KOINEU Logo
No Image

Alexander Dalgarno and the development of astrochemistry

1. ์ฒœ๋ฌธํ™”ํ•™์˜ ์—ญ์‚ฌ์  ๋ฐฐ๊ฒฝ ๊ด€์ธก ํ˜๋ช… : 1970๋…„๋Œ€ ์ดˆ, ์ ์™ธ์„ ยท์ž์™ธ์„ ยท์ „ํŒŒ ๋“ฑ ๋น„๊ฐ€์‹œ๊ด‘์„  ์˜์—ญ์—์„œ์˜ ๊ณ ๊ฐ๋„ ๊ด€์ธก์ด ๊ฐ€๋Šฅํ•ด์ง€๋ฉด์„œ, ์ด์ „์— โ€œ๋ฌด๊ธฐ์„ฑโ€์ด๋ผ๊ณ  ์—ฌ๊ฒจ์กŒ๋˜ ์„ฑ๊ฐ„ ๊ตฌ๋ฆ„์— ์ˆ˜์‹ญ ์ข… ์ด์ƒ์˜ ๋ถ„์ž๊ฐ€ ์กด์žฌํ•จ์ด ํ™•์ธ๋˜์—ˆ๋‹ค. ํ•™์ œ๊ฐ„ ํ•„์š”์„ฑ : ์ฒœ๋ฌธํ•™, ๋ฌผ๋ฆฌํ™”ํ•™, ์ด๋ก  ๋ฌผ๋ฆฌํ•™์ด ๊ฒฐํ•ฉํ•ด์•ผ๋งŒ ๋ณต์žกํ•œ ๋ฐ˜์‘ ๋ฉ”์ปค๋‹ˆ์ฆ˜๊ณผ ์—๋„ˆ์ง€ ์ „๋‹ฌ ๊ณผ์ •์„ ์„ค๋ช…ํ•  ์ˆ˜ ์žˆ์—ˆ์œผ๋ฉฐ, ์ด๋Š” ์ฒœ๋ฌธํ™”ํ•™์ด๋ผ๋Š” ์ƒˆ๋กœ์šด ํ•™๋ฌธ์˜ ํƒ„์ƒ์„ ์ด‰์ง„ํ–ˆ๋‹ค. 2. ์•Œ๋ ‰์‚ฐ๋” ๋‹ฌ๊ฐ€๋…ธ์˜ ํ•ต์‹ฌ ๊ธฐ์—ฌ | ๋ถ„์•ผ | ์ฃผ์š” ์—ฐ๊ตฌยท์„ฑ๊ณผ | ์ฒœ๋ฌธํ™”ํ•™์— ๋ฏธ์นœ ์˜ํ–ฅ | | | | | | ์›์žยท๋ถ„์ž ์ถฉ๋Œ ์ด๋ก  |

No Image

Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ Longโ€‘tail ์ƒํ™ฉ : ๊ธฐ์กด ๋ชจ๋ฐฉ ํ•™์Šต ๊ธฐ๋ฐ˜ ์ž์œจ์ฃผํ–‰์€ ๋ฐ์ดํ„ฐ๊ฐ€ ํ’๋ถ€ํ•œ ์ผ๋ฐ˜ ์ƒํ™ฉ์—์„œ๋Š” ๊ฐ•๋ ฅํ•˜์ง€๋งŒ, ๋“œ๋ฌผ๊ณ  ์œ„ํ—˜ํ•œ ์ƒํ™ฉ์—์„œ๋Š” ๋ผ๋ฒจ์ด ๋ถ€์กฑํ•ด ์ธ๊ณผ์  ํŒ๋‹จ์ด ์–ด๋ ค์›Œ์ง. ์ธ๊ณผ ์ถ”๋ก ์˜ ํ•„์š”์„ฑ : ์ธ๊ฐ„ ์šด์ „์ž๋Š” โ€œ์™œโ€ ํŠน์ • ํ–‰๋™์„ ํ•˜๋Š”์ง€์— ๋Œ€ํ•œ ์ธ๊ณผ์  ์ดํ•ด๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์œ„ํ—˜์„ ํšŒํ”ผํ•œ๋‹ค. ์ด๋ฅผ ๋ชจ๋ธ์— ๋„์ž…ํ•˜๋ฉด ํ•ด์„ ๊ฐ€๋Šฅ์„ฑ ๊ณผ ์•ˆ์ „์„ฑ ์„ ๋™์‹œ์— ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ์Œ. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๊ธฐ์—ฌ ๋‚ด์šฉ | ์˜์˜ | | | | | | 1 | CoC ๋ฐ์ดํ„ฐ์…‹ ๊ตฌ์ถ• (์ž๋™ ๋ผ๋ฒจ๋ง + ์ธ๊ฐ„โ€‘์ธโ€‘๋ฃจํ”„) | ์‹ค์ œ ์šด์ „ ํ–‰๋™๊ณผ ์ง

No Image

An item is worth one token in Multimodal Large Language Models-based Sequential Recommendation

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ LLM ๊ธฐ๋ฐ˜ SR์˜ ํ•œ๊ณ„ : ๊ธฐ์กด LLM์€ ํ…์ŠคํŠธ์— ํŠนํ™”๋ผ ์žˆ์–ด ์ด๋ฏธ์ง€ยท๋™์˜์ƒ ๋“ฑ ๋ฉ€ํ‹ฐ๋ชจ๋‹ฌ ์ •๋ณด๋ฅผ ์ง์ ‘ ํ™œ์šฉํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ํŠนํžˆ ์•„์ดํ…œ ์„ค๋ช…์ด ๊ธธ๊ณ  ์ค‘๋ณต๋  ๊ฒฝ์šฐ ํ† ํฐ ์ˆ˜๊ฐ€ ๊ธ‰์ฆํ•ด ๋น„์šฉ์ด ๋น„ํšจ์œจ์ ์ด๋‹ค. ์‹œํ€€์Šค ๊ธธ์ด์™€ ์ฃผ์˜ ๋ฉ”์ปค๋‹ˆ์ฆ˜ : Transformer ๊ธฐ๋ฐ˜ ๋ชจ๋ธ์€ ์ž…๋ ฅ ๊ธธ์ด๊ฐ€ ๊ธธ์–ด์งˆ์ˆ˜๋ก ์ดˆ๊ธฐ ํ† ํฐ์— ๋Œ€ํ•œ attention ๊ฐ€์ค‘์น˜๊ฐ€ ๊ฐ์†Œํ•˜๋Š” โ€œlongโ€‘range dependencyโ€ ๋ฌธ์ œ์— ์ง๋ฉดํ•œ๋‹ค. ์ด๋Š” ์‚ฌ์šฉ์ž ํ–‰๋™์ด ์ˆ˜๋ฐฑยท์ˆ˜์ฒœ ๋‹จ๊ณ„์— ์ด๋ฅด๋Š” ์‹ค์ œ ์„œ๋น„์Šค ์‹œ๋‚˜๋ฆฌ์˜ค์—์„œ ํฐ ์žฅ์• ๊ฐ€ ๋œ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์—ฌ

Model
No Image

Anchors in the Machine: Behavioral and Attributional Evidence of Anchoring Bias in LLMs

1. ์—ฐ๊ตฌ ๋™๊ธฐ์™€ ๋ฐฐ๊ฒฝ ์ธ์ง€ ํŽธํ–ฅ๊ณผ LLM : ์ธ๊ฐ„์˜ ํŒ๋‹จ ํŽธํ–ฅ์„ LLM์— ๊ทธ๋Œ€๋กœ ํˆฌ์˜ํ•˜๋Š” ๊ฒƒ์ด ๊ณผ์—ฐ ๋ชจ๋ธ ๋‚ด๋ถ€์˜ ํ™•๋ฅ  ๊ตฌ์กฐ ๋ณ€ํ™”๋ฅผ ์˜๋ฏธํ•˜๋Š”์ง€ ์—ฌ๋ถ€๋Š” ์•„์ง ๋…ผ์Ÿ ์ค‘์ด๋‹ค. ์•ต์ปค๋ง์€ โ€œ์ดˆ๊ธฐ๊ฐ’์ด ์ดํ›„ ํŒ๋‹จ์— ๊ณผ๋„ํ•˜๊ฒŒ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋Š”โ€ ํ˜„์ƒ์œผ๋กœ, LLM์ด ํ”„๋กฌํ”„ํŠธ ๋‚ด ์ˆซ์žยท๋ฌธ๊ตฌ๋ฅผ ์–ด๋–ป๊ฒŒ ๊ฐ€์ค‘์น˜๋กœ ๋ณ€ํ™˜ํ•˜๋Š”์ง€๋ฅผ ์‚ดํ”ผ๊ธฐ์— ์ ํ•ฉํ•œ ํ…Œ์ŠคํŠธ๋ฒ ๋“œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ์˜ ํ•œ๊ณ„ : ๋Œ€๋ถ€๋ถ„์˜ ์„ ํ–‰ ์—ฐ๊ตฌ๋Š” โ€œ์ •๋‹ต๋ฅ  ๊ฐ์†Œโ€ ํ˜น์€ โ€œ์ถœ๋ ฅ ํ˜•ํƒœ ๋ณ€ํ™”โ€ ์ •๋„๋งŒ ๋ณด๊ณ ํ–ˆ์œผ๋ฉฐ, ๋‚ด๋ถ€ ํ™•๋ฅ  ๋ถ„ํฌ ์™€ ๊ท€์ธ ๋ฉ”์ปค๋‹ˆ์ฆ˜ ์„ ์ •๋Ÿ‰ํ™”ํ•˜์ง€ ๋ชปํ–ˆ๋‹ค. ์ด๋Š” LLM์ด ์‹ค์ œ๋กœ โ€˜ํŽธํ–ฅ๋œ ์‚ฌ๊ณ โ€™๋ฅผ ํ•˜๋Š”์ง€

