฐ๋‚˜ ํฐ ํŒŒ๋ผ๋ฏธํ„ฐ ์ ํ”„๊ฐ€ ๋ฐœ์ƒํ•œ๋‹ค. ๊ธฐ์กด ํ•ด๊ฒฐ์ฑ…(๋ถ„์‚ฐ ๊ฐ์†Œ, ์ œ์–ด๋ณ€๋Ÿ‰, ์ ์‘ ์„œ๋ธŒ์ƒ˜ํ”Œ๋ง ๋“ฑ)์€ ๋ฉ”๋ชจ๋ฆฌยท์—ฐ์‚ฐ ์˜ค๋ฒ„ํ—ค๋“œ๊ฐ€ ํฌ๊ฑฐ๋‚˜ ๊ตฌํ˜„์ด ๋ณต์žกํ•˜๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด

Statistics Machine Learning
No Image

RoCE BALBOA: Service-enhanced Data Center RDMA for SmartNICs

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋ฐ์ดํ„ฐ์„ผํ„ฐ ๋„คํŠธ์›Œํฌ ๋ณ‘๋ชฉ : ๋Œ€๊ทœ๋ชจ ML ํŠธ๋ ˆ์ด๋‹ยท์ถ”๋ก , ๋ฐ์ดํ„ฐ ๋ถ„์„ ์›Œํฌ๋กœ๋“œ๋Š” ์ „ํ†ต์ ์ธ TCP/IP ์Šคํƒ์—์„œ ๋ฐœ์ƒํ•˜๋Š” ๋ณต์ˆ˜์˜ ๋ฉ”๋ชจ๋ฆฌ ๋ณต์‚ฌ์™€ CPU ๊ฐœ์ž…์œผ๋กœ ์ธํ•ด ๋„คํŠธ์›Œํฌ ์ง€์—ฐ์ด ํฌ๊ฒŒ ์ฆ๊ฐ€ํ•œ๋‹ค. RDMA์˜ ๋ถ€์ƒ : RDMA๋Š” ํ˜ธ์ŠคํŠธ ๋ฉ”๋ชจ๋ฆฌ์™€ ๋„คํŠธ์›Œํฌ ์‚ฌ์ด์˜ ์ง์ ‘ ๋ฐ์ดํ„ฐ ์ „์†ก์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•˜์—ฌ CPU ์˜ค๋ฒ„ํ—ค๋“œ์™€ ๋ ˆ์ดํ„ด์‹œ๋ฅผ ํฌ๊ฒŒ ๋‚ฎ์ถ˜๋‹ค. ํŠนํžˆ RoCE(RDMA over Converged Ethernet)๋Š” ๊ธฐ์กด ์ด๋”๋„ท ์ธํ”„๋ผ์™€ ํ˜ธํ™˜๋ผ ํด๋ผ์šฐ๋“œ ํ™˜๊ฒฝ์— ์ ํ•ฉํ•˜๋‹ค. ์Šค๋งˆํŠธNICยท์ธโ€‘๋„คํŠธ์›Œํฌ ์ปดํ“จํŒ… : ๋„คํŠธ์›Œํฌ ์ธํ„ฐํŽ˜์ด

Data
No Image

ROIX-Comp: Optimizing X-ray Computed Tomography Imaging Strategy for Data Reduction and Reconstruction

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฐ์ดํ„ฐ ํญ์ฆ : ์ตœ์‹  DIFRASยทIMX661 ๊ฒ€์ถœ๊ธฐ ๋“ฑ์€ 10 GB/s(โ‰ˆ 899 TB/์ผ) ์ˆ˜์ค€์˜ ์ŠคํŠธ๋ฆผ์„ ์ƒ์„ฑํ•œ๋‹ค. ์ „ํ†ต์  ์••์ถ• ํ•œ๊ณ„ : ์ผ๋ฐ˜ ๋ชฉ์  ์••์ถ•๊ธฐ(Gzip, Zstd ๋“ฑ)๋Š” Xโ€‘CT ํŠน์œ ์˜ ๊ณ ๋™์  ๋ฒ”์œ„ยท๋…ธ์ด์ฆˆยท๊ณต๊ฐ„ ์ƒ๊ด€์„ฑ์„ ํ™œ์šฉํ•˜์ง€ ๋ชปํ•œ๋‹ค. ROI ๊ธฐ๋ฐ˜ ์ ‘๊ทผ : ๋Œ€๋ถ€๋ถ„์˜ ์Šค์บ”์—์„œ ์‹ค์ œ ๊ณผํ•™์ ยท์ง„๋‹จ์  ๊ด€์‹ฌ ์˜์—ญ์€ ์ „์ฒด ๋ถ€ํ”ผ์˜ ๊ทนํžˆ ์ผ๋ถ€์ด๋ฉฐ, ๋ฐฐ๊ฒฝ์€ ์ €ํ•ด์ƒ๋„ยท๊ณ ์••์ถ•์ด ๊ฐ€๋Šฅํ•˜๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ๋ฒ• | ๋‹จ๊ณ„ | ์ฃผ์š” ๋‚ด์šฉ | ๊ธฐ๋Œ€ ํšจ๊ณผ | | | | | | Preโ€‘processing | โ€ข ๋ฐฐ๊ฒฝ ์ฐจ๊ฐ

Data Image Processing Electrical Engineering and Systems Science
No Image

ROSA: Roundabout Optimized Speed Advisory with Multi-Agent Trajectory Prediction in Multimodal Traffic

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ผ์šด๋“œ์–ด๋ฐ”์›ƒ์€ ๊ต์ฐจ๋กœ๋ณด๋‹ค ๋ณต์žกํ•œ ํ๋ฆ„๊ณผ ์ง€์†์ ์ธ ์ƒํ˜ธ์ž‘์šฉ์„ ์š”๊ตฌํ•œ๋‹ค. ํŠนํžˆ ๋ฉ€ํ‹ฐ๋ชจ๋‹ฌ ๊ตํ†ต (์ฐจ๋Ÿ‰ + ๋ณดํ–‰์žยท์ž์ „๊ฑฐ)์—์„œ๋Š” ์ธ๊ฐ„ ์šด์ „์ž๊ฐ€ ๋ˆˆ๋งž์ถคยท๋ฐ”๋””๋žญ๊ท€์ง€๋ฅผ ํ™œ์šฉํ•ด ๋ถˆํ™•์‹ค์„ฑ์„ ํ•ด์†Œํ•˜๋Š” ๋ฐ˜๋ฉด, ์ž๋™ํ™” ์ฐจ๋Ÿ‰์€ ์ด๋Ÿฌํ•œ ๋น„์–ธ์–ด์  ์‹ ํ˜ธ๋ฅผ ํ™œ์šฉํ•  ์ˆ˜ ์—†์–ด ์œ„ํ—˜์ด ์ฆ๋Œ€๋œ๋‹ค. ๊ธฐ์กด ์ธํ”„๋ผ ๊ธฐ๋ฐ˜ ํ•ด๊ฒฐ์ฑ…(์šฐ์„  ๋ณดํ–‰์ž ํšก๋‹จ๋ณด๋„, ์‹ ํ˜ธ๋“ฑ)์€ ์ •์  ์ด๋ฉฐ, ๊ตํ†ต ํ๋ฆ„์„ ์ €ํ•ดํ•˜๊ณ  ๋ฐฐ์ถœ๋Ÿ‰์„ ์ฆ๊ฐ€์‹œํ‚จ๋‹ค. ๋”ฐ๋ผ์„œ ๋™์ ์ธ ์˜ˆ์ธก + ์‹ค์‹œ๊ฐ„ ์†๋„ ์กฐ์ • ์ด ๊ฐ€๋Šฅํ•œ ์‹œ์Šคํ…œ์ด ์ ˆ์‹คํžˆ ์š”๊ตฌ๋œ๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ (Contributions) | ๋ฒˆํ˜ธ |

Computer Science Multiagent Systems
No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‹ค์ค‘ ๋ ˆ์ดํŠธ ์Œ์„ฑ ํ’ˆ์งˆ ํ‰๊ฐ€์˜ ๋‚œ์ œ ํ˜„์žฌ ๋Œ€๋ถ€๋ถ„์˜ SSL ๊ธฐ๋ฐ˜ SQA ๋ชจ๋ธ์€ 16 kHz ์ „์šฉ ์‚ฌ์ „ํ•™์Šต์„ ์‚ฌ์šฉํ•ด 24 kHzยท48 kHz์™€ ๊ฐ™์€ ๊ณ ํ•ด์ƒ๋„ ์Œ์„ฑ์˜ ๊ณ ์ฃผํŒŒ ์„ฑ๋ถ„์„ ๋ฌด์‹œํ•œ๋‹ค. ๊ณ ์ฃผํŒŒ๋Š” ํŠนํžˆ ๊ณ ์Œ์—ญ๋Œ€ ์žก์Œ, ์™œ๊ณก, ์ฝ”๋ฑ ์†์‹ค ๋“ฑ์„ ๊ฐ์ง€ํ•˜๋Š” ๋ฐ ์ค‘์š”ํ•œ ์—ญํ• ์„ ํ•˜๋ฏ€๋กœ, ์ด๋ฅผ ๋ฌด์‹œํ•˜๋ฉด MOS ์˜ˆ์ธก ์ •ํ™•๋„๊ฐ€ ๋–จ์–ด์ง„๋‹ค. ๋ฐ์ดํ„ฐ ๋ถ€์กฑ MOS ๋ผ๋ฒจ์ด ์žˆ๋Š” ๋‹ค์ค‘ ๋ ˆ์ดํŠธ ๋ฐ์ดํ„ฐ๋Š” ๋งค์šฐ ์ œํ•œ์ ์ด๋ฉฐ, AudioMOS 2025 ์ฑŒ๋ฆฐ์ง€ ๋ฐ์ดํ„ฐ์…‹(์ด 800๊ฐœ ์ƒ˜ํ”Œ) ์ •๋„๊ฐ€ ์ „๋ถ€์ด๋‹ค. ๋ผ๋ฒจ๋ง ๋น„์šฉ๊ณผ ์ธ๊ฐ„ ์ฒญ์ทจ์ž ํ‰๊ฐ€์˜ ์ฃผ๊ด€

Learning Audio Processing Electrical Engineering and Systems Science
No Image

Safe hypotheses testing with application to order restricted inference

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์ˆœ์„œ ์ œํ•œ ์ถ”๋ก (ORI) ์€ ์„ ํ˜•/๋ณผ๋ก ์ œ์•ฝ์„ ์ด์šฉํ•ด ํ†ต๊ณ„์  ํšจ์œจ์„ฑ์„ ๋†’์ด๋Š” ์ „ํ†ต์  ๋ฐฉ๋ฒ•์ด๋ฉฐ, Barlow et al. (1972), Robertson et al. (1988) ๋“ฑ์—์„œ ์ฒด๊ณ„ํ™”๋˜์—ˆ๋‹ค. Type III ์˜ค๋ฅ˜ ๋Š” ์ œ์•ฝ์ด ์‹ค์ œ ํŒŒ๋ผ๋ฏธํ„ฐ ๊ณต๊ฐ„์— ํฌํ•จ๋˜์ง€ ์•Š์„ ๋•Œ ๋ฐœ์ƒํ•œ๋‹ค. ๊ธฐ์กด ๋ฌธํ—Œ์—์„œ๋Š” ์ด ํ˜„์ƒ์ด โ€œ๋“œ๋ฌผ๋‹คโ€๊ณ  ์–ธ๊ธ‰ํ–ˆ์ง€๋งŒ, ์ €์ž๋Š” Type A ๋ฌธ์ œ(์˜ˆ: ฮธโˆˆC)์—์„œ ๊ฑฐ์˜ ๋ณดํŽธ์ ์œผ๋กœ ๋‚˜ํƒ€๋‚œ๋‹ค๊ณ  ์ฃผ์žฅํ•œ๋‹ค(์˜ˆ 2.1, 2.2). ์ด๋Ÿฌํ•œ ์˜ค๋ฅ˜๋Š” ํŠนํžˆ ๊ณ ์ฐจ์›ยท๋‹ค๋ณ€๋Ÿ‰ ์ƒํ™ฉ์—์„œ ์œ„ํ—˜๋„๊ฐ€ ๊ธ‰์ฆํ•œ๋‹ค๋Š” ์ ์„ ๊ฐ•์กฐ

Statistics
No Image

Sample size and power determination for assessing overall SNP effects in joint modeling of longitudinal and time-to-event data

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์—ฐ์† ๋ฐ”์ด์˜ค๋งˆ์ปค์™€ ์‚ฌ๊ฑด ๋ฐ์ดํ„ฐ์˜ ๊ฒฐํ•ฉ ์€ ๋‹น๋‡จ๋ณ‘ยท์‹ฌํ˜ˆ๊ด€ ์งˆํ™˜ ๋“ฑ์—์„œ ์œ„ํ—˜ ์˜ˆ์ธก ์ •ํ™•๋„๋ฅผ ํฌ๊ฒŒ ๋†’์ธ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ๋ชจ๋ธ๋ง ๋ฐฉ๋ฒ• ์— ์ง‘์ค‘ํ–ˆ์œผ๋‚˜, ์ƒ˜ํ”Œ ์‚ฌ์ด์ฆˆยท๊ฒ€์ •๋ ฅ ์„ค๊ณ„๋Š” ๊ฑฐ์˜ ๋‹ค๋ฃจ์ง€ ์•Š์•˜๋‹ค. ์œ ์ „ํ•™ ์—ฐ๊ตฌ์—์„œ๋Š” SNP๊ฐ€ ์ง์ ‘ยท๊ฐ„์ ‘(๋ฐ”์ด์˜ค๋งˆ์ปค ๋งค๊ฐœ) ๋‘ ๊ฒฝ๋กœ๋กœ ์‚ฌ๊ฑด ์œ„ํ—˜์— ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ์–ด, ์ „์ฒด ํšจ๊ณผ๋ฅผ ํ•œ ๋ฒˆ์— ๊ฒ€์ •ํ•˜๋Š” ๊ฒƒ์ด ์‹ค์งˆ์  ์˜๋ฏธ๊ฐ€ ํฌ๋‹ค. 2. ์ œ์•ˆ๋œ ํ†ต๊ณ„ ๋ชจ๋ธ | ๊ตฌ์„ฑ ์š”์†Œ | ์„ค๋ช… | | | | | Longitudinal subโ€‘model | (Y i(t) eta i(t)+varepsil

Data Statistics Model
No Image

Scaling Laws for Masked-Reconstruction Transformers on Single-Cell Transcriptomics

1. ์—ฐ๊ตฌ์˜ ์˜์˜์™€ ๋ฐฐ๊ฒฝ ์Šค์ผ€์ผ๋ง ๋ฒ•์น™์˜ ์ผ๋ฐ˜ํ™” : ๊ธฐ์กด ์—ฐ๊ตฌ(Kaplan et al., 2020; Hestness et al., 2017)๋Š” ์–ธ์–ดยท์‹œ๊ฐ ๋ถ„์•ผ์—์„œ ํŒŒ์›Œโ€‘law๊ฐ€ ๋ณดํŽธ์ ์ž„์„ ๋ณด์—ฌ์คฌ๋‹ค. ๋ณธ ๋…ผ๋ฌธ์€ ์ด๋ฅผ ์ƒ๋ช…๊ณผํ•™ , ํŠนํžˆ ๋‹จ์ผ ์„ธํฌ ์ „์‚ฌ์ฒด ์— ์ตœ์ดˆ๋กœ ์ ์šฉํ•จ์œผ๋กœ์จ, ๋ฐ์ดํ„ฐยท๋ชจ๋ธ ๊ทœ๋ชจ๊ฐ€ ์ถฉ๋ถ„ํ•  ๋•Œ ๋™์ผํ•œ ํ˜„์ƒ์ด ๋‚˜ํƒ€๋‚œ๋‹ค๋Š” ์ค‘์š”ํ•œ ์ฆ๊ฑฐ๋ฅผ ์ œ๊ณตํ•œ๋‹ค. ๋‹จ์ผ ์„ธํฌ ๊ธฐ์ดˆ ๋ชจ๋ธ์˜ ํ˜„์ฃผ์†Œ : scVI, scGPT, Geneformer ๋“ฑ ์—ฌ๋Ÿฌ ๋Œ€๊ทœ๋ชจ ์‚ฌ์ „ํ•™์Šต ๋ชจ๋ธ์ด ๋“ฑ์žฅํ–ˆ์ง€๋งŒ, ํ”„๋ฆฌํŠธ๋ ˆ์ธ ์†์‹ค์ด ๋ชจ๋ธ ํฌ๊ธฐ์— ๋”ฐ๋ผ ์–ด๋–ป๊ฒŒ ๋ณ€ํ•˜๋Š”์ง€ ๋Š” ์ฒด๊ณ„

Computer Science Machine Learning
No Image

Scaling limits for some Mittag-Leffler queues

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋ฌด๊ฑฐ์šด ๊ผฌ๋ฆฌ ํ˜„์ƒ : ๊ธˆ์œต, ํ†ต์‹ , ์˜๋ฃŒ ๋“ฑ ๋‹ค์–‘ํ•œ ๋ถ„์•ผ์—์„œ ๋„์ฐฉยท์„œ๋น„์Šค ์‹œ๊ฐ„์ด ๋ฌดํ•œ ํ‰๊ท ์„ ๊ฐ–๋Š” ๊ฒฝ์šฐ๊ฐ€ ๋ณด๊ณ ๋˜๊ณ  ์žˆ๋‹ค. ๊ธฐ์กด $M/M/1$, $M/G/1$, $G/G/1$ ์ด๋ก ์€ ์œ ํ•œ ํ‰๊ท  ๊ฐ€์ •์— ํฌ๊ฒŒ ์˜์กดํ•˜๋ฏ€๋กœ, ์ด๋Ÿฌํ•œ ์ƒํ™ฉ์„ ๋‹ค๋ฃจ๊ธฐ์—” ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค. Mittagโ€‘Leffler ๋ถ„ํฌ : ์ง€์ˆ˜๋ถ„ํฌ์˜ ์ž์—ฐ์Šค๋Ÿฌ์šด ์ผ๋ฐ˜ํ™”์ด๋ฉฐ, ์•ˆ์ •์  ์„œ๋ธŒ์˜ค๋””๋„ค์ดํ„ฐ(inverse stable subordinator)์˜ ๋ผํ”Œ๋ผ์Šค ๋ณ€ํ™˜์œผ๋กœ ์ •์˜๋œ๋‹ค. ํŒŒ์›Œโ€‘๋ฒ•์น™ ๊ผฌ๋ฆฌ๋ฅผ ๊ฐ€์ง€๋ฉด์„œ๋„ ํ”„๋ž™ํƒˆ ๋ฏธ๋ถ„ ๋ฐฉ์ •์‹(Caputo ๋ฏธ๋ถ„)์˜ ๊ณ ์œ ํ•จ์ˆ˜๋ผ๋Š” ํŠน

Mathematics
No Image

Scaling Open Discrete Audio Foundation Models with Interleaved Semantic, Acoustic, and Text Tokens

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ | ๊ธฐ์กด ์ ‘๊ทผ | ํ•œ๊ณ„ | | | | | LLMโ€‘centric (SALMONN, Qwen3โ€‘Omni ๋“ฑ) | ํ…์ŠคํŠธ ๋ฐฑ๋ณธ์— ์˜ค๋””์˜ค ๋ชจ๋“ˆ์„ ๋ถ€์ฐฉ โ†’ ์˜๋ฏธ ๋ณ‘๋ชฉ, ๊ณ ์Œ์งˆ ์˜ค๋””์˜ค ์ƒ์„ฑ ์–ด๋ ค์›€ | | Semanticโ€‘only (TWIST, SpiritLM ๋“ฑ) | ์˜๋ฏธ ํ† ํฐ๋งŒ ์‚ฌ์šฉ โ†’ ์Œํ–ฅ ๋””ํ…Œ์ผ ์†์‹ค, ๊ณ ํ’ˆ์งˆ TTSยทASR ํ•œ๊ณ„ | | Native audio (Moshi, Llamaโ€‘Mimi) | ์Œํ–ฅ ํ† ํฐ ์ง์ ‘ ๋ชจ๋ธ๋งํ•˜์ง€๋งŒ ํ…์ŠคํŠธ์™€์˜ ํ†ตํ•ฉ ์ด ๋ถ€์กฑ, ์Šค์ผ€์ผ๋ง ์—ฐ๊ตฌ ๋ถ€์žฌ | ๋”ฐ๋ผ์„œ ์Œ์„ฑยท์Œํ–ฅยทํ…์ŠคํŠธ๋ฅผ ํ•˜๋‚˜์˜ ์‹œํ€€

Computer Science Sound Model
No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๊ธฐ์กด ์‹œ๋‚˜๋ฆฌ์˜ค ์ ‘๊ทผ๋ฒ•๊ณผ์˜ ์ฐจ๋ณ„์  | ๊ตฌ๋ถ„ | ๊ธฐ์กด ์‹œ๋‚˜๋ฆฌ์˜ค ์ ‘๊ทผ๋ฒ• (Campi & Garatti, 2008โ€‘โ€ฆ) | ๋ณธ ๋…ผ๋ฌธ์˜ ํ™•์žฅ | | | | | | ๋ชฉํ‘œ | ์„ค๊ณ„ ์‹œ์ ์— ์ •์˜๋œ ์ ํ•ฉ์„ฑ(์˜ˆ: ์•ˆ์ •์„ฑ) ๋ณด์žฅ | ์„ค๊ณ„ ํ›„ ์ถ”๊ฐ€ ์ ํ•ฉ์„ฑ(์˜ˆ: ๋งˆ์ง„, ๋น„๋ณผ๋ก ๋ชฉํ‘œ) ๊ฒ€์ฆ | | ๋ฐ์ดํ„ฐ ์‚ฌ์šฉ | ์„ค๊ณ„ยท์ธ์ฆ์„ ๋™์ผ ์ƒ˜ํ”Œ๋กœ ์ˆ˜ํ–‰ | ๋™์ผ ์ƒ˜ํ”Œ๋งŒ ์‚ฌ์šฉํ•ด ์‚ฌํ›„ ์œ„ํ—˜ ์ƒยทํ•˜ํ•œ์„ ๊ณ„์‚ฐ | | ๊ฐ€์ • | ๋น„ํ‡ดํ™”(nonโ€‘degeneracy) ๊ฐ€์ •์ด ํ•„์š” (์ผ๋ถ€ ์—ฐ๊ตฌ์—์„œ๋Š” ์ œ๊ฑฐ) | ๋น„ํ‡ดํ™” ๊ฐ€์ • ์—†์ด๋„ ์ƒํ•œ ์ œ๊ณต, ํ•˜ํ•œ์€ ์ถ”๊ฐ€ ๊ฐ€์ • ํ•„์š”

Statistics
No Image

SCENE OTA-FD: Self-Centering Noncoherent Estimator for Over-the-Air Federated Distillation

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ OTAโ€‘FD ๋Š” ์—ฐํ•ฉ ํ•™์Šต์—์„œ ๋ชจ๋ธ ํŒŒ๋ผ๋ฏธํ„ฐ ๋Œ€์‹  ์†Œํ”„ํŠธ ๋ผ๋ฒจ(์˜ˆ์ธก ํ™•๋ฅ )์„ ๊ตํ™˜ํ•จ์œผ๋กœ์จ ํ†ต์‹  ๋น„์šฉ์„ ํฌ๊ฒŒ ์ ˆ๊ฐํ•œ๋‹ค. ๊ธฐ์กด ๋Œ€๋ถ€๋ถ„์˜ OTAโ€‘FD ์„ค๊ณ„๋Š” ์ฝ”ํžˆ์–ด๋ŸฐํŠธ ๋ฐฉ์‹์œผ๋กœ, ๋งค ๋ผ์šด๋“œ๋งˆ๋‹ค CSI(์ฑ„๋„ ์ƒํƒœ ์ •๋ณด)์™€ ์œ„์ƒ ์ •๋ ฌ์„ ์œ„ํ•ด ํŒŒ์ผ๋Ÿฟ ์‹ ํ˜ธ๋ฅผ ์ „์†กํ•œ๋‹ค. ์ด๋Š” ํŒŒ์ผ๋Ÿฟ ์˜ค๋ฒ„ํ—ค๋“œ, CFO/์œ„์ƒ ์žก์Œ์— ๋Œ€ํ•œ ๋ฏผ๊ฐ๋„, ๊ทธ๋ฆฌ๊ณ  ์งง์€ ์ฝ”ํžˆ์–ด๋Ÿฐ์Šค ๊ตฌ๊ฐ„์—์„œ์˜ ๋น„ํšจ์œจ์„ฑ์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ๋…ผ๋ฌธ์€ ์ด๋Ÿฌํ•œ ํ•œ๊ณ„๋ฅผ ์™„์ „ํžˆ ๋น„๋™๊ธฐ์‹(nonโ€‘coherent) ์ ‘๊ทผ์œผ๋กœ ์ „ํ™˜ํ•จ์œผ๋กœ์จ, ํŒŒ์ผ๋Ÿฟ ๋น„์šฉ์„ 0์œผ๋กœ ๋งŒ๋“ค๊ณ  ํ•˜๋“œ์›จ์–ด ๊ตฌํ˜„์„ ๋‹จ์ˆœํ™”ํ•˜๋Š”

Electrical Engineering and Systems Science
No Image

ScenicRules: An Autonomous Driving Benchmark with Multi-Objective Specifications and Abstract Scenarios