No Image

Applying Time Series Deep Learning Models to Forecast the Growth of Perennial Ryegrass in Ireland

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํƒ„์†Œ ํก์ˆ˜์›์œผ๋กœ์„œ ์ดˆ์ง€์˜ ์ค‘์š”์„ฑ : ์ดˆ์ง€๋Š” ์ „ ์„ธ๊ณ„ ํƒ„์†Œ ์ €์žฅ๋Ÿ‰์˜ ์•ฝ 10%๋ฅผ ์ฐจ์ง€ํ•˜๋ฉฐ, ๊ธฐํ›„ ๋ณ€ํ™” ์™„ํ™”์— ํ•ต์‹ฌ์ ์ธ ์—ญํ• ์„ ํ•œ๋‹ค. ์•„์ผ๋žœ๋“œ ๋‚™๋†์—…์˜ ํ˜„ํ™ฉ : ๋‚™๋†์—…์€ GDP์˜ 3~4%๋ฅผ ์ฐจ์ง€ํ•˜์ง€๋งŒ, ์‚ฌ๋ฃŒ ๋น„์šฉ ์ƒ์Šนยท๊ธฐํ›„ ๋ณ€๋™์„ฑ ๋“ฑ์œผ๋กœ ์ˆ˜์ต์„ฑ ์•…ํ™”์™€ ํ™˜๊ฒฝ ๋ถ€๋‹ด์ด ๋™์‹œ์— ๋Œ€๋‘๋˜๊ณ  ์žˆ๋‹ค. ๊ธฐ์กด ๋ชจ๋ธ์˜ ํ•œ๊ณ„ : ๋ฉ”์นด๋‹ˆ์ฆ˜ ๊ธฐ๋ฐ˜(์˜ˆ: ์„ฑ์žฅ ๋‹จ๊ณ„ ๋ชจ๋ธ, ๊ธฐ์ƒโ€‘์‹์ƒ ์—ฐ๊ณ„ ๋ชจ๋ธ)์€ ๋ณต์žกํ•œ ํŒŒ๋ผ๋ฏธํ„ฐ ํŠœ๋‹๊ณผ ํ˜„์žฅ ์ ์šฉ ๋น„์šฉ์ด ๋†’์•„ ์‹ค๋ฌด์—์„œ ํ™œ์šฉ๋„๊ฐ€ ๋‚ฎ๋‹ค. 2. ๋ฐ์ดํ„ฐ ํŠน์„ฑ ๋‹จ๋ณ€๋Ÿ‰ ์‹œ๊ณ„์—ด : ์ดˆ์ง€ ๋†’์ด(Grass Height)

Model Learning
No Image

Approaching Low-Cost Cardiac Intelligence with Semi-Supervised Knowledge Distillation

| ๋ถ„์„ ํ•ญ๋ชฉ | ๋‚ด์šฉ ๋ฐ ํ‰๊ฐ€ | | | | | ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ | โ€ข ํ˜„์žฌ ๊ณ ์„ฑ๋Šฅ ์‹ฌ์žฅ AI๋Š” ๋Œ€๊ทœ๋ชจ ๋ผ๋ฒจ๋ง๋œ ECG ๋ฐ์ดํ„ฐ์™€ ๊ณ ์„ฑ๋Šฅ GPU๊ฐ€ ์ „์ œ๋ผ ๋ณ‘์›ยท์—ฐ๊ตฌ์†Œ ์ˆ˜์ค€์— ๊ตญํ•œ๋จ.<br>โ€ข ์›จ์–ด๋Ÿฌ๋ธ” ๋””๋ฐ”์ด์Šค ๋ณด๊ธ‰์œผ๋กœ 1โ€‘lead ECG๊ฐ€ ์ผ์ƒ์—์„œ ์ˆ˜์ง‘ ๊ฐ€๋Šฅํ•˜์ง€๋งŒ, ์ •๋ณด๋Ÿ‰ยท๋…ธ์ด์ฆˆ ์ธก๋ฉด์—์„œ ๊ธฐ์กด 12โ€‘lead ECG ๊ธฐ๋ฐ˜ ๋ชจ๋ธ๋ณด๋‹ค ์—ด์œ„์— ์žˆ์Œ. <br>โ€ข ๋”ฐ๋ผ์„œ โ€œ์ €๋น„์šฉยท๊ณ ์„ฑ๋Šฅโ€์„ ๋™์‹œ์— ๋‹ฌ์„ฑํ•˜๋Š” LCCI๊ฐ€ ์‹ค์งˆ์ ์ธ ์˜๋ฃŒ ํ˜„์žฅ์— ํ•„์ˆ˜์ ์ž„. | | ํ•ต์‹ฌ ๊ธฐ์—ฌ | 1. Regionโ€‘aware Distillation : ์‹ฌ์ „๋„ ํŒŒํ˜•์—

No Image

Are Agents Probabilistic Automata? A Trace-Based, Memory-Constrained Theory of Agentic AI

1. ์—ฐ๊ตฌ ๋™๊ธฐ์™€ ๋ฐฐ๊ฒฝ ์—์ด์ „ํŠธ AI์˜ ๊ฒ€์ฆ ํ•„์š”์„ฑ : ๋Œ€ํ˜• ์–ธ์–ด ๋ชจ๋ธ(LLM) ๊ธฐ๋ฐ˜ ์—์ด์ „ํŠธ๊ฐ€ ์‹ค์ œ ํ™˜๊ฒฝ์—์„œ ํ–‰๋™ํ•  ๋•Œ, ์•ˆ์ „ยท์‹ ๋ขฐ์„ฑ์„ ๋ณด์žฅํ•˜๊ธฐ ์œ„ํ•œ ํ˜•์‹์  ๊ฒ€์ฆ์ด ๊ธ‰์„ ๋ฌด์ด๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ ์ •์ฑ… ์ž์ฒด (์˜ˆ: RL ์ •์ฑ…)์˜ ์•ˆ์ „์„ฑ์„ ๋‹ค๋ฃจ์ง€๋งŒ, ๊ตฌํ˜„ ๋ ˆ๋ฒจ(๋ฉ”๋ชจ๋ฆฌยท์ œ์–ด ํ๋ฆ„) ๊นŒ์ง€ ํฌ๊ด„ํ•˜๋Š” ์ฒด๊ณ„๋Š” ๋ถ€์กฑํ–ˆ๋‹ค. ์ž๋™์ž ์ด๋ก ๊ณผ ํŠธ๋ ˆ์ด์Šค ์˜๋ฏธ๋ก  : ์ž๋™์ž ์ด๋ก ์€ ์–ธ์–ด(ํŠธ๋ ˆ์ด์Šค)์˜ ๊ตฌ์กฐ ์™€ ์ „์ด ์‹œ์Šคํ…œ ์„ ์—ฐ๊ฒฐํ•ด ๊ฒ€์ฆ์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•œ๋‹ค. ์ด ๋…ผ๋ฌธ์€ ์—์ด์ „ํŠธโ€‘ํ™˜๊ฒฝ ์ƒํ˜ธ์ž‘์šฉ์„ ํ™•๋ฅ ์  ํŠธ๋ ˆ์ด์Šค๋กœ ๋ชจ๋ธ๋งํ•จ์œผ๋กœ์จ, ๊ธฐ์กด ๊ฒฐ์ •๋ก ์  ์ž๋™์ž ๊ฒ€์ฆ์„ ํ™•๋ฅ ์ 

No Image

Atomic-Scale Surface Imaging of bulk Epitaxial CsPbBr3 Perovskite Single Crystals on Mica using Light Assisted Scanning Tunneling Microscopy at Low-Temperature (80 K)

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ „์ด๊ธˆ์† ํ• ๋ผ์ด๋“œ ํผ๋กœ๋ธŒ์Šค์นด์ดํŠธ(MHP) ๋Š” ๋†’์€ ํก์ˆ˜ ๊ณ„์ˆ˜ยท๊ธด ์บ๋ฆฌ์–ด ํ™•์‚ฐ ๊ฑฐ๋ฆฌยท๊ฐ„๋‹จํ•œ ํ•ฉ์„ฑ๋ฒ• ๋•๋ถ„์— ํƒœ์–‘์ „์ง€, LED, ๊ด‘์ด‰๋งค, ์„ผ์„œ ๋“ฑ ๋‹ค์–‘ํ•œ ๋ถ„์•ผ์—์„œ ํ•ต์‹ฌ ์†Œ์žฌ๋กœ ๋ถ€์ƒํ•˜๊ณ  ์žˆ๋‹ค. ํŠนํžˆ ์ „ํ˜€ ์œ ๊ธฐ ์„ฑ๋ถ„์ด ์—†๋Š” CsPbBrโ‚ƒ ๋Š” ์—ดยท๊ตฌ์กฐ์  ์•ˆ์ •์„ฑ์ด ๋›ฐ์–ด๋‚˜๋ฉฐ, 2.3 eV ์ •๋„์˜ ๋„“์€ ๋ฐด๋“œ๊ฐญ๊ณผ ๋†’์€ ๊ด‘๋ฐœ๊ด‘ ํšจ์œจ์„ ๊ฐ€์ ธ ์ฒญโ€‘๋…น์ƒ‰ ๋ฐœ๊ด‘ ๋ฐ ์•ˆ์ •์ ์ธ ๊ด‘๊ฒ€์ถœ๊ธฐ์— ์ ํ•ฉํ•˜๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ํ‘œ๋ฉดยท๊ณ„๋ฉด ์ด ์ „ํ•˜ ์žฌ๊ฒฐํ•ฉ, ์•ˆ์ •์„ฑ, ์—๋„ˆ์ง€ ์ •๋ ฌ ๋“ฑ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์ด ์ ์  ๋” ์ค‘์š”ํ•ด์ง€๋ฉด์„œ, ์›์ž ์ˆ˜์ค€์—์„œ์˜ ํ‘œ๋ฉด ๊ตฌ์กฐ ๊ทœ๋ช…์€ ์žฅ์น˜