1๏ธโƒฃ ์—ฐ๊ตฌ ๋™๊ธฐ์™€ ํ•„์š”์„ฑ ๋‹ค๋ชฉ์  ์ถฉ๋Œ : ์‹ค์ œ ๋„๋กœ์—์„œ๋Š” ์ถฉ๋Œ ํšŒํ”ผ vs ์ฐจ์„  ์œ ์ง€, ์Šน์ฐจ๊ฐ vs ๋„๋กœ ๊ทœ์น™ ๋“ฑ ๋™์‹œ ๋งŒ์กฑ์ด ๋ถˆ๊ฐ€๋Šฅ ํ•œ ์ƒํ™ฉ์ด ๋นˆ๋ฒˆํžˆ ๋ฐœ์ƒํ•œ๋‹ค(๊ทธ๋ฆผ 1aยทb). ์šฐ์„ ์ˆœ์œ„ ๋ช…์‹œ์˜ ๋ถ€์žฌ : ๊ธฐ์กด ๋ฒค์น˜๋งˆํฌ๋Š” ๋‹จ์ผ ๋ชฉํ‘œ(์˜ˆ: ์ถฉ๋Œ์œจ) ํ˜น์€ ๊ฐ€์ค‘์น˜ ๊ธฐ๋ฐ˜ ๋‹ค๋ชฉ์  ํ‰๊ฐ€์ง€ํ‘œ๋งŒ ์ œ๊ณต, ๋ช…์‹œ์  ์šฐ์„ ์ˆœ์œ„ ๊ด€๊ณ„ ๊ฐ€ ์—†์–ด์„œ ์‹ค์ œ ์˜์‚ฌ๊ฒฐ์ • ๊ณผ์ •์„ ํ‰๊ฐ€ํ•˜๊ธฐ ์–ด๋ ค์›€. ์‹œ๋‚˜๋ฆฌ์˜ค ํ‘œํ˜„์˜ ํ•œ๊ณ„ : ๋Œ€๊ทœ๋ชจ ์‹ค๋ฐ์ดํ„ฐ๋Š” ํฌ๊ด„์„ฑยท์žฌํ˜„์„ฑ ์ด ๋–จ์–ด์ง€๊ณ , ๊ฒ€์ฆ์— ๋น„ํšจ์œจ์ ์ด๋‹ค. ๋”ฐ๋ผ์„œ ๋ชฉํ‘œยท์šฐ์„ ์ˆœ์œ„ยท์‹œ๋‚˜๋ฆฌ์˜ค ๋ฅผ ๋™์‹œ์— ๋‹ค๋ฃจ๋Š” ํ†ตํ•ฉ ํ”„๋ ˆ์ž„์›Œํฌ๊ฐ€ ์ ˆ์‹คํžˆ ์š”๊ตฌ๋œ๋‹ค.

Robotics Computer Science
No Image

scPilot: Large Language Model Reasoning Toward Automated Single-Cell Analysis and Discovery

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‹จ์ผ์„ธํฌ ์˜ค๋ฏน์Šค ๊ฐ€ ๊ธ‰์ฆํ•˜๋ฉด์„œ ์ „ํ†ต์ ์ธ ๋ถ„์„ ํŒŒ์ดํ”„๋ผ์ธ์€ ์ธ๊ฐ„ ์ „๋ฌธ๊ฐ€์˜ ์•”๋ฌต์  ํŒ๋‹จ ์— ํฌ๊ฒŒ ์˜์กดํ•˜๊ณ  ์žˆ๋‹ค(์˜ˆ: ํŒŒ๋ผ๋ฏธํ„ฐ ํŠœ๋‹, ๊ฒฐ๊ณผ ํ•ด์„). ๊ธฐ์กด LLM ํ™œ์šฉ ์‚ฌ๋ก€๋Š” ํˆด ์—์ด์ „ํŠธ ์ˆ˜์ค€์— ๋จธ๋ฌผ๋Ÿฌ, LLM์ด ์ž์—ฐ์–ด ์„ค๋ช…์„ ์ œ๊ณตํ•˜์ง€๋งŒ ์‹ค์ œ ๋ฐ์ดํ„ฐ์™€ ์—ฐ์‚ฐ ์‚ฌ์ด์˜ ๋…ผ๋ฆฌ์  ์—ฐ๊ฒฐ๊ณ ๋ฆฌ๋Š” ๋ถ€์กฑํ–ˆ๋‹ค. ๋”ฐ๋ผ์„œ โ€œomicsโ€‘native reasoningโ€ ์ด๋ผ๋Š” ๊ฐœ๋…์„ ๋„์ž…ํ•ด LLM์ด ๋ฐ์ดํ„ฐ โ†’ ๊ฐ€์„ค โ†’ ์ฆ๊ฑฐ โ†’ ์ˆ˜์ • ์˜ ์ˆœํ™˜์„ ์ง์ ‘ ์ˆ˜ํ–‰ํ•˜๋„๋ก ํ•˜๋Š” ๊ฒƒ์ด ํ•ต์‹ฌ ๋ชฉํ‘œ์ด๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ Omicsโ€‘Native Reas

Computer Science Artificial Intelligence Analysis Model
No Image

SeaSpoofFinder -- Potential GNSS Spoofing Event Detection Using AIS

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ GNSS ์Šคํ‘ธํ•‘ ์œ„ํ˜‘ ์€ 20๋…„ ๋„˜๊ฒŒ ํ•™๊ณ„ยท์‚ฐ์—…๊ณ„์—์„œ ๊ฒฝ๊ณ ๋˜์–ด ์™”์œผ๋ฉฐ, ์ตœ๊ทผ ํ•ญ๊ณต ๋ถ„์•ผ์—์„œ ADSโ€‘B ๋ฐ์ดํ„ฐ๋ฅผ ์ด์šฉํ•œ ๋Œ€๊ทœ๋ชจ ๋ชจ๋‹ˆํ„ฐ๋ง ์‚ฌ๋ก€๊ฐ€ ๋ณด๊ณ ๋˜์—ˆ๋‹ค. ํ•ด์–‘ ๋ถ„์•ผ์—์„œ๋Š” AIS ๊ฐ€ ์ „ ์„ธ๊ณ„์ ์œผ๋กœ ์‹ค์‹œ๊ฐ„ ์œ„์น˜ ์ •๋ณด๋ฅผ ์ œ๊ณตํ•˜๋ฏ€๋กœ, ๋™์ผํ•œ ์ ‘๊ทผ๋ฒ•์„ ์ ์šฉํ•  ์ˆ˜ ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ๋Š” ๊ณต๊ฐœยท๋ฌด๋ฃŒ ์ธ AIS ์ŠคํŠธ๋ฆผ์„ ํ™œ์šฉํ•ด ์Šคํ‘ธํ•‘ ์ง•ํ›„๋ฅผ ํƒ์ง€ํ•˜๊ณ , ์ด๋ฅผ ์‹œ๊ฐํ™”ยท๊ณต์œ ํ•จ์œผ๋กœ์จ ํ•ด์–‘ ๋ณด์•ˆ ์ปค๋ฎค๋‹ˆํ‹ฐ์— ์ดˆ๊ธฐ ๊ฒฝ๋ณด ์ฒด๊ณ„๋ฅผ ์ œ๊ณตํ•œ๋‹ค๋Š” ์ ์—์„œ ์˜๋ฏธ๊ฐ€ ํฌ๋‹ค. 2. ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘ ๋ฐ ์ „์ฒ˜๋ฆฌ | ํ•ญ๋ชฉ | ๋‚ด์šฉ | | | | | ๋ฐ์ดํ„ฐ ์†Œ์Šค |

Electrical Engineering and Systems Science Detection
No Image

Secure High-Resolution ISAC via Multi-Layer Intelligent Metasurfaces: A Layered Optimization Framework

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ISAC์€ 5Gยท6G ์‹œ๋Œ€์— ํ†ต์‹ ยท๊ฐ์ง€๋ฅผ ํ†ตํ•ฉํ•จ์œผ๋กœ์จ ์ŠคํŽ™ํŠธ๋Ÿผ ํšจ์œจ๊ณผ ์ธํ”„๋ผ ๋น„์šฉ ์ ˆ๊ฐ์„ ๋ชฉํ‘œ๋กœ ํ•œ๋‹ค. ํ•˜์ง€๋งŒ ๊ฐ์ง€ ํ•ด์ƒ๋„์™€ ํ†ต์‹  ๋ณด์•ˆ ์‚ฌ์ด์˜ ํŠธ๋ ˆ์ด๋“œ์˜คํ”„๊ฐ€ ์‹ฌ๊ฐํ•œ ์ œ์•ฝ์œผ๋กœ ์ž‘์šฉํ•œ๋‹ค. ํŠนํžˆ ๋‹ค์ค‘ ์‚ฌ์šฉ์žยท๋‹ค์ค‘ eavesdropper ํ™˜๊ฒฝ์—์„œ ๊ธฐ์กด ๋‹จ์ผ์ธต RIS๋Š” ํŒŒ๋™ ์กฐ์ž‘ ์ž์œ ๋„๊ฐ€ ๋ถ€์กฑํ•ด ๊ณ ์ •๋ฐ€ ๋น”ํฌ๋ฐ์ด ์–ด๋ ค์› ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ์ ์ธตํ˜• ์ง€๋Šฅํ˜• ๋ฉ”ํƒ€์„œํŽ˜์ด์Šค(SIM) : ๋‹ค์ธต ๊ตฌ์กฐ๋ฅผ ํ†ตํ•ด ๋ถ€ํ”ผ(๋ณผ๋ฅ˜๋ฉ”ํŠธ๋ฆญ) ์•„๋‚ ๋กœ๊ทธ ์—ฐ์‚ฐ์ด ๊ฐ€๋Šฅํ•ด, ๋‹จ์ผ์ธต RIS๊ฐ€ ์ œ๊ณตํ•˜๋Š” ์œ„์ƒ ๋ณ€์กฐ๋ฅผ ๋„˜์–ด ๋ณตํ•ฉ์ ์ธ ๊ณต๊ฐ„ ํ•„ํ„ฐ๋งยท์ดˆ์  ์กฐ์ ˆ์„

Framework Electrical Engineering and Systems Science
No Image

SELEBI: Percussion-aware Time Stretching via Selective Magnitude Spectrogram Compression by Nonstationary Gabor Transform

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์œ„์ƒ ๋ณด์ฝ”๋”(PV) ๋Š” STFT โ†’ ์œ„์ƒ ๋ณ€ํ˜• โ†’ OLA(Overlapโ€‘Add) ์ˆœ์œผ๋กœ ๋™์ž‘ํ•˜๋ฉฐ, ํ†ค์„ฑ ์„ฑ๋ถ„ ์—๋Š” ํšจ๊ณผ์ ์ด์ง€๋งŒ ํผ์ปค์‹œ๋ธŒ(์ถฉ๊ฒฉ์„ฑ) ์„ฑ๋ถ„ ์—๋Š” โ€œํผ์ปค์…˜ ์Šค๋ฏธ์–ด๋งโ€์ด๋ผ๋Š” ์‹ฌ๊ฐํ•œ ์™œ๊ณก์„ ์ผ์œผํ‚จ๋‹ค. ๊ธฐ์กด์˜ ํผ์ปค์…˜โ€‘์ธ์‹ ๋ฐฉ๋ฒ•์€ ํฌ๊ฒŒ ๋‘ ๊ฐ€์ง€๋กœ ๊ตฌ๋ถ„๋œ๋‹ค. Class A : ํ†คโ€‘ํผ์ปค์‹œ๋ธŒ ๋ถ„๋ฆฌ๋ฅผ ์‚ฌ์ „ ์ˆ˜ํ–‰ ํ›„ ๊ฐ๊ฐ ์ฒ˜๋ฆฌ โ†’ ๋ถ„๋ฆฌ ์˜ค๋ฅ˜๊ฐ€ ์ƒˆ๋กœ์šด ์•„ํ‹ฐํŒฉํŠธ๋ฅผ ์œ ๋ฐœ. Class B : magnitudeโ€‘phase ๋ถˆ์ผ์น˜๋ฅผ ์™„ํ™”ํ•˜๋ ค ์‹œ๋„ํ•˜์ง€๋งŒ, ์—ฌ์ „ํžˆ magnitude๊ฐ€ ์‹œ๊ฐ„์ ์œผ๋กœ ํผ์ ธ ์žˆ์–ด ์™„์ „ํ•œ ํ•ด๊ฒฐ์ด ์–ด๋ ค

Audio Processing Electrical Engineering and Systems Science
No Image

Self-Organized Bioelectricity via Collective Pump Alignment: Physical Origin of Chemiosmosis

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์งˆ๋ฌธ ํ™”ํ•™์‚ผํˆฌ์™€ ๋น„ํ‰ํ˜• : ๋ฏธ์ฒผ(Mitchell)์˜ ํ™”ํ•™์‚ผํˆฌ ์ด๋ก ์€ ๋ง‰์„ ํ†ตํ•œ ์ด์˜จ ์ „๋ฅ˜๊ฐ€ ์„ธํฌ ์—๋„ˆ์ง€ ์ „ํ™˜์˜ ๊ทผ์›์ž„์„ ๋ฐํžˆ์ง€๋งŒ, ์ดˆ๊ธฐ ์›์‹œ ์„ธํฌ๊ฐ€ ์ด๋Ÿฌํ•œ ์ „๋ฅ˜๋ฅผ ์–ด๋–ป๊ฒŒ ์Šค์Šค๋กœ ์ƒ์„ฑํ–ˆ๋Š”์ง€๋Š” ๋ฏธํ•ด๊ฒฐ ๋ฌธ์ œ์ด๋‹ค. ์ƒ๋ช… ๊ธฐ์› ์—ฐ๊ตฌ์™€์˜ ์ฐจ๋ณ„์  : ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ž๋™์ด‰๋งค ๋„คํŠธ์›Œํฌ, ๋ฆฌํ”ผ๋“œ ๋ง‰, ๋น„ํ‰ํ˜• ๊ตฌ๋™ ์กฐ๊ฑด ๋“ฑ์— ์ดˆ์ ์„ ๋งž์ท„๋‹ค. ๋ณธ ๋…ผ๋ฌธ์€ ํŽŒํ”„โ€‘์ „์œ„ ์ƒํ˜ธ๊ฐ•ํ™” ๋ผ๋Š” ์ƒˆ๋กœ์šด ๋ฌผ๋ฆฌ์  ํ”ผ๋“œ๋ฐฑ ๋ฃจํ”„๋ฅผ ์ œ์‹œํ•œ๋‹ค. 2. ๋ชจ๋ธ ๊ตฌ์„ฑ | ์š”์†Œ | ๋ฌผ๋ฆฌ์  ์˜๋ฏธ | ์ˆ˜ํ•™์  ํ‘œํ˜„ | | | | | | ํŽŒํ”„ ๋ฐฉํ–ฅ (P i) | ๋‚ด๋ถ€(+) ํ˜น