Condensed Matter
No Image

AutoBench: Automating LLM Evaluation through Reciprocal Peer Assessment

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ •์  ๋ฒค์น˜๋งˆํฌ์˜ ํ•œ๊ณ„ : ํ…Œ์ŠคํŠธ์…‹ ์˜ค์—ผ, ์ตœ์‹  ๋ชจ๋ธ์— ๋Œ€ํ•œ ์ ์‹œ์„ฑ ๋ถ€์กฑ, ๋„๋ฉ”์ธ ํŽธํ–ฅ ๋“ฑ. ๋™์ ยท์ž๊ธ‰์ž์กฑํ˜• ํ‰๊ฐ€ : ๋ชจ๋ธ ์ž์ฒด๊ฐ€ ํ‰๊ฐ€ ๋ฐ์ดํ„ฐ๋ฅผ ์ƒ์„ฑยทํŒ๋‹จํ•จ์œผ๋กœ์จ ์ง€์†์ ์ธ ์—…๋ฐ์ดํŠธ์™€ ์˜ค์—ผ ๋ฐฉ์ง€๋ฅผ ๋ชฉํ‘œ. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด Reciprocal Peer Assessment : ๋ชจ๋ธ์ด ์„œ๋กœ ์งˆ๋ฌธ์„ ๋งŒ๋“ค๊ณ , ๋‹ต๋ณ€์„ ์ œ์ถœํ•˜๊ณ , ๋‹ค๋ฅธ ๋ชจ๋ธ์ด ์‹ฌ์‚ฌํ•œ๋‹ค๋Š” ์ˆœํ™˜ ๊ตฌ์กฐ. Iterative Weighting : ํ‰๊ฐ€์ž(์‹ฌ์‚ฌ์ž)๋“ค์˜ ์‹ ๋ขฐ๋„๋ฅผ ๋ฐ˜๋ณต์ ์œผ๋กœ ์—…๋ฐ์ดํŠธํ•˜์—ฌ, ์ผ๊ด€๋œ ํ‰๊ฐ€์ž๋ฅผ ๋” ํฐ ๊ฐ€์ค‘์น˜๋กœ ๋ฐ˜์˜ํ•œ๋‹ค. Consensusโ€‘B

No Image

Autograder+: A Multi-Faceted AI Framework for Rich Pedagogical Feedback in Programming Education

1. ์—ฐ๊ตฌ์˜ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํ”„๋กœ๊ทธ๋ž˜๋ฐ ๊ต์œก์˜ ๊ธ‰์„ฑ์žฅ : ์˜จ๋ผ์ธ ์ฝ”์Šค, ๋ถ€ํŠธ์บ ํ”„, ๊ณ ๋“ฑยท๋Œ€ํ•™ ์ˆ˜์ค€์˜ CS ๊ต์œก ํ™•๋Œ€๋กœ ๋Œ€๋Ÿ‰์˜ ์ฝ”๋“œ ์ œ์ถœ์ด ๋ฐœ์ƒํ•œ๋‹ค. ์ „ํ†ต์  ์ž๋™ ์ฑ„์ ๊ธฐ์˜ ํ•œ๊ณ„ : ์ •๋‹ต/์˜ค๋‹ต๋งŒ ์ œ๊ณต, ํ•™์ƒ์˜ ์˜ค๋ฅ˜ ์›์ธยทํ•™์Šต ๋‹จ๊ณ„ ํŒŒ์•… ์–ด๋ ค์›€, ํ”ผ๋“œ๋ฐฑ์˜ ๊ฐœ์ธํ™” ๋ถ€์กฑ. AIยทLLM์˜ ๋ถ€์ƒ : GPTโ€‘4ยทClaude ๋“ฑ ๋Œ€ํ˜• ์–ธ์–ด ๋ชจ๋ธ์ด ์ž์—ฐ์–ด ์ดํ•ดยท์ƒ์„ฑ์—์„œ ๋›ฐ์–ด๋‚œ ์„ฑ๋Šฅ์„ ๋ณด์—ฌ, ๊ต์œก ํ”ผ๋“œ๋ฐฑ์— ์ ์šฉ ๊ฐ€๋Šฅ์„ฑ์ด ์ปค์กŒ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๊ธฐ์—ฌ ๋‚ด์šฉ | ์˜์˜ | | | | | | 1 | Fineโ€‘tuned LLM ๊ธฐ๋ฐ˜ ์ž๋™ ํ”ผ๋“œ๋ฐฑ :

Framework
No Image

Benchmarking Educational LLMs with Analytics: A Case Study on Gender Bias in Feedback

1. ์—ฐ๊ตฌ ์„ค๊ณ„์™€ ๋ฐฉ๋ฒ•๋ก  | ์š”์†Œ | ๋‚ด์šฉ | ๊ฐ•์  | ํ•œ๊ณ„ | | | | | | | ๋ฐ์ดํ„ฐ | AES 2.0 ์ฝ”ํผ์Šค โ†’ 600๊ฐœ ์‹ค์ œ ํ•™์ƒ ์—์„ธ์ด | ์‹ค์ œ ๊ต์œก ํ˜„์žฅ ๋ฐ์ดํ„ฐ ์‚ฌ์šฉ์œผ๋กœ ์™ธ์  ํƒ€๋‹น์„ฑ ํ™•๋ณด | ์˜์–ด ๊ธฐ๋ฐ˜ ์ฝ”ํผ์Šค์ด๋ฉฐ, ํ•œ๊ตญ์–ดยท๋‹ค๋ฌธํ™” ๋งฅ๋ฝ์— ์ง์ ‘ ์ ์šฉ ์–ด๋ ค์›€ | | ๋Œ€์กฐ๊ตฐ ์ƒ์„ฑ | (i) Lexiconโ€‘based ์„ฑ๋ณ„ ์šฉ์–ด ๊ต์ฒด (์•”์‹œ์ ) <br> (ii) ํ”„๋กฌํ”„ํŠธ์— ์„ฑ๋ณ„ ๋ฐฐ๊ฒฝ ์‚ฝ์ž… (๋ช…์‹œ์ ) | ๋‘ ์ฐจ์›์˜ โ€œcounterfactualโ€ ์ ‘๊ทผ๋ฒ•์œผ๋กœ ํŽธํ–ฅ ์›์ธ ๊ตฌ๋ถ„ ๊ฐ€๋Šฅ | ์šฉ์–ด ๊ต์ฒด๊ฐ€ ๋ฌธ๋งฅ์— ๋”ฐ๋ผ ๋ถ€์ž์—ฐ์Šค๋Ÿฌ์šธ ์ˆ˜ ์žˆ์Œ; ํ”„

No Image

Best Practices for Biorisk Evaluations on Open-Weight Bio-Foundation Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์˜คํ”ˆ ์›จ์ดํŠธ ๋ชจ๋ธ์˜ ๊ธ‰๋ถ€์ƒ : ๋Œ€๊ทœ๋ชจ ๋ฐ”์ด์˜ค ๋ฐ์ดํ„ฐ์™€ ์ปดํ“จํŒ… ํŒŒ์›Œ์˜ ๊ฒฐํ•ฉ์œผ๋กœ, ์˜คํ”ˆ ์›จ์ดํŠธ ๋ชจ๋ธ์€ ์œ ์ „์ฒด ํ•ด์„, ์‹ ์•ฝ ํ›„๋ณด ๋ฌผ์งˆ ์˜ˆ์ธก, ํ•ฉ์„ฑ์ƒ๋ฌผํ•™ ์„ค๊ณ„ ๋“ฑ ๋‹ค์–‘ํ•œ ๋ถ„์•ผ์— ํ˜์‹ ์„ ๊ฐ€์ ธ์˜ค๊ณ  ์žˆ๋‹ค. ๋ฐ”์ด์˜ค๋ฆฌ์Šคํฌ์˜ ๋ณตํ•ฉ์„ฑ : ์ด๋Ÿฌํ•œ ๋ชจ๋ธ์ด ์ƒ์„ฑํ•˜๊ฑฐ๋‚˜ ์ œ์•ˆํ•˜๋Š” ์ƒ๋ฌผํ•™์  ์ •๋ณด๋Š” ์•…์˜์  ์‚ฌ์šฉ(์˜ˆ: ๋ณ‘์›์ฒด ์„ค๊ณ„, ๋…์„ฑ ๋ฌผ์งˆ ํ•ฉ์„ฑ) ํ˜น์€ ์˜๋„์น˜ ์•Š์€ ์•ˆ์ „ ๋ฌธ์ œ(์˜ˆ: ์˜ค์ง„, ๋ถ€์ •ํ™•ํ•œ ์น˜๋ฃŒ ๊ถŒ๊ณ )๋ฅผ ์ดˆ๋ž˜ํ•  ์œ„ํ—˜์ด ์กด์žฌํ•œ๋‹ค. ๊ทœ์ œยท์œค๋ฆฌ์  ๊ฒฉ์ฐจ : ๊ธฐ์กด์˜ AI ์œค๋ฆฌยท์•ˆ์ „ ํ”„๋ ˆ์ž„์›Œํฌ๋Š” ์ฃผ๋กœ ์ผ๋ฐ˜ AI์— ์ดˆ์ ์„ ๋งž์ถ”์—ˆ์œผ๋ฉฐ, ๋ฐ”์ด์˜ค

Model
No Image

Beyond Fact Retrieval: Episodic Memory for RAG with Generative Semantic Workspaces

| ๊ตฌ๋ถ„ | ๋‚ด์šฉ | ํ‰๊ฐ€ยท์‹œ์‚ฌ์  | | | | | | ๋ฌธ์ œ ์ •์˜ | LLM์˜ ์ปจํ…์ŠคํŠธ ์œˆ๋„์šฐ ํ•œ๊ณ„ <br> ๊ธด ๋ฌธ์„œยท์‹œํ€€์Šค์—์„œ ์„ฑ๋Šฅ ์ €ํ•˜ <br> ๊ธฐ์กด RAG์€ โ€œ์‚ฌ์‹คโ€ ์ค‘์‹ฌ, ์„œ์‚ฌยท์—ํ”ผ์†Œ๋“œ ์ถ”์  ๋ถ€์žฌ | ํ˜„์žฌ ์ƒ์šฉ LLMยทRAG ์‹œ์Šคํ…œ์ด ์žฅ๊ธฐ ์ถ”๋ก ์— ์ทจ์•ฝํ•จ์„ ์ •ํ™•ํžˆ ์งš์Œ. ํŠนํžˆ โ€œ์‹œ๊ฐ„ยท๊ณต๊ฐ„์— ๊ณ ์ •๋œ ๋‚ด๋Ÿฌํ‹ฐ๋ธŒโ€๊ฐ€ ํ•„์š”ํ•œ ์‘์šฉ(๋ฒ•๋ฅ , ์˜๋ฃŒ, ์Šคํ† ๋ฆฌํ…”๋ง ๋“ฑ)์—์„œ ํฐ ๊ฒฉ์ฐจ๊ฐ€ ์กด์žฌํ•จ. | | ์ œ์•ˆ ๋ฐฉ๋ฒ• | Generative Semantic Workspace (GSW) <br>1. Operator : ๊ด€์ฐฐ โ†’ ์˜๋ฏธ ๊ตฌ์กฐ(์˜ˆ: ์‚ฌ๊ฑด, ํ–‰์œ„

No Image

Beyond Fixed Depth: Adaptive Graph Neural Networks for Node Classification Under Varying Homophily

[Cโ€‹atchy Title KO] โ€œ๊นŠ์ด ๊ณ ์ •์€ ์ด์ œ ๊ทธ๋งŒ! ์ง€์—ญ ๋™์งˆ์„ฑ์— ๋งž์ถ˜ ์ ์‘ํ˜• ๊ทธ๋ž˜ํ”„ ์‹ ๊ฒฝ๋งโ€ [โ€‹Abstract KO] ๊ทธ๋ž˜ํ”„ ์‹ ๊ฒฝ๋ง(GNN)์€ ๋…ธ๋“œ ๋ถ„๋ฅ˜์—์„œ ๋›ฐ์–ด๋‚œ ์„ฑ๊ณผ๋ฅผ ๋ณด์—ฌ์™”์ง€๋งŒ, ์—ฐ๊ฒฐ๋œ ๋…ธ๋“œ๊ฐ€ ์„œ๋กœ ๋‹ค๋ฅธ ๋ผ๋ฒจ์„ ๊ฐ–๋Š” ์ด์งˆ(heterophilic) ๊ทธ๋ž˜ํ”„์—์„œ๋Š” ์„ฑ๋Šฅ์ด ๊ธ‰๊ฒฉํžˆ ๋–จ์–ด์ง„๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋“ค์€ ์ด์งˆ์„ฑ์„ ์™„ํ™”ํ•˜๊ธฐ ์œ„ํ•œ ๋‹ค์–‘ํ•œ ๋ฉ”์ปค๋‹ˆ์ฆ˜์„ ์ œ์‹œํ–ˆ์ง€๋งŒ, ๋ชจ๋“  ๋…ธ๋“œ์— ๋™์ผํ•œ ์ง‘๊ณ„ ๊นŠ์ด(aggregation depth)๋ฅผ ์ ์šฉํ•œ๋‹ค๋Š” ๊ทผ๋ณธ์ ์ธ ํ•œ๊ณ„ ๋ฅผ ์•ˆ๊ณ  ์žˆ๋‹ค. ์‹ค์ œ ๊ทธ๋ž˜ํ”„์—์„œ๋Š” ๊ฐ ๋…ธ๋“œ๊ฐ€ ์œ„์น˜ํ•œ ์ง€์—ญ์˜ ๋™์งˆ์„ฑ ์ˆ˜์ค€๊ณผ ์ด

Network
No Image

Beyond Procedure: Substantive Fairness in Conformal Prediction

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์ ˆ์ฐจ์  ๊ณต์ •์„ฑ vs. ์‹ค์งˆ์  ๊ณต์ •์„ฑ : ๊ธฐ์กด CP ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ ์ ˆ์ฐจ์  ๋ณด์žฅ(์˜ˆ: ๋™์ผ ์ปค๋ฒ„๋ฆฌ์ง€)๋งŒ์„ ๋‹ค๋ฃจ์—ˆ์œผ๋ฉฐ, ์ด๋Ÿฌํ•œ ๋ณด์žฅ์ด downstream ์˜์‚ฌ๊ฒฐ์ •์—์„œ์˜ ํ˜•ํ‰์„ฑ์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ๊ฐ„๊ณผํ–ˆ๋‹ค. ๊ณ ์œ„ํ—˜ ๋ถ„์•ผ (ํ—ฌ์Šค์ผ€์–ด, ๊ธˆ์œต ๋“ฑ)์—์„œ๋Š” ๋ถˆํ™•์‹ค์„ฑ ์ •๋Ÿ‰ํ™”๊ฐ€ ์‹ ๋ขฐ์„ฑ ํ™•๋ณด์— ํ•ต์‹ฌ์ด๋ฏ€๋กœ, CP์˜ ์„ค๊ณ„๊ฐ€ ์‹ค์งˆ์  ๊ฒฐ๊ณผ์— ๋ฏธ์น˜๋Š” ํŒŒ๊ธ‰ํšจ๊ณผ๋ฅผ ์ดํ•ดํ•˜๋Š” ๊ฒƒ์ด ํ•„์ˆ˜์ ์ด๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๋‚ด์šฉ | ์˜์˜ | | | | | |โ‘ | LLMโ€‘inโ€‘theโ€‘loop ๊ณต์ •์„ฑ ํ‰๊ฐ€ ํ”„๋ ˆ์ž„์›Œํฌ ๊ตฌ์ถ• | ์ธ๊ฐ„ ์‹คํ—˜์˜ ๋น„์šฉยท์‹œ๊ฐ„ยทํ”ผ๋กœ๋„

Machine Learning Statistics
Bi-Objective Evolutionary Optimization for Large-Scale Open Pit Mine Scheduling Problem under Uncertainty with Chance Constraints

Bi-Objective Evolutionary Optimization for Large-Scale Open Pit Mine Scheduling Problem under Uncertainty with Chance Constraints

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ OPMSP์˜ ๋ณตํ•ฉ์„ฑ : ๋ธ”๋ก ์ฑ„๊ตด ์ˆœ์„œ, ๊ธฐ๊ฐ„ ํ• ๋‹น, ์„ ํ–‰ ๊ด€๊ณ„, ๋งค์žฅ๋Ÿ‰ ์ œ์•ฝ ๋“ฑ ๋‹ค์ค‘ ์ œ์•ฝ์ด ๋™์‹œ์— ์กด์žฌํ•œ๋‹ค. ๋ถˆํ™•์‹ค์„ฑ : ๊ด‘์ƒ ํ’ˆ์งˆ(grade)ยทํ•จ๋Ÿ‰์ด ํ™•๋ฅ ์ ์œผ๋กœ ๋ณ€๋™ํ•˜๋ฏ€๋กœ, ๊ฒฐ์ •๋ก ์  ๋ชจ๋ธ์€ ์‹ค์ œ ์ƒ์‚ฐ๋Ÿ‰ยท์ˆ˜์ต์„ ๊ณผ๋Œ€/๊ณผ์†Œ ํ‰๊ฐ€ํ•œ๋‹ค. ํ™•๋ฅ ์  ์ œ์•ฝ(Chance Constraints) : ์ œ์•ฝ์ด ์ผ์ • ํ™•๋ฅ (์˜ˆ: 95%) ์ด์ƒ ๋งŒ์กฑํ•˜๋„๋ก ํ•จ์œผ๋กœ์จ ์œ„ํ—˜์„ ์ •๋Ÿ‰ํ™”ํ•œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ํŠน์ • ์‹ ๋ขฐ์ˆ˜์ค€์„ ๊ณ ์ •ํ•˜๊ณ  ๋‹จ์ผ ๋ชฉํ‘œ(NPV)๋งŒ ์ตœ์ ํ™”ํ–ˆ์œผ๋‚˜, ์‹ ๋ขฐ์ˆ˜์ค€ ์„ ํƒ์ด ๊ฒฐ๊ณผ์— ํฐ ์˜ํ–ฅ์„ ๋ฏธ์นœ๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๊ตฌ๋ถ„