Physics
No Image

Service Orchestration in the Computing Continuum: Structural Challenges and Vision

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์ปดํ“จํŒ… ์—ฐ์†์ฒด(Continuum) ๊ฐœ๋… : IoT ์„ผ์„œ ๋ฐ์ดํ„ฐ์™€ AR ๊ธฐ๋ฐ˜ ์Šค๋งˆํŠธ ์‹œํ‹ฐ ๋””์ง€ํ„ธ ํŠธ์œˆ ๋“ฑ ์ดˆ์‹ค์‹œ๊ฐ„ ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜์„ ์œ„ํ•ด ์—ฃ์ง€์™€ ํด๋ผ์šฐ๋“œ๊ฐ€ ์œ ๊ธฐ์ ์œผ๋กœ ๊ฒฐํ•ฉ๋œ ์ธํ”„๋ผ. ์˜ค์ผ€์ŠคํŠธ๋ ˆ์ด์…˜ ๋‚œ์ œ : 1. ์ด์งˆ์„ฑ โ€“ ๋””๋ฐ”์ด์Šคยท์ž์› ์‚ฌ์–‘, ๋ฒค๋” ์ •์ฑ…, ๋ฒ•์  ์ œ์•ฝ์ด ์„œ๋กœ ๋‹ค๋ฆ„. 2. ๋™์ ์„ฑ โ€“ ๋””๋ฐ”์ด์Šค์˜ ์ž…ยท์ถœยท์†Œ๋ฉธ, ์„œ๋น„์Šค ์š”๊ตฌ์‚ฌํ•ญ(SLO) ๋ณ€๋™์ด ์‹ค์‹œ๊ฐ„์œผ๋กœ ๋ฐœ์ƒ. 3. ๋ถ€๋ถ„ ๊ด€์ธก โ€“ ๊ฐ ์ฐธ์—ฌ์ž๋Š” ์ „์ฒด ์‹œ์Šคํ…œ ์ƒํƒœ๋ฅผ ์™„์ „ํ•˜๊ฒŒ ์•Œ ์ˆ˜ ์—†์œผ๋ฉฐ, ์ œํ•œ๋œ ๊ด€์ธก๋งŒ์œผ๋กœ ํ˜‘์—…ํ•ด์•ผ ํ•จ. 2. ๊ตฌ์กฐ์  ๊ฐ€์„ค(Hypotheses

Computer Science Distributed Computing
No Image

SIT-LMPC: Safe Information-Theoretic Learning Model Predictive Control for Iterative Tasks

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋ฐ˜๋ณต ์ž‘์—… (quadrotor ๋ ˆ์ด์‹ฑ, ์‚ฌ์กฑ๋ณดํ–‰ ๋กœ๋ด‡, ์ˆ˜์ˆ  ๋กœ๋ด‡ ๋“ฑ)์€ ๋ฐ์ดํ„ฐ ์žฌํ™œ์šฉ ์„ ํ†ตํ•ด ์„ฑ๋Šฅ์„ ์ ์ง„์ ์œผ๋กœ ๊ฐœ์„ ํ•ด์•ผ ํ•˜๋Š”๋ฐ, ๊ธฐ์กด ILCยทLMPC ์ ‘๊ทผ๋ฒ•์€ ์„ ํ˜•ยทํ™•์ •์  ์‹œ์Šคํ…œ์— ๊ตญํ•œ๋˜๊ฑฐ๋‚˜ ๋ณด์ˆ˜์ ์ธ ์ œ์•ฝ ์ฒ˜๋ฆฌ (์˜ˆ: ๊ณ ์ • ๋น„์šฉ, CEM ๊ธฐ๋ฐ˜ ์ƒ˜ํ”Œ๋ง)๋กœ ์ธํ•ด ์‹ค์ œ ๋ณต์žกํ•œ ๋น„์„ ํ˜•ยทํ™•๋ฅ ์  ํ™˜๊ฒฝ์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ต๋‹ค. MPPI ๋Š” ์ƒ˜ํ”Œ๋ง ๊ธฐ๋ฐ˜์œผ๋กœ KLโ€‘๋‹ค์ด๋ฒ„์ „์Šค๋ฅผ ์ตœ์†Œํ™”ํ•ด ๊ฐ•์ธํ•œ ์ œ์–ด๋ฅผ ์ œ๊ณตํ•˜์ง€๋งŒ, ์ œ์•ฝ ์ฒ˜๋ฆฌ ๊ฐ€ ๋ฏธํกํ•˜๊ณ , ์ œ์•ฝ์„ ๋น„์šฉ์— ์ง์ ‘ ์‚ฝ์ž…ํ•˜๋ฉด ๊ณผ๋ณด์ˆ˜ ๊ฐ€ ๋œ๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๊ธฐ์—ฌ ๋‚ด์šฉ |

Robotics Computer Science Learning Model
No Image

Smart Reaction Templating: A Graph-Based Method for Automated Molecular Dynamics Input Generation

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ˆ˜์ž‘์—… ํ…œํ”Œ๋ฆฟ์˜ ํ•œ๊ณ„ : ๊ธฐ์กด LAMMPS REACTION ํŒจํ‚ค์ง€์—์„œ ์‚ฌ์šฉ๋˜๋Š” ํ…œํ”Œ๋ฆฟ์€ ์ „๋ฌธ๊ฐ€๊ฐ€ ์ง์ ‘ ์›์ž ๋งคํ•‘๊ณผ ๋ฐ˜์‘ ๋ถ€์œ„๋ฅผ ์ •์˜ํ•ด์•ผ ํ•˜๋ฉฐ, ํŠนํžˆ ๋Œ€์นญ์„ฑ์ด๋‚˜ ๋™์ผ ์›์ž ๊ตฌ๋ถ„์ด ์–ด๋ ค์šด ์‹œ์Šคํ…œ์—์„œ ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•˜๊ธฐ ์‰ฝ๋‹ค. ๊ณ ์ฒ˜๋ฆฌ๋Ÿ‰ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ์š”๊ตฌ : ์žฌ๋ฃŒ ์„ค๊ณ„ยท๋จธ์‹ ๋Ÿฌ๋‹ ๊ธฐ๋ฐ˜ ์ „์‚ฐ ํ™”ํ•™์—์„œ๋Š” ์ˆ˜์ฒœ~์ˆ˜๋งŒ ๊ฐœ์˜ ๋ฐ˜์‘์„ ๋™์‹œ์— ๋‹ค๋ฃจ์–ด์•ผ ํ•˜๋ฏ€๋กœ ์ž๋™ํ™”๊ฐ€ ํ•„์ˆ˜์ ์ด๋‹ค. 2. ํ•ต์‹ฌ ๋ฐฉ๋ฒ•๋ก  | ๋‹จ๊ณ„ | ์„ค๋ช… | ๊ทธ๋ž˜ํ”„ ์ด๋ก  ์ ์šฉ | | | | | | ๋ถ„์ž โ†’ ๊ทธ๋ž˜ํ”„ ๋ณ€ํ™˜ | ์›์ž๋ฅผ ์ •์ , ๊ฒฐํ•ฉ์„ ๊ฐ„์„ ์œผ๋กœ ๋ณ€ํ™˜. ์›์ž ์ข…๋ฅ˜ยท์ „

No Image

Software Engineering Agents for Embodied Controller Generation : A Study in Minigrid Environments

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ SWEโ€‘Agents ๋Š” ๊ธฐ์กด์— ์ฝ”๋“œ ์ž๋™ ์™„์„ฑ, ๋ฒ„๊ทธ ์ˆ˜์ •, ํ…Œ์ŠคํŠธ ์ƒ์„ฑ ๋“ฑ ์ •ํ˜•ํ™”๋œ ์†Œํ”„ํŠธ์›จ์–ด ์ž‘์—…์— ์ ์šฉ๋ผ ์™”์Œ. Embodied AI ๋Š” ๋กœ๋ด‡ยท์—์ด์ „ํŠธ๊ฐ€ ๋ฌผ๋ฆฌยท๊ฐ€์ƒ ํ™˜๊ฒฝ์—์„œ ํ–‰๋™์„ ์ˆ˜ํ–‰ํ•˜๋„๋ก ์š”๊ตฌ๋˜๋ฉฐ, ๋™์  ํ™˜๊ฒฝ ํƒ์ƒ‰ ๊ณผ ์‹ค์‹œ๊ฐ„ ํ”ผ๋“œ๋ฐฑ ์ด ํ•ต์‹ฌ. ๊ธฐ์กด SWEโ€‘Agents ์—ฐ๊ตฌ๋Š” ์ •์  ์ฝ”๋“œ ์—๋งŒ ์ดˆ์ ์„ ๋งž์ท„๊ธฐ ๋•Œ๋ฌธ์—, ๋™์  ์ •๋ณด(์‹œ๋ฎฌ๋ ˆ์ด์…˜, ์„ผ์„œ ๋ฐ์ดํ„ฐ) ๋ฅผ ํ™œ์šฉํ•˜๋Š” ๋Šฅ๋ ฅ์ด ๊ฒ€์ฆ๋˜์ง€ ์•Š์Œ. ๋”ฐ๋ผ์„œ ๋ณธ ๋…ผ๋ฌธ์€ โ€œ์ •์  vs ๋™์  ์ •๋ณด ์ ‘๊ทผโ€ ์ด๋ผ๋Š” ์ƒˆ๋กœ์šด ์ฐจ์›์„ ๋„์ž…ํ•ด SWEโ€‘Agents์˜ ๋ฒ”์šฉ์„ฑ์„ ์‹œํ—˜ํ•œ๋‹ค

No Image

Space-filling lattice designs for computer experiments

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ maximin / minimax ์„ค๊ณ„ ๋Š” ์ „ํ†ต์ ์œผ๋กœ ์ปดํ“จํ„ฐ ์‹คํ—˜์—์„œ ๋„๋ฆฌ ์‚ฌ์šฉ๋ผ ์™”์œผ๋ฉฐ, ๊ฐ๊ฐ ์  ๊ฐ„ ๊ฑฐ๋ฆฌ ์™€ ์ „์ฒด ์˜์—ญ ์ปค๋ฒ„ ๋ฅผ ์ตœ์ ํ™”ํ•œ๋‹ค. ๋‘ ๋ชฉํ‘œ๋ฅผ ๋™์‹œ์— ๋งŒ์กฑ์‹œํ‚ค๋Š” quasiโ€‘uniformity ๋Š” ์ตœ๊ทผ kernel interpolation , RBF approximation , Gaussian process ๋“ฑ ๋‹ค์–‘ํ•œ ์ˆ˜์น˜ ๋ฐฉ๋ฒ•์—์„œ ํ•ต์‹ฌ ํ’ˆ์งˆ ๊ธฐ์ค€์œผ๋กœ ๋ถ€์ƒํ•˜๊ณ  ์žˆ๋‹ค. ๊ธฐ์กด QMC ์ˆ˜์—ด(์˜ˆ: Sobolโ€™, Halton)์€ lowโ€‘discrepancy ํŠน์„ฑ์€ ๋›ฐ์–ด๋‚˜์ง€๋งŒ, separation radiu