No Image

Bilateral parking procedures

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๊ณ ์ „์ ์ธ ์ฃผ์ฐจ ํ•จ์ˆ˜๋Š” ํ•ด์‹œ๋ง, ๋น„๊ต๋ถˆ๊ฐ€๋Šฅ ํŒŒํ‹ฐ์…˜, ์ดˆํ‰๋ฉด ๋ฐฐ์—ด ๋“ฑ๊ณผ ๊นŠ์€ ์—ฐ๊ด€์„ ๊ฐ€์ง€๊ณ  ์žˆ์–ด ์กฐํ•ฉ๋ก ยท๋Œ€์ˆ˜ํ•™ ์–‘์ชฝ์—์„œ ํ™œ๋ฐœํžˆ ์—ฐ๊ตฌ๋˜์–ด ์™”๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ โ€œ์˜ค๋ฅธ์ชฝ์œผ๋กœ๋งŒ ์ด๋™โ€ํ•˜๋Š” ๊ทœ์น™์— ๊ตญํ•œ๋˜์—ˆ์œผ๋ฉฐ, ์ด๋ฅผ ํ™•์žฅํ•˜๋Š” ์‹œ๋„๋Š” ๋“œ๋ฌผ๋‹ค. ์ €์ž๋“ค์€ โ€œ์–‘์ชฝ์œผ๋กœ ์ด๋™ ๊ฐ€๋Šฅโ€์ด๋ผ๋Š” ๊ฐ€์žฅ ์ง๊ด€์ ์ธ ์ผ๋ฐ˜ํ™”๋ฅผ ํ†ตํ•ด ๋ณดํŽธ์„ฑ(universality) ํ˜„์ƒ์„ ํƒ๊ตฌํ•œ๋‹ค๋Š” ์ ์—์„œ ์ƒˆ๋กญ๋‹ค. ์ด๋Š” โ€œ์ง€์—ญ์„ฑ(locality)โ€์ด๋ผ๋Š” ์ œ์•ฝ์„ ๋‘์–ด ์‹ค์ œ ๋ฌผ๋ฆฌ์  ๋ชจ๋ธ๊ณผ๋„ ์–ด๋А ์ •๋„ ์ผ์น˜ํ•˜๋„๋ก ๋งŒ๋“ ๋‹ค. 2. ํ•ต์‹ฌ ์ •์˜์™€ ๋ชจ๋ธ ์–‘๋ฐฉํ–ฅ ์ ˆ์ฐจ : ํ˜„

Mathematics
BitRL-Light: 1-bit LLM Agents with Deep Reinforcement Learning for Energy-Efficient Smart Home Lighting Optimization

BitRL-Light: 1-bit LLM Agents with Deep Reinforcement Learning for Energy-Efficient Smart Home Lighting Optimization

BitRLโ€‘Light๋Š” ํ˜„์žฌ ์Šค๋งˆํŠธ ํ™ˆ ์กฐ๋ช… ๋ถ„์•ผ์—์„œ ๊ฐ€์žฅ ํฐ ๋‘ ๊ฐ€์ง€ ๊ณผ์ œ, ์ฆ‰ ์—๋„ˆ์ง€ ํšจ์œจ๊ณผ ์‚ฌ์šฉ์ž ๊ฒฝํ—˜์„ ๋™์‹œ์— ๋งŒ์กฑ์‹œํ‚ค๋Š” ๋ฐฉ๋ฒ•์„ ์ œ์‹œํ•œ๋‹ค๋Š” ์ ์—์„œ ํ•™์ˆ ์ ยท์‚ฐ์—…์  ์˜๋ฏธ๊ฐ€ ํฌ๋‹ค. ์ฒซ์งธ, 1๋น„ํŠธ ์–‘์žํ™” LLM์„ ํ™œ์šฉํ•จ์œผ๋กœ์จ ๋ชจ๋ธ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ 8๋ฐฐ ์ด์ƒ ์••์ถ•ํ•˜๊ณ , ์—ฐ์‚ฐ๋Ÿ‰์„ ํฌ๊ฒŒ ๊ฐ์†Œ์‹œ์ผœ ์—ฃ์ง€ ๋””๋ฐ”์ด์Šค์—์„œ ์‹ค์‹œ๊ฐ„ ์ถ”๋ก ์ด ๊ฐ€๋Šฅํ•˜๋„๋ก ๋งŒ๋“  ์ ์€ ์–‘์žํ™” ๊ธฐ์ˆ ์˜ ์‹ค์šฉ์„ฑ์„ ์ž…์ฆํ•œ๋‹ค. ํŠนํžˆ ๋ผ์ฆˆ๋ฒ ๋ฆฌ ํŒŒ์ด 4์™€ ๊ฐ™์€ ์ €๊ฐ€ํ˜• ARM ๋ณด๋“œ์—์„œ 71.4๋ฐฐ์˜ ์ „๋ ฅ ์ ˆ๊ฐ๊ณผ 5.07๋ฐฐ์˜ ์†๋„ ํ–ฅ์ƒ์„ ๋‹ฌ์„ฑํ–ˆ๋‹ค๋Š” ์‹คํ—˜ ๊ฒฐ๊ณผ๋Š”, ํด๋ผ์šฐ๋“œ ์˜์กด์„ฑ์„ ์ตœ์†Œํ™”ํ•˜๊ณ  ๋ฐ์ดํ„ฐ

Computer Science Learning Artificial Intelligence
No Image

BMW: Bayesian Model-Assisted Adaptive Phase II Clinical Trial Design for Win Ratio Statistic

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์Šน๋ฅ (WR) ํ†ต๊ณ„์˜ ๋ถ€์ƒ : ๋ณตํ•ฉ์ง€ํ‘œ๋ฅผ ๋‹จ์ˆœํžˆ โ€œ์ฒซ ์‚ฌ๊ฑดโ€์œผ๋กœ๋งŒ ์ฒ˜๋ฆฌํ•˜๋Š” ์ „ํ†ต์  ๋ฐฉ๋ฒ•์˜ ํ•œ๊ณ„๋ฅผ ๊ทน๋ณตํ•˜๊ณ , ์ž„์ƒ์  ์šฐ์„ ์ˆœ์œ„๋ฅผ ๋ฐ˜์˜ํ•œ๋‹ค๋Š” ์ ์—์„œ ์ตœ๊ทผ ๋‹ค์ˆ˜์˜ 2์ƒยท3์ƒ ์‹œํ—˜์— ์ฑ„ํƒ๋˜๊ณ  ์žˆ๋‹ค. ๋ฒ ์ด์ง€์•ˆ ์ ์‘ํ˜• ๋””์ž์ธ๊ณผ์˜ ๋ถˆ์ผ์น˜ : ๊ธฐ์กด ๋ฒ ์ด์ง€์•ˆ ๋””์ž์ธ(BOP2 ๋“ฑ)์€ ์›์‹œ ๋ฐ์ดํ„ฐ(์˜ˆ: ์ดํ•ญ, ์‹œ๊ฐ„โ€‘์ƒ์กด)์—์„œ ์ง์ ‘ ์‚ฌํ›„ ํ™•๋ฅ ์„ ์ถ”์ •ํ•œ๋‹ค. WR์€ ์Œ๋ณ„ ๋น„๊ต ๊ฒฐ๊ณผ์˜ ๋น„์œจ์ด๋ฏ€๋กœ, ๋™์ผํ•œ WR ๊ฐ’์„ ์ƒ์„ฑํ•  ์ˆ˜ ์žˆ๋Š” ๋ฌด์ˆ˜ํžˆ ๋งŽ์€ ๋ฐ์ดํ„ฐ ์ƒ์„ฑ ๋ฉ”์ปค๋‹ˆ์ฆ˜์ด ์กด์žฌํ•œ๋‹ค. ์ด๋Š” ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ๊ธฐ๋ฐ˜ ์„ค๊ณ„ ์ตœ์ ํ™”์— ํฐ ์žฅ์• ๊ฐ€ ๋œ๋‹ค. 2.