Statistics
No Image

Spanning the Visual Analogy Space with a Weight Basis of LoRAs

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์‹œ๊ฐ์  ์œ ์ถ” ๋Š” โ€œa : aโ€ฒ :: b : bโ€ฒโ€ ํ˜•ํƒœ์˜ ๋ณ€ํ™˜์„ ํ•™์Šตยท์ ์šฉํ•˜๋Š” ๋ฌธ์ œ๋กœ, ํ…์ŠคํŠธ๋กœ ํ‘œํ˜„ํ•˜๊ธฐ ์–ด๋ ค์šด ์Šคํƒ€์ผยทํฌ์ฆˆยท๊ตฌ์กฐ ๋ณ€ํ™˜์„ ์ง์ ‘ ์‹œ์—ฐ์œผ๋กœ ์ง€์ •ํ•  ์ˆ˜ ์žˆ๋‹ค. ๊ธฐ์กด ๋‹จ์ผ LoRA ์ ‘๊ทผ๋ฒ•(์˜ˆ: RelationAdapter, VisualCloze)์€ ํŒŒ๋ผ๋ฏธํ„ฐ ํšจ์œจ์„ฑ์€ ๋†’์ง€๋งŒ, ํ•˜๋‚˜์˜ ์ €์ฐจ์› ์–ด๋Œ‘ํ„ฐ๊ฐ€ ๋ชจ๋“  ๋ณ€ํ™˜์„ ์••์ถ•ํ•˜๋ ค๋‹ค ๋ณด๋‹ˆ ๋ณ€ํ™˜ ๋‹ค์–‘์„ฑ ์„ ์ถฉ๋ถ„ํžˆ ํฌ์ฐฉํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์ตœ๊ทผ ์—ฐ๊ตฌ์—์„œ ๋‹ค์ˆ˜ LoRA ๊ฐ€ ์˜๋ฏธ๋ก ์  ๊ณต๊ฐ„์„ ํ˜•์„ฑํ•˜๊ณ , ์„ ํ˜• ๋ณด๊ฐ„์ด ์ƒˆ๋กœ์šด ๋ณ€ํ™˜์„ ์ƒ์„ฑํ•œ๋‹ค๋Š” ์‚ฌ์‹ค์„ ๋ฐœ๊ฒฌํ–ˆ์œผ๋ฉฐ, ์ด๋ฅผ ์‹œ๊ฐ์  ์œ ์ถ”์—

Computer Science Computer Vision
No Image

Sparse Additive Model Pruning for Order-Based Causal Structure Learning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ˆœ์„œ ๊ธฐ๋ฐ˜ ์ธ๊ณผ ํ•™์Šต ์€ ์œ„์ƒ ์ˆœ์„œ๋ฅผ ๋ฏธ๋ฆฌ ์ถ”์ •ํ•จ์œผ๋กœ์จ DAG ํƒ์ƒ‰ ๊ณต๊ฐ„์„ ์ง€์ˆ˜์ ์œผ๋กœ ์ถ•์†Œํ•œ๋‹ค๋Š” ์žฅ์ ์ด ์žˆ๋‹ค. ํ•˜์ง€๋งŒ ์™„์ „ ์—ฐ๊ฒฐ๋œ DAG์—์„œ ์‹ค์ œ ๋ถ€๋ชจ๋ฅผ ์ฐพ๋Š” pruning ๋‹จ๊ณ„ ๊ฐ€ ์ „์ฒด ํŒŒ์ดํ”„๋ผ์ธ์˜ ๋ณ‘๋ชฉ์ด ๋œ๋‹ค. ๊ธฐ์กด CAMโ€‘pruning์€ GAM ์„ ๊ฐ ๋ณ€์ˆ˜๋งˆ๋‹ค ๋ฐ˜๋ณต ์ ํ•ฉํ•˜๊ณ , ๋‹ค์ค‘ ๊ฐ€์„ค ๊ฒ€์ • ์„ ์ˆ˜ํ–‰ํ•œ๋‹ค. ๊ณ ์ฐจ์›(์ˆ˜์ฒœ ๋ณ€์ˆ˜) ์ƒํ™ฉ์—์„œ ์‹œ๊ฐ„ ๋ณต์žก๋„๊ฐ€ O(dยทnยทk) (d: ๋ณ€์ˆ˜ ์ˆ˜, n: ์ƒ˜ํ”Œ, k: GAM ๋ณต์žก๋„) ์ˆ˜์ค€์œผ๋กœ ๊ธ‰์ฆํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด | ์š”์†Œ | ๊ธฐ์กด ๋ฐฉ๋ฒ• | ์ œ์•ˆ ๋ฐฉ๋ฒ• (SART

Statistics Learning Machine Learning Model
No Image

Spatiotemporal noise stabilizes unbounded diversity in strongly-competitive communities

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ œ๊ธฐ May์˜ ๋‹ค์–‘์„ฑโ€‘์•ˆ์ •์„ฑ ์—ญ์„ค : ๋ฌด์ž‘์œ„ ์ƒํ˜ธ์ž‘์šฉ์„ ๊ฐ–๋Š” ํฐ ์ปค๋ฎค๋‹ˆํ‹ฐ์ผ์ˆ˜๋ก ์•ˆ์ •์ ์ธ ๊ณต์กด์ด ์–ด๋ ค์›Œ์ง„๋‹ค(์ข… ์ˆ˜ โ†‘ โ†’ ์•ˆ์ •์„ฑ โ†“). ๊ธฐ์กด ํ•ด๊ฒฐ์ฑ…์€ ๊ตฌ์กฐํ™”๋œ ์ƒํ˜ธ์ž‘์šฉ ํ–‰๋ ฌ(ํฌ์†Œ์„ฑ, ๋ชจ๋“ˆ์„ฑ ๋“ฑ) ํ˜น์€ ์ •์  ํ™˜๊ฒฝ ๋ณ€๋™ ์— ์˜์กดํ–ˆ์ง€๋งŒ, ๊ฐ•ํ•œ ๊ฒฝ์Ÿ ์ƒํ™ฉ์—์„œ๋Š” ์ถฉ๋ถ„ํžˆ ์„ค๋ช…๋˜์ง€ ์•Š๋Š”๋‹ค. ์‹œ๊ณต๊ฐ„์  ํ™˜๊ฒฝ ๋ณ€๋™ (์˜ˆ: ๊ธฐํ›„, ์˜์–‘์†Œ ๋ณ€๋™)๊ณผ ๊ณต๊ฐ„์  ์ด์งˆ์„ฑ (ํŒจ์น˜ ๊ฐ„ ์ƒํ˜ธ์ž‘์šฉ ์ฐจ์ด)์€ ์ž์—ฐ๊ณ„์—์„œ ํ”ํžˆ ๊ด€์ฐฐ๋˜์ง€๋งŒ, ๋‘ ์š”์†Œ๊ฐ€ ๋™์‹œ์— ์ž‘์šฉํ•  ๋•Œ์˜ ํšจ๊ณผ๋Š” ์•„์ง ์ฒด๊ณ„์ ์œผ๋กœ ์—ฐ๊ตฌ๋˜์ง€ ์•Š์•˜๋‹ค. 2. ๋ชจ๋ธ๋ง ์ ‘๊ทผ ์ผ๋ฐ˜ํ™”๋œ ๋กœํ‹€์นดโ€‘๋ณผํ…Œ๋ผ(

Quantitative Biology
No Image

Spec-Gloss Surfels and Normal-Diffuse Priors for Relightable Glossy Objects

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ด‘ํƒ ๋ฌผ์ฒด ๋ณต์›์˜ ๋‚œ์ œ : ๊ด‘ํƒ ๋ฌผ์ฒด๋Š” ๋ฏธ์„ธํ•œ ๋ฐ˜์‚ฌ์™€ ๋ณต์žกํ•œ BRDF ํŠน์„ฑ ๋•Œ๋ฌธ์— ํ˜•ํƒœ์™€ ์žฌ์งˆ์„ ๋™์‹œ์— ์ถ”์ •ํ•˜๊ธฐ๊ฐ€ ์–ด๋ ต๋‹ค. ๊ธฐ์กด ๋ฐฉ๋ฒ•๋“ค์€ ์ข…์ข… Lambertian + Phong ๊ฐ™์€ ๋‹จ์ˆœ ๋ชจ๋ธ์— ์˜์กดํ•˜๊ฑฐ๋‚˜, diffuseโ€‘specular ๋ฅผ ํ•˜๋‚˜์˜ ํŒŒ๋ผ๋ฏธํ„ฐ๋กœ ๋ฌถ์–ด ํ‘œํ˜„ํ•œ๋‹ค. ์ด๋Š” ์‹ค์ œ ์žฌ์งˆ์˜ ๋ฏธ์„ธํ•œ ๋ณ€ํ™”๋ฅผ ํฌ์ฐฉํ•˜์ง€ ๋ชปํ•œ๋‹ค. Gaussian Splatting : ์ตœ๊ทผ 3D ์žฌ๊ตฌ์„ฑ์—์„œ 2D Gaussian Splatting ์ด ๋น ๋ฅธ ๋ Œ๋”๋ง๊ณผ ๋†’์€ ํ•ด์ƒ๋„ ์žฌ๊ตฌ์„ฑ์— ๊ฐ•์ ์„ ๋ณด์—ฌ์ฃผ์ง€๋งŒ, ๊ธฐ์กด ๊ตฌํ˜„์€ ์ฃผ๋กœ diffu

No Image

Stability and convergence of multi-converter systems using projection-free power-limiting droop control

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ „๋ ฅ ์‹œ์Šคํ…œ์˜ ํŒจ๋Ÿฌ๋‹ค์ž„ ์ „ํ™˜ : ๋™๊ธฐ์‹ ๋ฐœ์ „๊ธฐ โ†’ ์ „๋ ฅ ์ „์ž ๊ธฐ๋ฐ˜ ์žฌ์ƒยท์ €์žฅ ์‹œ์Šคํ…œ. ์ด๋กœ ์ธํ•ด ์ฃผํŒŒ์ˆ˜ยท์ „์•• ๋™์—ญํ•™ ์ด ๋ฐ€๋ฆฌ์ดˆ ์ˆ˜์ค€์œผ๋กœ ๋นจ๋ผ์ง€๊ณ , ์ „๋ ฅยท์ „๋ฅ˜ ์ œํ•œ ์ด๋ผ๋Š” ๋ฌผ๋ฆฌ์  ์ œ์•ฝ์ด ๊ฐ•์กฐ๋œ๋‹ค. ๊ทธ๋ฆฌ๋“œโ€‘ํฌ๋ฐ(gridโ€‘forming) ์ปจ๋ฒ„ํ„ฐ ๋Š” ๋ฏธ๋ž˜ ์ „๋ ฅ๋ง์˜ ํ•ต์‹ฌ์œผ๋กœ ๊ธฐ๋Œ€๋˜์ง€๋งŒ, ๊ธฐ์กด ๋“œ๋กญ, VSM, dVOC ๋“ฑ์€ ์ œ์•ฝ(์ „๋ ฅ ํ•œ๊ณ„) ์„ ๊ณ ๋ คํ•˜์ง€ ์•Š์€ ์ฑ„ ์•ˆ์ •์„ฑ ๋ถ„์„์ด ์ง„ํ–‰๋ผ ์™”๋‹ค. ์ „๋ ฅ ์ œํ•œ ๋“œ๋กญ ์ œ์–ด ๋Š” ๊ธฐ์กด์— ํ”„๋กœ์ ์…˜ ๊ธฐ๋ฐ˜ (projectionโ€‘based)์œผ๋กœ ๊ตฌํ˜„๋ผ, ์—ฐ์†์ ์ธ ๋ฏธ๋ถ„ ๋ฐฉ์ •์‹์ด ์•„๋‹Œ ํˆฌ์‚ฌ

System Electrical Engineering and Systems Science
No Image

State Feedback Control of State-Delayed LPV Systems using Dynamics IQCs

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์‹œ๊ฐ„ ์ง€์—ฐ ์‹œ์Šคํ…œ ์€ ์ œ์กฐ ๊ณต์ •, ์‹ ๊ฒฝ๋ง, ํผ์ง€ ๋™์—ญํ•™ ๋“ฑ ๋‹ค์–‘ํ•œ ๋ถ„์•ผ์—์„œ ํ•„์—ฐ์ ์œผ๋กœ ๋‚˜ํƒ€๋‚˜๋ฉฐ, ์ง€์—ฐ์ด ์‹œ์Šคํ…œ์˜ ๋ถˆ์•ˆ์ •์„ฑ๊ณผ ์„ฑ๋Šฅ ์ €ํ•˜๋ฅผ ์ดˆ๋ž˜ํ•œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ •์ (์ƒ์ˆ˜) ์ง€์—ฐ ์— ์ดˆ์ ์„ ๋งž์ถ”์—ˆ์œผ๋ฉฐ, ์‹œ๊ฐ„โ€‘๊ฐ€๋ณ€ ์ง€์—ฐ ์— ๋Œ€ํ•œ ๋ถ„์„ยทํ•ฉ์„ฑ์€ ์•„์ง ๋ณด์ˆ˜์„ฑ์ด ํฌ๊ณ  ๋น„๋ณผ๋ก(BMI) ํ˜•ํƒœ์˜ ์กฐ๊ฑด์ด ๋งŽ์•„ ์‹ค์šฉ์  ์ ์šฉ์ด ์–ด๋ ค์› ๋‹ค. LPV(Linear Parameterโ€‘Varying) ์‹œ์Šคํ…œ ์€ ํŒŒ๋ผ๋ฏธํ„ฐ ๋ณ€๋™์„ ๋ช…์‹œ์ ์œผ๋กœ ๋ชจ๋ธ๋งํ•จ์œผ๋กœ์จ ๋„“์€ ๋ฒ”์œ„์˜ ๋น„์„ ํ˜•ยท์‹œ์‹œ์„ฑ ์‹œ์Šคํ…œ์„ ํฌ๊ด„ํ•˜์ง€๋งŒ, ์ง€์—ฐ๊ณผ ํŒŒ๋ผ๋ฏธํ„ฐ ๋ณ€๋™์ด ๋™์‹œ์— ์กด์žฌํ• 

System Electrical Engineering and Systems Science
No Image

STRAND: Sequence-Conditioned Transport for Single-Cell Perturbations

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์œ ์ „์žโ€‘๋ ˆ๋ฒจ ๊ต๋ž€ ๋ชจ๋ธ์˜ ํ•œ๊ณ„ : ๋™์ผ ์œ ์ „์ž๋ฅผ ๋Œ€์ƒ์œผ๋กœ ํ•˜๋Š” ๋‹ค์–‘ํ•œ CRISPRโ€‘i/a, ๋ฒ ์ด์Šค ํŽธ์ง‘ ๋“ฑ์€ ์ •ํ™•ํ•œ ์—ผ๊ธฐ ์ขŒ์œ„๋ฅผ ์กฐ์ž‘ํ•˜์ง€๋งŒ, ๊ธฐ์กด ๋ชจ๋ธ์€ ์ด๋ฅผ ๋ชจ๋‘ ๋™์ผํ•œ โ€˜์œ ์ „์žโ€™ ์‹๋ณ„์ž๋กœ ๋งคํ•‘ํ•œ๋‹ค. ์ด๋กœ ์ธํ•ด (โ‘ ) ๋™์ผ ์œ ์ „์ž์˜ ์„œ๋กœ ๋‹ค๋ฅธ ์กฐ์ ˆ ์š”์†Œ(๊ฐ•ํ™”์ž, ๋Œ€์ฒด TSS)์—์„œ ๋ฐœ์ƒํ•˜๋Š” ์ „์‚ฌ ์ฐจ์ด๋ฅผ ํฌ์ฐฉํ•˜์ง€ ๋ชปํ•˜๊ณ , (โ‘ก) ๋น„์ฝ”๋”ฉ ์˜์—ญ(์ „์ฒด ์œ ์ „์ฒด์˜ 98 %)์— ๋Œ€ํ•œ ์˜ˆ์ธก์ด ๋ถˆ๊ฐ€๋Šฅํ•˜๋‹ค. ์‹œํ€€์Šคโ€‘๋ ˆ๋ฒจ ์˜ˆ์ธก์˜ ํ•„์š”์„ฑ : ์‹ค์ œ ๊ต๋ž€ ํšจ๊ณผ๋Š” ์ˆ˜๋ฐฑ ๊ฐœโ€‘์ˆ˜์ฒœ ๊ฐœ ์—ผ๊ธฐ ์„œ์—ด์— ์˜ํ•ด ๋น„์„ ํ˜•์ ์œผ๋กœ ๊ฒฐ์ •๋˜๋ฉฐ, ์„ธํฌ

Quantitative Biology
No Image

StrokeNeXt: A Siamese-encoder Approach for Brain Stroke Classification in Computed Tomography Imagery

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‡Œ์กธ์ค‘ ์€ ๊ธ‰์„ฑยท๊ณ ์œ„ํ—˜ ์งˆํ™˜์œผ๋กœ ์กฐ๊ธฐ ์ง„๋‹จ์ด ์น˜๋ฃŒ ์„ฑํŒจ๋ฅผ ์ขŒ์šฐํ•œ๋‹ค. CT๋Š” ์‘๊ธ‰์‹ค์—์„œ ๊ฐ€์žฅ ํ”ํžˆ ์‚ฌ์šฉ๋˜๋Š” ์˜์ƒ modality์ด์ง€๋งŒ, ์ „๋ฌธ ๋ฐฉ์‚ฌ์„ ๊ณผ ์˜์‚ฌ์˜ ํŒ๋… ์€ ์‹œ๊ฐ„๊ณผ ์ธ๋ ฅ ๋น„์šฉ์ด ๋งŽ์ด ๋“ ๋‹ค. ๊ธฐ์กด AI ๊ธฐ๋ฐ˜ ์ ‘๊ทผ๋ฒ•์€ ์„ฑ๋Šฅโ€‘ํšจ์œจ์„ฑ ํŠธ๋ ˆ์ด๋“œ์˜คํ”„ (๋†’์€ ์ •ํ™•๋„ vs. ๋†’์€ ์—ฐ์‚ฐ ๋น„์šฉ) ๋ฌธ์ œ๋ฅผ ์•ˆ๊ณ  ์žˆ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ๋“€์–ผโ€‘๋ธŒ๋žœ์น˜ ConvNeXt | ์š”์†Œ | ์„ค๋ช… | ์žฅ์  | | | | | | ๋‘ ๊ฐœ์˜ ๋…๋ฆฝ ConvNeXt ์ธ์ฝ”๋” | ๋™์ผ ์ด๋ฏธ์ง€ โ†’ ํŒŒ๋ผ๋ฏธํ„ฐ ๋น„๊ณต์œ  โ†’ ์„œ๋กœ ๋‹ค๋ฅธ ์ตœ์ ํ™” ๊ฒฝ๋กœ |

Image Processing Electrical Engineering and Systems Science
No Image

Structural barriers of the discrete Hasimoto map applied to protein backbone geometry

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๊ธฐํ•˜ํ•™์  ์ ‘๊ทผ : ํ”„๋ ˆ๋„ท ํ”„๋ ˆ์ž„์„ ์ด์šฉํ•ด ๊ณก๋ฅ  (kappa)์™€ ๋น„ํ‹€๋ฆผ (tau)๋ฅผ ๋‚ด๋ถ€ ์ขŒํ‘œ๋กœ ์‚ผ๋Š” ๋ฐฉ๋ฒ•์€ 3์ฐจ์› ๊ณจ๊ฒฉ์„ 2๊ฐœ์˜ ์‹ค์ˆ˜ ํ•„๋“œ๋กœ ์™„์ „ํžˆ ๊ธฐ์ˆ ํ•œ๋‹ค๋Š” ์ ์—์„œ ๋งค๋ ฅ์ ์ด๋‹ค. ํ•˜์‹œ๋ชจํ†  ๋ณ€ํ™˜์˜ ์„ฑ๊ณต ์‚ฌ๋ก€ : ์œ ์ฒด์—ญํ•™์—์„œ ์™€๋ฅ˜์„ (vortex filament)์˜ ๋กœ์ปฌ ์ธ๋•์…˜ ๊ทผ์‚ฌ(LIA)๋ฅผ NLS๋กœ ๋ณ€ํ™˜ํ•ด ์™„์ „ ์ ๋ถ„์„ฑ์„ ํ™•๋ณดํ•œ ๊ฒƒ์ด ์˜๊ฐ์ด ๋˜์—ˆ์œผ๋ฉฐ, ์ด๋ฅผ ๋‹จ๋ฐฑ์งˆ ๊ณจ๊ฒฉ์— ์ ์šฉํ•˜๋ ค๋Š” ์‹œ๋„๊ฐ€ ์—ฌ๋Ÿฌ ์ฐจ๋ก€ ์ด์–ด์กŒ๋‹ค. ์˜ˆ์ธก vs. ์„œ์ˆ  : ๊ธฐ์กด ์—ฐ๊ตฌ๋“ค์€ DNLS์™€ ์†”๋ฆฌํ†ค์„ ์ด์šฉํ•ด ์ด๋ฏธ ์•Œ๋ ค์ง„ ๊ตฌ์กฐ๋ฅผ ์žฌํ˜„ํ•˜๋Š”

Quantitative Biology
No Image

Structural coarse-graining enables noise-robust functional connectivity and reveals hidden inter-subject variability

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ƒ˜ํ”Œ๋ง ์ œ์•ฝ : fMRI๋Š” ์ผ๋ฐ˜์ ์œผ๋กœ 5โ€“15 ๋ถ„ ์ •๋„์˜ ์Šค์บ” ์‹œ๊ฐ„์— ์ˆ˜๋ฐฑ ๊ฐœ์˜ ๋‡Œ ์˜์—ญ(๋˜๋Š” ์ˆ˜์ฒœ ๊ฐœ์˜ voxel)์—์„œ ์ˆ˜๋ฐฑ ๊ฐœ์˜ ์‹œ๊ฐ„์ ๋งŒ์„ ์ œ๊ณตํ•œ๋‹ค. ์ด๋•Œ (T ll N)์ด๋ฉด ์ƒ๊ด€ ํ–‰๋ ฌ์˜ ๊ณ ์œ ๊ฐ’ ์ŠคํŽ™ํŠธ๋Ÿผ์ด Marchenkoโ€‘Pastur(MP) ์žก์Œ ๊ฒฝ๊ณ„์™€ ํฌ๊ฒŒ ๊ฒน์ณ ์‹ค์ œ ์‹ ํ˜ธ๋ฅผ ๊ตฌ๋ถ„ํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ๊ธฐ์กด ํ•ด๊ฒฐ์ฑ…์˜ ํ•œ๊ณ„ : ์ž„๊ณ„๊ฐ’(threshold) ์ ์šฉ, ์ŠคํŒŒ์Šค์„ฑ ๊ฐ€์ •, ์ •๊ทœํ™” ๋“ฑ์€ ์ž„์˜์˜ ๊ฐ€์ •์„ ๋„์ž…ํ•˜๊ฑฐ๋‚˜ ๊ณ„์‚ฐ ๋น„์šฉ์ด ํฌ๊ฒŒ ์ฆ๊ฐ€ํ•œ๋‹ค. ํŠนํžˆ ์ „์—ญ ์‹ ํ˜ธ ํšŒ๊ท€(global signal regression

Condensed Matter
No Image

Structural grouping of extreme value models via graph fused lasso

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ทน๊ฐ’ ๋ถ„์„(EVT) ์€ ํ‰๊ท ยท์ค‘์‹ฌ๊ฐ’๋ณด๋‹ค ๊ผฌ๋ฆฌ(๊ทน๋‹จ๊ฐ’) ํŠน์„ฑ์— ์ดˆ์ ์„ ๋งž์ถ”์–ด ์œ„ํ—˜ ํ‰๊ฐ€, ๊ธฐํ›„ ๋ณ€๋™์„ฑ ๋“ฑ์—์„œ ํ•„์ˆ˜์ ์ด๋‹ค. ์ผ๋ฐ˜ํ™” ํŒŒ๋ ˆํ†  ๋ถ„ํฌ(GPD) ๋Š” ์ดˆ๊ณผ๊ฐ’(Threshold exceedance) ๋ชจ๋ธ๋ง์— ๋„๋ฆฌ ์“ฐ์ด๋ฉฐ, ํŠนํžˆ ํ˜•ํƒœ ํŒŒ๋ผ๋ฏธํ„ฐ(ฮณ) ๊ฐ€ ๊ผฌ๋ฆฌ ๋‘๊ป˜์™€ ๊ทน๊ฐ’ ์ง€์ˆ˜(extremeโ€‘value index)๋ฅผ ๊ฒฐ์ •ํ•œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ์—์„œ๋Š” ํด๋Ÿฌ์Šคํ„ฐ๋ณ„ (์˜ˆ: ๊ธฐํ›„ ๊ด€์ธก์†Œ) GPD๋ฅผ ๋…๋ฆฝ์ ์œผ๋กœ ์ถ”์ •ํ•˜๊ฑฐ๋‚˜, ์ง€์—ญ ๋นˆ๋„ ๋ถ„์„(regional frequency analysis) ๊ณผ ๊ฐ™์ด ์‚ฌ์ „ ์ •์˜๋œ ๊ตฐ์ง‘์— ๋ฐ์ดํ„ฐ๋ฅผ

Statistics Model
No Image

Supershear-subshear-supershear rupture sequence during the 2025 Mandalay Earthquake in Myanmar

1. ์—ฐ๊ตฌ ๋ฐฉ๋ฒ•๋ก ์˜ ๊ฐ•์  ๋‹ค์ค‘ ๋ฐ์ดํ„ฐ ์œตํ•ฉ : ์˜์ƒ, ๊ฐ•์ง„ ๊ธฐ๋ก, ์ˆ˜์น˜ ์‹œ๋ฎฌ๋ ˆ์ด์…˜, ์œ„์„ฑ ์˜์ƒ์„ ๋™์‹œ์— ํ™œ์šฉํ•จ์œผ๋กœ์จ ํŒŒ์—ด ์†๋„ ๋ณ€ํ™”๋ฅผ ๋‹ค๊ฐ๋„๋กœ ๊ฒ€์ฆํ•˜์˜€๋‹ค. ํŠนํžˆ ์˜์ƒ์—์„œ ์ง์ ‘ SํŒŒ๋ฅผ ๊ด€์ธกํ•œ ์ ์€ ๊ธฐ์กด ์ง€์ง„ํ•™ ์—ฐ๊ตฌ์—์„œ๋Š” ๋“œ๋ฌผ์–ด, ํŒŒ์—ด ์†๋„ ์ถ”์ •์— ์ƒˆ๋กœ์šด ๊ทผ๊ฑฐ๋ฅผ ์ œ๊ณตํ•œ๋‹ค. ๋™์—ญํ•™ยท์šด๋™ํ•™ ๋ชจ๋ธ๋ง : ์˜์ƒ์—์„œ ์ถ”์ถœํ•œ ๋‹จ์ธต๋ฒ•์„  ๊ฐ€์†๋„์™€ kinematic rupture ๋ชจ๋ธ์„ ๋น„๊ตํ•จ์œผ๋กœ์จ ํŒŒ์—ด ์†๋„ ๋ณ€ํ™”๋ฅผ ๋ฌผ๋ฆฌ์ ์œผ๋กœ ์„ค๋ช…ํ•œ๋‹ค. ์ด๋Š” ํŒŒ์—ด ์†๋„์™€ ์ „์œ„ ๋ถ„ํฌ ์‚ฌ์ด์˜ ์ธ๊ณผ๊ด€๊ณ„๋ฅผ ์ •๋Ÿ‰ํ™”ํ•˜๋Š” ๋ฐ ์œ ์šฉํ•˜๋‹ค. 2. ์ดˆ์Œ์†โ€‘์•„์Œ์† ์ „์ด ๋ฉ”์ปค๋‹ˆ์ฆ˜ ์‘๋ ฅ ๊ฐ•ํ•˜