Statistics Model
No Image

BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ EEG ํŠน์„ฑ : ๋ฐ€๋ฆฌ์ดˆ ์ˆ˜์ค€์˜ ์‹œ๊ฐ„ ํ•ด์ƒ๋„์™€ ์ € SNR, ๋น„์ •์ƒ์„ฑ, ํ”ผํ—˜์ž ๊ฐ„ ๋ณ€์ด์„ฑ์œผ๋กœ ์ธํ•ด ๊ธฐ์กด ๋”ฅ๋Ÿฌ๋‹ ๋ชจ๋ธ์ด ์ผ๋ฐ˜ํ™”์— ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค. ๊ธฐ์กด SSL ์ ‘๊ทผ : BENDR(๋Œ€์กฐ ํ•™์Šต), LaBraM, Brainโ€‘BERT, Brant ๋“ฑ์€ ์ฃผ๋กœ ๋‹จ์ผ ๋„๋ฉ”์ธ(์‹œ๊ฐ„ ํ˜น์€ ์ฃผํŒŒ์ˆ˜) ํ† ํฌ๋‚˜์ด์ € ์™€ ๋‹จ์ผ ๋ ˆ๋ฒจ VQ ๋ฅผ ์‚ฌ์šฉํ•ด ์ •๋ณด ์†์‹ค์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ํ•ต์‹ฌ ๋ฌธ์ œ : (1) ์ŠคํŽ™ํŠธ๋กœโ€‘ํ…œํฌ๋Ÿด ๊ฒฐํ•ฉ ๋ฏธํฌ์ฐฉ, (2) ์–‘์žํ™” ์šฉ๋Ÿ‰ ์ œํ•œ ๋ฐ ์ž”์ฐจ ์ฝ”๋“œ ๊ฐ„ ๋…๋ฆฝ์„ฑ ์œผ๋กœ ์ธํ•œ ๋น„ํšจ์œจ์  ํ•™์Šต. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๊ธฐ์—ฌ | ํ•ต์‹ฌ

Model Electrical Engineering and Systems Science
No Image

Can a Small Model Learn to Look Before It Leaps? Dynamic Learning and Proactive Correction for Hallucination Detection

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ํ™˜๊ฐ ๋ฌธ์ œ : LLM์ด ์‚ฌ์‹ค๊ณผ ๋‹ค๋ฅธ ์ •๋ณด๋ฅผ ์ƒ์„ฑํ•˜๋Š” ํ˜„์ƒ์€ ์‹ ๋ขฐ์„ฑ ์ €ํ•˜์™€ ์•ˆ์ „ ์œ„ํ—˜์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ๊ธฐ์กด ์ ‘๊ทผ๋ฒ•์˜ ํ•œ๊ณ„ : 1. ๊ณ ์ • ๊ฒ€์ฆ ์ „๋žต โ†’ ํ™˜๊ฒฝ ๋ณ€ํ™”์— ์ทจ์•ฝ, ๋„๊ตฌ ํ˜ธ์ถœ ์ตœ์ ํ™” ์–ด๋ ค์›€. 2. ํ์‡„ํ˜• LLM ํƒ์ง€๊ธฐ โ†’ ๋†’์€ ์ •ํ™•๋„์ง€๋งŒ ๋น„์šฉยท์ ‘๊ทผ์„ฑ ๋ฌธ์ œ. 3. ๊ต์‚ฌโ€‘ํ•™์ƒ ํŒŒ์ธํŠœ๋‹ โ†’ ๋น„์šฉ ์ ˆ๊ฐ์€ ๋˜์ง€๋งŒ ์ „๋žต ๊ณ ์ •์„ฑ ์œ ์ง€. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ LEAP ํ”„๋ ˆ์ž„์›Œํฌ | ๊ตฌ์„ฑ ์š”์†Œ | ์—ญํ•  | ์ฃผ์š” ๋ฉ”์ปค๋‹ˆ์ฆ˜ | | | | | | Teacher Model | ๋™์  ์ „๋žต ์ƒ์„ฑยท์กฐ์ • | ์‹คํ–‰ ์‹คํŒจ ์‹œ ์ „๋žต์„

Model Learning Detection
No Image

CombiGraph-Vis: A Curated Multimodal Olympiad Benchmark for Discrete Mathematical Reasoning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ LLM์˜ ์ฆ๋ช… ๋Šฅ๋ ฅ ๊ธ‰์„ฑ์žฅ : GPTโ€‘4, Gemini 2.5 Pro ๋“ฑ ์ตœ์‹  ๋ชจ๋ธ์ด IMO ์ˆ˜์ค€์˜ ์ฆ๋ช…์„ ์ž๋™ ์ƒ์„ฑํ•  ์ˆ˜ ์žˆ๊ฒŒ ๋œ ๊ฒƒ์€, ๊ธฐ์กด โ€œ์ •๋‹ต ์—ฌ๋ถ€โ€ ํ‰๊ฐ€๋ฅผ ๋„˜์–ด ์ฆ๋ช… ํ’ˆ์งˆ ์„ ์ •๋Ÿ‰ํ™”ํ•  ํ•„์š”์„ฑ์„ ์ œ๊ธฐํ•œ๋‹ค. ์ฑ„์  ์ž๋™ํ™”์˜ ํ•„์š”์„ฑ : ์˜ฌ๋ฆผํ”ผ์•„๋“œ์™€ ๊ฐ™์€ ๊ณ ๋‚œ์ด๋„ ์ˆ˜ํ•™ ๊ฒฝ์‹œ์—์„œ๋Š” ์ธ๊ฐ„ ์ฑ„์ ๊ด€์ด ์ˆ˜์‹ญ~์ˆ˜๋ฐฑ ๊ฑด์˜ ํ’€์ด๋ฅผ ์ผ๊ด€๋˜๊ฒŒ ํ‰๊ฐ€ํ•˜๊ธฐ ์–ด๋ ค์›Œ, ์ž๋™ ์ฑ„์  ์‹œ์Šคํ…œ์ด ๊ต์œกยท์—ฐ๊ตฌ ํ˜„์žฅ์—์„œ ํฐ ๊ฐ€์น˜๋ฅผ ์ œ๊ณตํ•œ๋‹ค. 2. ๋ฐ์ดํ„ฐ ๊ตฌ์„ฑ ๋ฐ ๋ผ๋ฒจ๋ง ์ „๋žต | ๋ฐ์ดํ„ฐ์…‹ | ์ถœ์ฒ˜ | ๊ทœ๋ชจ | ํ‰๊ฐ€ ์ฒ™๋„ | ๋ผ๋ฒจ๋ง ๋ฐฉ์‹ | | |

Comparing Baseline and Day-1 Diffusion MRI Using Multimodal Deep Embeddings for Stroke Outcome Prediction

Comparing Baseline and Day-1 Diffusion MRI Using Multimodal Deep Embeddings for Stroke Outcome Prediction

์ด ๋…ผ๋ฌธ์€ ๊ธ‰์„ฑ ํ—ˆํ˜ˆ์„ฑ ๋‡Œ์กธ์ค‘(AIS) ํ™˜์ž์˜ ์žฅ๊ธฐ ๊ธฐ๋Šฅ ํšŒ๋ณต์„ ์กฐ๊ธฐ์— ์˜ˆ์ธกํ•˜๊ธฐ ์œ„ํ•œ ์ƒˆ๋กœ์šด ๋‹ค์ค‘๋ชจ๋‹ฌ ์ธ๊ณต์ง€๋Šฅ ํŒŒ์ดํ”„๋ผ์ธ์„ ์ œ์‹œํ•œ๋‹ค. ๋จผ์ € ์—ฐ๊ตฌ ๋Œ€์ƒ์€ 74๋ช…์˜ AIS ํ™˜์ž๋กœ, ๊ฐ ํ™˜์ž๋Š” ์น˜๋ฃŒ ์ „(์‹œ๊ฐ„ 0, J0)๊ณผ ์น˜๋ฃŒ ํ›„ 24์‹œ๊ฐ„(J1) ๋‘ ์‹œ์ ์—์„œ ํ™•์‚ฐ MRI์˜ ํ‘œ์ค€ํ™”๋œ ADC ์˜์ƒ์„ ํš๋“ํ•˜์˜€๋‹ค. ์ด๋Ÿฌํ•œ ์ง์„ ์ด๋ฃฌ ์˜์ƒ ๋ฐ์ดํ„ฐ๋Š” 3์ฐจ์› ResNetโ€‘50 ๋„คํŠธ์›Œํฌ์— ์ž…๋ ฅ๋˜์–ด ๊ณ ์ฐจ์› ์ž„๋ฒ ๋”ฉ ๋ฒกํ„ฐ๋ฅผ ์ƒ์„ฑํ–ˆ์œผ๋ฉฐ, ์ด๋Š” ๊ธฐ์กด์˜ ์ „ํ†ต์ ์ธ ์˜์ƒ ํŠน์ง•(์˜ˆ: ๋ณ‘๋ณ€ ๋ถ€ํ”ผ)๊ณผ๋Š” ๋ณ„๊ฐœ์˜ ์ •๋ณด ํ๋ฆ„์„ ์ œ๊ณตํ•œ๋‹ค. ์ž„๋ฒ ๋”ฉ๊ณผ ์ž„์ƒ ๋ณ€์ˆ˜(์—ฐ๋ น, NIHSS

Image Processing Electrical Engineering and Systems Science
No Image

Computational TIRF enables optical sectioning beyond the evanescent field for widefield fluorescence microscopy

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์™€์ด๋“œํ•„๋“œ ํ˜„๋ฏธ๊ฒฝ์˜ ํ•œ๊ณ„ : ์ €์ถ•๋„(NA)์™€ ์–•์€ ๊นŠ์ด ๋•Œ๋ฌธ์— ์ดˆ์  ์™ธ ๋ฐฐ๊ฒฝ์ด ํฌ๊ฒŒ ์„ž์ด๋ฉฐ, ํŠนํžˆ ๋ผ๋ฒจ๋ง์ด ์กฐ๋ฐ€ํ•œ ์„ธํฌ๋‚˜ ์กฐ์ง์—์„œ ์‹ ํ˜ธโ€‘๋Œ€โ€‘์žก์Œ๋น„(SNR)๊ฐ€ ๊ธ‰๊ฒฉํžˆ ์ €ํ•˜๋œ๋‹ค. TIRF์˜ ์žฅ์ ๊ณผ ์ œ์•ฝ : ์ „๋ฐ˜์‚ฌ์— ์˜ํ•ด 100 ~ 200 nm ์ด๋‚ด์˜ ์–•์€ ํ‘œ๋ฉด๋งŒ์„ ์„ ํƒ์ ์œผ๋กœ ๊ด€์ฐฐํ•  ์ˆ˜ ์žˆ์–ด ๋ฐฐ๊ฒฝ ์–ต์ œ์— ๋›ฐ์–ด๋‚˜์ง€๋งŒ, ๊นŠ์ด ์ œํ•œ๊ณผ ํŠน์ˆ˜ ๊ด‘ํ•™ ๋ถ€ํ’ˆ(ํ”„๋ฆฌ์ฆ˜, ๊ณ  NA ๋ชฉํ‘œ ๋“ฑ)์ด ํ•„์š”ํ•˜๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ์ปดํ“จํ…Œ์ด์…”๋„ TIRF(cTIRF) ๋ฌผ๋ฆฌโ€‘์ธํฌ๋ฉ”๋“œ ๋”ฅ๋Ÿฌ๋‹ : ์ „ํ†ต์ ์ธ ๋ฐ์ดํ„ฐโ€‘์ฃผ๋„ ํ•™์Šต์ด ์•„๋‹Œ, ์‹ค์ œ ๊ด‘ํ•™

Conformational landscapes in cryo-ET data based on MD simulations

Conformational landscapes in cryo-ET data based on MD simulations

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ Cryoโ€‘ET์˜ ๊ณ ์œ  ๋ฌธ์ œ : ๋‚ฎ์€ SNR, missingโ€‘wedge, ๊ทธ๋ฆฌ๊ณ  ๋ณต์žกํ•œ ์„ธํฌ ๋‚ด ํ™˜๊ฒฝ์€ ์ „ํ†ต์ ์ธ โ€˜ํด๋ž˜์Šค ๋ถ„๋ฅ˜ โ†’ ํ‰๊ท โ€™ ํŒŒ์ดํ”„๋ผ์ธ์„ ์ œํ•œํ•œ๋‹ค. ํŠนํžˆ ์ž…์ž ์ˆ˜๊ฐ€ ์ ์€ ๊ฒฝ์šฐ, ์ด์‚ฐ์ ์ธ ํด๋ž˜์Šค ๊ตฌ๋ถ„์ด ๋ถˆ๊ฐ€๋Šฅํ•ด ์—ฐ์†์ ์ธ ๊ตฌ์กฐ ๋ณ€์ด๋ฅผ ํฌ์ฐฉํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ์—ฐ์†์  ๋ณ€์ด์˜ ์ค‘์š”์„ฑ : ์ตœ๊ทผ ๋‹ค์ˆ˜์˜ cryโ€‘ET ์—ฐ๊ตฌ๊ฐ€ โ€˜์—ฐ์†์ ์ธ ๋ณ€์ด(continuous heterogeneity)โ€™๋ฅผ ๋ณด๊ณ ํ•˜๊ณ  ์žˆ์œผ๋ฉฐ, ์ด๋Š” ๊ธฐ์กด ๊ณ ์ •๋œ ๊ตฌ์กฐ ๋ชจ๋ธ๋กœ๋Š” ์„ค๋ช…ํ•  ์ˆ˜ ์—†๋Š” ๊ธฐ๋Šฅ์ ยท๋™์—ญํ•™์  ์ •๋ณด๋ฅผ ๋‹ด๊ณ  ์žˆ๋‹ค. 2. MD ์‹œ๋ฎฌ๋ ˆ์ด์…˜์ด ์ œ

Data Quantitative Biology
No Image

Construction of a classification model for dementia among Brazilian adults aged 50 and over

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์‚ฌํšŒยท๊ฒฝ์ œ์  ๋งฅ๋ฝ : ๋ธŒ๋ผ์งˆ์€ ์ €์†Œ๋“ยท๊ณ ๋ถˆํ‰๋“ฑ ์‚ฌํšŒ๊ตฌ์กฐ์™€ ๊ณ ๋ น ์ธ๊ตฌ ๋น„์ค‘์ด ๊ธ‰์ฆํ•˜๊ณ  ์žˆ์–ด, ๊ณ ๊ฐ€์˜ ์˜์ƒยท๋ฐ”์ด์˜ค๋งˆ์ปค ๊ธฐ๋ฐ˜ ์ง„๋‹จ์ด ํ˜„์‹ค์ ์œผ๋กœ ์ œํ•œ๋œ๋‹ค. ๊ตญ์ œ์  ํŠธ๋ Œ๋“œ : ์ €ยท์ค‘์†Œ๋“ ๊ตญ๊ฐ€(LMIC)์—์„œ ์น˜๋งค ๋ฐœ์ƒ๋ฅ ์ด ๊ธ‰์ฆํ•˜๊ณ  ์žˆ์œผ๋ฉฐ, ์œ„ํ—˜ ์š”์ธ ์ˆ˜์ •์œผ๋กœ 45% ์ด์ƒ์˜ ์‚ฌ๋ก€๋ฅผ ์˜ˆ๋ฐฉ ๊ฐ€๋Šฅํ•˜๋‹ค๋Š” ๊ทผ๊ฑฐ๊ฐ€ ์žˆ๋‹ค. ์—ฐ๊ตฌ ๊ณต๋ฐฑ : ๊ธฐ์กด ์˜ˆ์ธก ๋ชจ๋ธ์€ ๋Œ€๋ถ€๋ถ„ ๊ณ ์†Œ๋“ ๊ตญ๊ฐ€ ๋ฐ์ดํ„ฐ์— ๊ธฐ๋ฐ˜ํ•˜๊ฑฐ๋‚˜, ๋ธŒ๋ผ์งˆ ๋‚ด์—์„œ๋„ ๊ณต๋ฌด์›ยท๊ณ ํ•™๋ ฅ ์ง‘๋‹จ์— ๊ตญํ•œ๋œ ์ž๋ฃŒ๋ฅผ ์‚ฌ์šฉํ•˜์˜€๋‹ค. ๋”ฐ๋ผ์„œ ์ „๊ตญ ๋Œ€ํ‘œ์„ฑ์„ ๊ฐ–์ถ˜ ๋ชจ๋ธ ๊ฐœ๋ฐœ์ด ์ ˆ์‹คํ–ˆ๋‹ค. 2. ์ž๋ฃŒยท๋ฐฉ๋ฒ• | ํ•ญ

Machine Learning Computer Science Model
No Image

Contour Integral Representations of Finite-part Integrals with Logarithmic Singularities

| ๊ตฌ๋ถ„ | ๋‚ด์šฉ | | | | | ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ | ์œ ํ•œ๋ถ€๋ถ„ ์ ๋ถ„(finiteโ€‘part integration)์€ ๋ฐœ์‚ฐ ์ ๋ถ„์„ โ€˜์œ ํ•œ ๋ถ€๋ถ„โ€™๋งŒ ์ถ”์ถœํ•ด ์˜๋ฏธ ์žˆ๋Š” ๊ฐ’์„ ๋ถ€์—ฌํ•˜๋Š” ๊ธฐ๋ฒ•์œผ๋กœ, ํŠนํžˆ Stieltjes ๋ณ€ํ™˜ยทMellin ๋ณ€ํ™˜ ๋“ฑ์—์„œ ํ•ต์‹ฌ ๋„๊ตฌ๋‹ค. ๊ธฐ์กด Galapon(2017)์˜ ๊ฒฐ๊ณผ๋Š” ๋กœ๊ทธ ํ•ญ์ด ์—†๋Š” ๊ฒฝ์šฐ์—๋งŒ ์ ์šฉ ๊ฐ€๋Šฅํ–ˆ์œผ๋ฉฐ, ๋กœ๊ทธ ํŠน์ด์„ฑ์€ ๋ฌผ๋ฆฌํ•™(์˜ˆ: Eulerโ€‘Heisenberg ๋ผ๊ทธ๋ž‘์ง€์•ˆ) ๋ฐ ํŠน์ˆ˜ ํ•จ์ˆ˜ ์ด๋ก ์—์„œ ์ž์ฃผ ๋‚˜ํƒ€๋‚œ๋‹ค. | | ์ฃผ์š” ๋ชฉํ‘œ | 1) ๋กœ๊ทธ ํŠน์ด์„ฑ์„ ์ฐจ์ˆ˜ (n)๊นŒ์ง€ ํฌํ•จํ•˜๋Š” ์œ ํ•œ๋ถ€๋ถ„ ์ ๋ถ„์— ๋Œ€ํ•œ ๋“ฑ๊ฐ€