No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๋…ธ์ถœ ๋น„์œจ(Prevalence) ์€ ์ •์ฑ…ยท์ถ”์ฒœ ์‹œ์Šคํ…œ์—์„œ โ€œํŠน์ • ์นดํ…Œ๊ณ ๋ฆฌ(์˜ˆ: ์Œ์‹ ๋ ˆ์‹œํ”ผ, AIโ€‘์ƒ์„ฑ ์ฝ˜ํ…์ธ )โ€๊ฐ€ ์ „์ฒด ํŠธ๋ž˜ํ”ฝ์— ์ฐจ์ง€ํ•˜๋Š” ๋น„์ค‘์„ ์˜๋ฏธํ•œ๋‹ค. ๋Œ€๊ทœ๋ชจ ํ”Œ๋žซํผ์—์„œ๋Š” ์ˆ˜์ฒœ ๊ฐœ์˜ ๋™์‹œ ์‹คํ—˜์ด ์ง„ํ–‰๋˜๋ฉฐ, ๊ฐ ์‹คํ—˜ยท์„ธ๊ทธ๋จผํŠธ๋งˆ๋‹ค ๋ฏธ์„ธํ•œ(โ‰ค1 %) ๋ณ€ํ™” ๋ฅผ ๊ฐ์ง€ํ•ด์•ผ ํ•œ๋‹ค. ๊ธฐ์กด LLM ๋ผ๋ฒจ๋ง์€ ๊ณ ์ •๋ฐ€ยท๊ณ ๋น„์šฉ ์ด๋ฉฐ, ๋งค ์‹คํ—˜๋งˆ๋‹ค ๋ผ๋ฒจ๋ง ํŒŒ์ดํ”„๋ผ์ธ์„ ์‹คํ–‰ํ•˜๋ฉด ๋น„์šฉยท์ง€์—ฐ ์ด ๊ธ‰์ฆํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ์„œ๋Ÿฌ๊ฒŒ์ดํŠธ ๊ธฐ๋ฐ˜ ์ธก์ • ํ”„๋ ˆ์ž„์›Œํฌ | ๋‹จ๊ณ„ | ์„ค๋ช… | | | | | ์˜คํ”„๋ผ์ธ ๋ผ๋ฒจ๋ง | PPSWOR

Statistics Applications
No Image

Synthetic-Powered Multiple Testing with FDR Control

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๋‹ค์ค‘ ๊ฒ€์ • ์€ ์œ ์ „์ฒดยท์‹ ์•ฝ ์Šคํฌ๋ฆฌ๋‹ยท์ด์ƒ์น˜ ํƒ์ง€ ๋“ฑ์—์„œ ์ˆ˜์ฒœ~์ˆ˜๋ฐฑ๋งŒ ๊ฐœ ๊ฐ€์„ค์„ ๋™์‹œ์— ๊ฒ€์ •ํ•œ๋‹ค. FDR (Benjaminiโ€‘Hochberg, 1995)๋Š” ์˜ค๋ฅ˜ ์ œ์–ด์™€ ๊ฒ€์ •๋ ฅ ์‚ฌ์ด์˜ ์ข‹์€ ์ ˆ์ถฉ์ ์œผ๋กœ ๋„๋ฆฌ ์“ฐ์ธ๋‹ค. ์‹ค์ œ ๋ฐ์ดํ„ฐ๋Š” ํ‘œ๋ณธ์ด ์ œํ•œ ์ ์ด๋ฉฐ, ๋ฐ˜๋ฉด ํ•ฉ์„ฑ ๋ฐ์ดํ„ฐ (๊ณผ๊ฑฐ ์‹คํ—˜, ์ž๋™ ๋ผ๋ฒจ๋ง, ์ƒ์„ฑ ๋ชจ๋ธ ๋“ฑ)๋Š” ์–‘์ด ํ’๋ถ€ํ•˜์ง€๋งŒ ๋ถ„ํฌ ๋ถˆํ™•์‹ค ํ•˜๊ณ  ํŽธํ–ฅ ๊ฐ€๋Šฅ ์„ฑ์ด ์žˆ๋‹ค. ๊ธฐ์กด ๋ฐฉ๋ฒ•์€ (i) ํ•ฉ์„ฑ ๋ฐ์ดํ„ฐ๋ฅผ ์ „ํ˜€ ์‚ฌ์šฉํ•˜์ง€ ์•Š์•„ ๊ฒ€์ •๋ ฅ์ด ๋‚ฎ๊ฑฐ๋‚˜, (ii) ๊ทธ๋Œ€๋กœ ํ’€์–ด ์‚ฌ์šฉํ•ด FDR์ด ํญ๋ฐœํ•˜๋Š” ๋‘ ๊ทน๋‹จ์— ๋จธ๋ฌธ๋‹ค.

Statistics
No Image

TAVAE: A VAE with Adaptable Priors Explains Contextual Modulation in the Visual Cortex

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ํ™•๋ฅ ์  ์ธ์ง€ ํ”„๋ ˆ์ž„์›Œํฌ : ๋‡Œ๊ฐ€ ํ™˜๊ฒฝ์„ ๋ชจ๋ธ๋งํ•˜๊ณ  ์‚ฌ์ „โ€‘ํ›„๋ฐฉ ์ถ”๋ก ์„ ์ˆ˜ํ–‰ํ•œ๋‹ค๋Š” ๊ฐ€์„ค์€ ์˜ค๋ž˜์ „๋ถ€ํ„ฐ ์ œ์‹œ๋ผ ์™”์œผ๋ฉฐ, V1์ด ๋‹จ์ˆœ ์„ ํ˜• ์‚ฌ์ „(์˜ˆ: ๊ฐ€๋ณด ํ•„ํ„ฐ)๊ณผ ์ƒ์œ„ ํ”ผ์งˆ์˜ ๋งฅ๋ฝ์  ์‚ฌ์ „ ์‚ฌ์ด์˜ ์—ฐ๊ฒฐ ๊ณ ๋ฆฌ ์—ญํ• ์„ ํ•œ๋‹ค๋Š” ์ฆ๊ฑฐ๊ฐ€ ๋ˆ„์ ๋ผ ์žˆ๋‹ค. ์ž‘์—…โ€‘ํŠน์ด์  ์‚ฌ์ „์˜ ๋ฏธ์ง€์  : ์ž์—ฐ ์ด๋ฏธ์ง€ ํ†ต๊ณ„์— ๊ธฐ๋ฐ˜ํ•œ ์‚ฌ์ „์€ ์ž˜ ์•Œ๋ ค์กŒ์ง€๋งŒ, ์ƒˆ๋กœ์šด ํ–‰๋™ ๊ณผ์ œ(์˜ˆ: ๋ฐฉํ–ฅ ๊ตฌ๋ถ„)์—์„œ V1์ด ์–ด๋–ป๊ฒŒ ์ž‘์—…โ€‘ํŠน์ด์  ์‚ฌ์ „์„ ํ˜•์„ฑํ•˜๊ณ  ํ™œ์šฉํ•˜๋Š”์ง€๋Š” ์•„์ง ์‹คํ—˜์ ์œผ๋กœ ๊ฒ€์ฆ๋˜์ง€ ์•Š์•˜๋‹ค. 2. ํ•ต์‹ฌ ๋ฐฉ๋ฒ•๋ก  | ๋‹จ๊ณ„ | ๋‚ด์šฉ | ์˜์˜ | | | | | | (

Quantitative Biology
No Image

Tensor Polarizability of the Nucleus and Angular Mixing in Muonic Deuterium

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ ๋ฎค์˜จ ์›์žยท์ด์˜จ์€ ์ „์ž๊ณ„๋ณด๋‹ค ๋ณด์–ด ๋ฐ˜๊ฒฝ์ด ์ž‘์•„ ํ•ต ๊ตฌ์กฐยท๋™์—ญํ•™ ํšจ๊ณผ๊ฐ€ ํฌ๊ฒŒ ์ฆํญ๋œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ์—์„œ๋Š” ์ „๊ธฐ(์Šค์นผ๋ผ) ํŽธ๊ทน์ด ์ฃผ๋œ ์ดˆ์ ์ด์—ˆ์œผ๋‚˜, ์Šคํ•€ > ยฝ์ธ ํ•ต(์˜ˆ: ์ค‘์ˆ˜์†Œ)์—์„œ๋Š” ํ…์„œ ํŽธ๊ทน์ด ์กด์žฌํ•œ๋‹ค๋Š” ์ ์„ ๊ฐ„๊ณผํ–ˆ๋‹ค. ํ…์„œ ํŽธ๊ทน์€ ํ•ต์˜ 2์ฐจ ์ „๊ธฐ ๋‹ค์ค‘๊ทน์ž์™€ ์ง์ ‘ ์—ฐ๊ฒฐ๋˜๋ฉฐ, ์ด๋Š” ํ•ต ์Šคํ•€๊ณผ ๊ถค๋„ ๊ฐ์šด๋™๋Ÿ‰ ์‚ฌ์ด์˜ ๋น„๋Œ€์นญ ๊ฒฐํ•ฉ์„ ์•ผ๊ธฐํ•œ๋‹ค. ๋”ฐ๋ผ์„œ ํ•˜์ดํผํŒŒ์ธ ๊ตฌ์กฐ์˜ ๋ฏธ์„ธํ•œ ๋ถ„ํ• ์„ ์ดํ•ดํ•˜๋Š” ๋ฐ ํ•„์ˆ˜์ ์ธ ์ƒˆ๋กœ์šด ๋ฌผ๋ฆฌ๋Ÿ‰์ด๋‹ค. 2. ํ•ต์‹ฌ ์ด๋ก ์  ์ „๊ฐœ ์ผ๋ฐ˜์‹ ๋„์ถœ : ์ €์ž๋Š” ๋‘ ์ž…์ž(๋ฎค์˜จ + ํ•ต) ์‹œ์Šคํ…œ์— ๋Œ€ํ•ด ์ „์ž๊ธฐ์ 

Physics
No Image

Terminalizations of quotients of Fano varieties of lines on cubic fourfolds

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ IHS ๋‹ค์–‘์ฒด ๋Š” ์นผ๋ผ๋น„โ€“์•ผ์šฐ(Kโ€‘trivial) ๋‹ค์–‘์ฒด ๋ถ„๋ฅ˜์—์„œ ํ•ต์‹ฌ์ ์ธ ์—ญํ• ์„ ํ•˜๋ฉฐ, ํ˜„์žฌ๊นŒ์ง€ ์•Œ๋ ค์ง„ ๋ถ€๋“œ๋Ÿฌ์šด ์‚ฌ๋ก€๋Š” K3 ํ‘œ๋ฉด์˜ ํž๋ฒ ๋ฅดํŠธ ์Šคํ‚ด, ์ผ๋ฐ˜ํ™”๋œ Kummer ๋‹ค์–‘์ฒด, ๊ทธ๋ฆฌ๊ณ  Oโ€™Grady์˜ 6ยท10 ์ฐจ์› ์˜ˆ์™ธ๋“ค๋ฟ์ด๋‹ค. ํŠน์ด์ ์ด ์žˆ๋Š” ๊ฒฝ์šฐ (ํ„ฐ๋ฏธ๋„ํ™”, Qโ€‘Gorenstein ๋“ฑ)์—๋Š” ๋ณด๋‹ค ํ’๋ถ€ํ•œ ์˜ˆ์‹œ๊ฐ€ ์กด์žฌํ•  ๊ฐ€๋Šฅ์„ฑ์ด ํฌ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ(

Mathematics
No Image

The geometry of online conversations and the causal antecedents of conflictual discourse

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ ๋””์ง€ํ„ธ ๊ณต๋ก ์žฅ์˜ ๊ฐˆ๋“ฑ : ๊ธฐํ›„๋ณ€ํ™”์™€ ๊ฐ™์€ ๊ณผํ•™ยท์ •์น˜์  ์ด์Šˆ๋Š” ์˜จ๋ผ์ธ ํ† ๋ก ์˜ ํ•ต์‹ฌ ์ฃผ์ œ์ด๋ฉฐ, ๊ฐˆ๋“ฑ์€ ๋‹จ์ˆœ ์˜๊ฒฌ ์ฐจ์ด๋ฅผ ๋„˜์–ด ํ‘œํ˜„ ๋ฐฉ์‹(ํ†ค, ์ •์ค‘ํ•จ, ๊ฐ์ •)๊นŒ์ง€ ํฌํ•จํ•œ๋‹ค๋Š” ์ ์„ ๊ธฐ์กด ๋ฌธํ—Œ์ด ๊ฐ•์กฐํ•œ๋‹ค. ๋Œ€ํ™” ๊ธฐํ•˜ํ•™(Geometry of Conversation) : ์ €์ž๋“ค์€ โ€œ์‹œ๊ฐ„ยท์ˆœ์„œยท๊ตฌ์กฐโ€๋ผ๋Š” ์‚ผ์ค‘์ถ•์„ โ€˜๋Œ€ํ™” ๊ธฐํ•˜ํ•™โ€™์ด๋ผ ์ •์˜ํ•˜๊ณ , ์ด ๊ตฌ์กฐ๊ฐ€ ๊ฐˆ๋“ฑ ์–ธ์–ด์˜ ๋ฐœ์ƒยท์ „ํŒŒ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์‹ค์ฆ์ ์œผ๋กœ ๊ฒ€์ฆํ•œ๋‹ค๋Š” ์ ์—์„œ ๋…์ฐฝ์ ์ด๋‹ค. 2. ๋ฐ์ดํ„ฐ ๋ฐ ์ฃผ์„ ๋ฐฉ๋ฒ• | ์š”์†Œ | ๋‚ด์šฉ | | | | | ๋ฐ์ดํ„ฐ | Internet

Computer Science Social Networks
No Image

The Implicit Bias of Adam and Muon on Smooth Homogeneous Neural Networks

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์•”๋ฌต์  ํŽธํ–ฅ ์€ โ€œ์™œ ๊ณผ๋Œ€ ํŒŒ๋ผ๋ฏธํ„ฐํ™”๋œ ๋ชจ๋ธ์ด ์ผ๋ฐ˜ํ™”์— ์„ฑ๊ณตํ•˜๋Š”๊ฐ€?โ€๋ผ๋Š” ์งˆ๋ฌธ์— ๋Œ€ํ•œ ํ•ต์‹ฌ ์ด๋ก ์  ์„ค๋ช…์ด๋‹ค. ๊ธฐ์กด์—๋Š” ๊ฒฝ์‚ฌํ•˜๊ฐ•๋ฒ• ์ด โ„“โ‚‚โ€‘๋งˆ์ง„์„ ์ตœ๋Œ€ํ™”ํ•œ๋‹ค๋Š” ๊ฒฐ๊ณผ๊ฐ€ ์ฃผ๋ฅ˜์˜€์œผ๋ฉฐ, ์ด๋Š” ์„ ํ˜•ยท๋‹จ์ธต ๋„คํŠธ์›Œํฌ์— ๊ตญํ•œ๋œ ๊ฒฝ์šฐ๊ฐ€ ๋งŽ์•˜๋‹ค. ํ˜„๋Œ€ ๋”ฅ๋Ÿฌ๋‹์—์„œ๋Š” Adam , AdamW , Muon ๋“ฑ ๋ชจ๋ฉ˜ํ…€ ๊ธฐ๋ฐ˜ ์˜ตํ‹ฐ๋งˆ์ด์ €๊ฐ€ ์‚ฌ์‹ค์ƒ ํ‘œ์ค€์ด ๋˜์—ˆ๋‹ค. ์ด๋“ค์˜ ๋™์ž‘ ๋ฉ”์ปค๋‹ˆ์ฆ˜์ด ๊ธฐ์กด ์ด๋ก ๊ณผ ์–ผ๋งˆ๋‚˜ ์ผ์น˜ํ•˜๋Š”์ง€๋Š” ์•„์ง ์ถฉ๋ถ„ํžˆ ๋ฐํ˜€์ง€์ง€ ์•Š์•˜๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๋‚ด์šฉ | ์˜๋ฏธ | | | | | | โ‘  | ์ •๊ทœํ™”๋œ ๊ธ‰๊ฒฝ์‚ฌ ํ•˜๊ฐ•(

Computer Science Network Machine Learning
No Image

The Information Geometry of Softmax: Probing and Steering

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์„ ํ˜• ํ‘œํ˜„ ๊ฐ€์„ค ์€ ๋งŽ์€ ํ•ด์„ยท์กฐ์ž‘ ์—ฐ๊ตฌ์˜ ๊ธฐ๋ฐ˜์ด์ง€๋งŒ, ๋Œ€๋ถ€๋ถ„ ์œ ํด๋ฆฌ๋“œ(ํ‰๋ฉด) ๊ธฐํ•˜ ๋ฅผ ์ „์ œํ•œ๋‹ค. ์‹ค์ œ๋กœ ์†Œํ”„ํŠธ๋งฅ์Šค์™€ ๊ฐ™์€ ํ™•๋ฅ  ๋ณ€ํ™˜์„ ๊ฑฐ์นœ ํ‘œํ˜„์€ ๋น„์œ ํด๋ฆฌ๋“œ ๊ตฌ์กฐ๋ฅผ ๊ฐ€์งˆ ๊ฐ€๋Šฅ์„ฑ์ด ํฌ๋‹ค. ์ €์ž๋Š” โ€œ๋‘ ํ‘œํ˜„์ด ๋น„์Šทํ•˜๋‹ค๋Š” ๊ฒƒ์€ ๊ทธ๋“ค์ด ๋งŒ๋“ค์–ด๋‚ด๋Š” ์ถœ๋ ฅ ํ™•๋ฅ ๋ถ„ํฌ๊ฐ€ ๋น„์Šทํ•˜๋‹คโ€๋Š” ์ง๊ด€์„ KL ๋ฐœ์‚ฐ ์„ ํ†ตํ•œ Bregman ๋ฐœ์‚ฐ ์œผ๋กœ ์ •๋Ÿ‰ํ™”ํ•œ๋‹ค. ์ด๋Š” ์ •๋ณด๊ธฐํ•˜ํ•™ ์ด ์ž์—ฐ์Šค๋Ÿฌ์šด ๊ฑฐ๋ฆฌยท๊ณก๋ฅ  ๊ฐœ๋…์„ ์ œ๊ณตํ•œ๋‹ค๋Š” ์ ์„ ๊ฐ•์กฐํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์ด๋ก ์  ๊ธฐ์—ฌ | ๋‚ด์šฉ | ์„ค๋ช… | ์˜๋ฏธ | | | | | | Bregman(๋“€์–ผ ํ”Œ๋žซ)

Computer Science Machine Learning
No Image

The invariance of the Auslander-Reiten Formula for hereditary algebras

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ Auslanderโ€‘Reiten ์ด๋ก  ์€ ๋ชจ๋“ˆ ๋ฒ”์ฃผ์—์„œ ์‚ฌ์ƒ๋“ค์˜ ์‚ฌํ›„(translation) ๊ตฌ์กฐ๋ฅผ ํŒŒ์•…ํ•˜๋Š” ํ•ต์‹ฌ ๋„๊ตฌ์ด๋ฉฐ, ํŠนํžˆ AR ๊ณต์‹ ์€ (operatorname{Ext}^{1})์™€ (operatorname{Hom}) ์‚ฌ์ด์˜ ์ด์ค‘์„ฑ(duality)์„ ๋ช…์‹œํ•œ๋‹ค. ๊ธฐ์กด ๋ฌธํ—Œ์—์„œ๋Š” AR ๊ณต์‹์ด (tau)์™€ (tau^{ }) ์‚ฌ์ด์˜ ์ž์—ฐ ๋™ํ˜• ์ž„์€ ์•Œ๋ ค์ ธ ์žˆ์œผ๋‚˜, ๊ณต์‹ ์ž์ฒด๊ฐ€ (tau)์— ๋Œ€ํ•ด ๋ถˆ๋ณ€ ์ด๋ผ๋Š” ๊ตฌ์ฒด์  ํ˜„์ƒ์€ ๋ช…์‹œ์ ์œผ๋กœ ๋‹ค๋ฃจ์–ด์ง€์ง€ ์•Š์•˜๋‹ค. ์ด ๋ถˆ๋ณ€์„ฑ์€ ์‚ฌ์ „ํˆฌ์‚ฌ ๋Œ€์ˆ˜ (

Mathematics
No Image

The Quantum Symmetric Simple Exclusion Process in the Continuum and Free Processes

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๊ณ ์ „ ๋น„ํ‰ํ˜• ํ˜„์ƒ์˜ ํ•œ๊ณ„ : ๊ณ ์ „์ ์ธ ๊ฑฐ์‹œ ๋ณ€๋™ ์ด๋ก (MFT)์€ ํ™•์‚ฐ ์‹œ์Šคํ…œ์˜ ์ „์†ก ๋ฐ ๋ณ€๋™์„ ์„ฑ๊ณต์ ์œผ๋กœ ๊ธฐ์ˆ ํ•˜์ง€๋งŒ, ์–‘์ž ์ฝ”ํžˆ๋Ÿฐ์Šคยท์–ฝํž˜ยท๊ฐ„์„ญ์„ ํฌํ•จํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์–‘์ž ํ™•์žฅ ํ•„์š”์„ฑ : ์–‘์ž ์‹œ์Šคํ…œ์—์„œ์˜ ๋น„ํ‰ํ˜• ํ˜„์ƒ์€ ์ตœ๊ทผ ์–‘์ž ํšŒ๋กœ์™€ ์–‘์ž ๋ฐฐ์ œ ๊ณผ์ • ์—ฐ๊ตฌ๋ฅผ ํ†ตํ•ด ์ ์ฐจ ์กฐ๋ช…๋˜๊ณ  ์žˆ๋‹ค. QSSEP๋Š” ์ด๋Ÿฌํ•œ ์–‘์ž ๋น„ํ‰ํ˜• ํ˜„์ƒ์„ ๋ชจ๋ธ๋งํ•˜๋Š” ๋Œ€ํ‘œ์ ์ธ ์‚ฌ๋ก€์ด๋ฉฐ, ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ด์‚ฐ ๊ฒฉ์ž์—์„œ ์‹œ์ž‘ํ•ด ์—ฐ์†๊ทนํ•œ์„ ์ทจํ•˜๋Š” ๋ฐฉ์‹์— ์˜์กดํ–ˆ๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๊ตฌ๋ถ„ | ๋‚ด์šฉ | ์˜๋ฏธ | | | | | | ์—ฐ์†์ฒด ์ง์ ‘ ์ •์˜ | Q

MATH-PH
No Image

The regular multivariate quadratic problem

1. ๋ฌธ์ œ ์ •์˜์™€ ์ •๊ทœ์„ฑ ์กฐ๊ฑด์˜ ์˜์˜ ๊ธฐ์กด ๋‹ค๋ณ€๋Ÿ‰ 2์ฐจ ๋ฐฉ์ •์‹(MQ) ๋ฌธ์ œ๋Š” ํ•ด์˜ ์กด์žฌ ์—ฌ๋ถ€๋งŒ์„ ํŒ๋‹จํ•œ๋‹ค. ์—ฌ๊ธฐ์„œ ์ œ์•ˆ๋œ ์ •๊ทœ MQ๋Š” ํ•ด๊ฐ€ ๋ธ”๋ก๋‹น ํ•˜๋‚˜์˜ ๋น„์˜ ์›์†Œ ๋งŒ์„ ๊ฐ–๋Š” ์ถ”๊ฐ€ ์ œ์•ฝ์„ ๋‘”๋‹ค. ์ด ์ œ์•ฝ์€ ํ•ด ๊ณต๊ฐ„์„ ํฌ๊ฒŒ ์–‡๊ฒŒ ๋งŒ๋“ค๋ฉฐ, ํŠนํžˆ ์ŠคํŒŒ์Šค(sparse) ๊ตฌ์กฐ ๋ฅผ ๊ฐ•์ œํ•œ๋‹ค. ์•”ํ˜ธํ•™์ ์œผ๋กœ๋Š” ํ‚ค ์ƒ์„ฑยท๋ณตํ˜ธํ™” ๊ณผ์ •์—์„œ ๊ตฌ์กฐ์  ์ •๋ณด๋ฅผ ์ œํ•œํ•จ์œผ๋กœ์จ ๊ณต๊ฒฉ ํ‘œ๋ฉด์„ ์ค„์ด๋Š” ํšจ๊ณผ๊ฐ€ ๊ธฐ๋Œ€๋œ๋‹ค. 2. NPโ€‘์™„์ „์„ฑ ์ฆ๋ช… ์ €์ž๋Š” ๊ธฐ์กด์˜ ์ผ๋ฐ˜ MQ ๋ฌธ์ œ์˜ NPโ€‘์™„์ „์„ฑ์„ ์ด์šฉํ•ด ์ •๊ทœ MQ ๋ฌธ์ œ๋กœ์˜ ๋‹คํ•ญ์‹ ์‹œ๊ฐ„ ๊ฐ์†Œ(reduction) ๋ฅผ ๊ตฌ์„ฑํ•œ๋‹ค. ํ•ต

< Category Statistics (Total: 2829) >

Electrical Engineering and Systems Science
100
General
731
General Relativity
22
HEP-EX
17
HEP-LAT
3
HEP-PH
39
HEP-TH
19
MATH-PH
36
NUCL-EX
2
NUCL-TH
5
Quantum Physics
41

Start searching

Enter keywords to search articles

โ†‘โ†“
โ†ต
ESC
โŒ˜K Shortcut