Mathematics
No Image

Controlled dripping from a grooved condensing plate

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์ˆ˜์ง ํ‘œ๋ฉด์—์„œ ๋ฌผ์ด ํ๋ฅด๋Š” ํ˜•ํƒœ(๋ฐฉ์šธ, ๋ฆฌ๋น„์—ํŠธ, ํ•„๋ฆ„)๋Š” ํ‘œ๋ฉด ์นœ์ˆ˜์„ฑยท์œ ์†์— ๋”ฐ๋ผ ๋‹ค์–‘ํ•˜๊ฒŒ ๋‚˜ํƒ€๋‚œ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ํ•˜๋‹จ ๊ฐ€์žฅ์ž๋ฆฌ์—์„œ ๋ฌผ์ด โ€˜์–ด๋””์„œ, ์–ธ์ œโ€™ ๋–จ์–ด์ง€๋Š”์ง€๋Š” ๊ฑฐ์˜ ์—ฐ๊ตฌ๋˜์ง€ ์•Š์•˜๋‹ค. ์‘์ถ• ์‹œ์Šคํ…œ(๋””์›Œ ์ปฌ๋ ‰ํ„ฐ, ์—ด๊ตํ™˜๊ธฐ)์—์„œ๋Š” ์ด ํƒˆ๋ฝ ๊ณผ์ •์ด ์—ดยท๋ฌผ ์ „๋‹ฌ ํšจ์œจ ์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์นœ๋‹ค. ๊ธฐ์กด์—๋Š” ์ค‘๋ ฅยท์ถฉ๊ฒฉ์— ์˜์กดํ•œ ๋ฌด์ž‘์œ„ ํƒˆ๋ฝ์ด ์ผ๋ฐ˜์ ์ด์—ˆ์œผ๋ฉฐ, ์ด๋ฅผ ๊ธฐํ•˜ํ•™์  ๊ตฌ์กฐ ๋กœ ์ œ์–ดํ•  ๊ฐ€๋Šฅ์„ฑ์€ ์ถฉ๋ถ„ํžˆ ํƒ๊ตฌ๋˜์ง€ ์•Š์•˜๋‹ค. 2. ์‹คํ—˜ ์„ค๊ณ„ ๋ฐ ๋ฐฉ๋ฒ•๋ก  | ํ•ญ๋ชฉ | ๋‚ด์šฉ | | | | | ์‘์ถ• ๋ฐฉ์‹ | ๋”ฐ๋œปยท์Šตํ•œ ๊ณต๊ธฐ(โ‰ˆ

Physics
No Image

Data Fusion-Enhanced Decision Transformer for Stable Cross-Domain Generalization

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๊ต์ฐจโ€‘๋„๋ฉ”์ธ ์‹œํ”„ํŠธ : ์‹ค์ œ ๋กœ๋ด‡ยท์‹œ๋ฎฌ๋ ˆ์ด์…˜ ํ™˜๊ฒฝ์€ ๋ณด์ƒ ๊ตฌ์กฐ, ์‹œ๊ฐ„ horizon, ๋ฌผ๋ฆฌ ํŒŒ๋ผ๋ฏธํ„ฐ(์ค‘๋ ฅ, ๊ด€์ ˆ ๊ธธ์ด ๋“ฑ)์—์„œ ํฌ๊ฒŒ ๋‹ฌ๋ผ์ง„๋‹ค. DT๋Š” ์‹œํ€€์Šคโ€‘ํˆฌโ€‘์‹œํ€€์Šค ๋ชจ๋ธ์ด๋ฏ€๋กœ ์ž…๋ ฅ ํ† ํฐ(RTG, ์ƒํƒœ, ํ–‰๋™)์˜ ์—ฐ์†์„ฑ์ด ๊นจ์ง€๋ฉด ์„ฑ๋Šฅ์ด ๊ธ‰๊ฒฉํžˆ ์ €ํ•˜๋œ๋‹ค. ๊ธฐ์กด ๋ฐฉ๋ฒ•์˜ ํ•œ๊ณ„ : 1. ๋‹จ์ผ ํ•„ํ„ฐ๋ง ๊ธฐ์ค€ (์˜ˆ: ์ƒํƒœ ๊ตฌ์กฐ ์ผ์น˜ ํ˜น์€ ํ–‰๋™ ๊ฐ€๋Šฅ์„ฑ)๋งŒ ์‚ฌ์šฉ โ†’ ์„ ํƒ๋œ ์กฐ๊ฐ ๊ฐ„ ์Šคํ‹ฐ์น˜ ๋ถˆ์ผ์น˜ ๋ฐœ์ƒ. 2. RTG ํ† ํฐ ์‚ฌ์šฉ : ๋ณด์ƒยท์‹œ๊ฐ„ horizon์ด ๋ฐ”๋€Œ๋ฉด RTG ๊ฐ’ ์ž์ฒด๊ฐ€ ๋น„๊ต ๋ถˆ๊ฐ€๋Šฅํ•ด์ ธ ํ† ํฐ ์˜๋ฏธ๊ฐ€ ํŒŒํŽธํ™”๋œ

Data
No Image

Data-driven sequential analysis of tipping in high-dimensional complex systems

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ธฐ์กด์˜ Critical Slowing Down(CSD) ๊ธฐ๋ฐ˜ ์กฐ๊ธฐ๊ฒฝ๋ณด๋Š” ์ผ์ฐจ์› ๊ด€์ธก์— ์ตœ์ ํ™”๋ผ ์žˆ์–ด ๊ณ ์ฐจ์›ยท๋‹ค๋ณ€๋Ÿ‰ ์‹œ์Šคํ…œ์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ต๊ณ , ๊ฐ€์—ญ์  ๋ณ€ํ™”์—๋„ ๋ฏผ๊ฐํ•ด false positive ๋ฅผ ์ดˆ๋ž˜ํ•œ๋‹ค. ๋„คํŠธ์›Œํฌ ๊ธฐ๋ฐ˜ยทPCA/EOF ๊ธฐ๋ฐ˜ ๋ฐฉ๋ฒ•์€ ๊ตฌ์กฐ์  ์„ ํƒ(์ž„๊ณ„๊ฐ’, ์ฐจ์› ์ˆ˜) ์— ํฌ๊ฒŒ ์ขŒ์šฐ๋˜๋ฉฐ, ์‹ค์ œ ์ง€๊ตฌ ์‹œ์Šคํ…œ์—์„œ๋Š” ๊ด€์ธก ๊ฒฐํ•ยท๋…ธ์ด์ฆˆ๊ฐ€ ์‹ฌํ•ด ์‹ ๋ขฐ๋„๊ฐ€ ๋–จ์–ด์ง„๋‹ค. ๋จธ์‹ ๋Ÿฌ๋‹ ๊ธฐ๋ฐ˜ ์ ‘๊ทผ๋ฒ•์€ ํ•™์Šต ๋ฐ์ดํ„ฐ์™€ ๋ชฉํ‘œ ์‹œ์Šคํ…œ ๊ฐ„ ๋ถˆ์ผ์น˜ ๋ฌธ์ œ์— ์ทจ์•ฝํ•˜๋‹ค. ๋”ฐ๋ผ์„œ ํ•™์Šตโ€‘ํ”„๋ฆฌ, ๋‹ค๋ณ€๋Ÿ‰, ๊ด€์ธกยท๋…ธ์ด์ฆˆ์— ๊ฐ•์ธ ํ•œ ํ‹ฐํ•‘ ์ง€ํ‘œ๊ฐ€

System Analysis Physics Data
No Image

Decoupled Internal Energy Regulation and Inertial Response Provision for Grid-Forming Multilevel-Converter-Based E-STATCOMs

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ €๊ด€์„ฑยท๊ณ ์žฌ์ƒ ์ „๋ ฅ๊ณ„ํ†ต : ์žฌ์ƒ์—๋„ˆ์ง€์˜ ๊ฐ„ํ—์„ฑ๊ณผ ์˜ˆ์ธก๋ถˆ๊ฐ€๋Šฅ์„ฑ์œผ๋กœ ์ „๋ ฅ ๊ณต๊ธ‰ยท์ˆ˜์š” ๋ถˆ์ผ์น˜๊ฐ€ ๋นˆ๋ฒˆํ•ด์ง€๊ณ , ๊ด€์„ฑ ๊ฐ์†Œ๋Š” ์ „์••ยท์ฃผํŒŒ์ˆ˜ ๊ธ‰๋ณ€์„ ์ดˆ๋ž˜ํ•œ๋‹ค. Eโ€‘STATCOM์˜ ์—ญํ•  : STATCOM์— ์—๋„ˆ์ง€ ์ €์žฅ(๋ฐฐํ„ฐ๋ฆฌยท์Šˆํผ์ปคํŒจ์‹œํ„ฐ ๋“ฑ)์„ ๊ฒฐํ•ฉํ•ด ์ „์•• ๋ณด์กฐ์™€ ๋™์‹œ์— ๊ทธ๋ฆฌ๋“œโ€‘ํฌ๋ฐ(GFM) ์ œ์–ด๋ฅผ ํ†ตํ•ด ๊ด€์„ฑ ์‘๋‹ต์„ ์ œ๊ณตํ•œ๋‹ค. DSโ€‘MC ํ† ํด๋กœ์ง€์˜ ์žฅ์  : ๋‹ค๋‹จ๊ณ„ ๋ณ€ํ™˜๊ธฐ์˜ ๋ชจ๋“ˆ์‹ ๊ตฌ์กฐ๋Š” ๊ณ ์ „์••ยท๊ณ ์ „๋ ฅ ์ ์šฉ์— ์œ ๋ฆฌํ•˜๊ณ , ์ค‘์•™ DCโ€‘๋งํฌ์— ์ €์žฅ์†Œ๋ฅผ ๋ฐฐ์น˜ํ•˜๋ฉด ์—ด๊ด€๋ฆฌยท์œ ์ง€๋ณด์ˆ˜๊ฐ€ ์šฉ์ดํ•ด์ง„๋‹ค. 2. ๊ธฐ์กด ์ œ์–ด ๋ฐฉ์‹์˜ ํ•œ๊ณ„ | ๊ตฌ๋ถ„

Electrical Engineering and Systems Science
No Image

Deep Unfolded BM3D: Unrolling Non-local Collaborative Filtering into a Trainable Neural Network

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์ €์„ ๋Ÿ‰ CT ๋Š” ํ™˜์ž ๋ฐฉ์‚ฌ์„  ํ”ผํญ์„ ๊ฐ์†Œ์‹œํ‚ค์ง€๋งŒ, ๊ฐ•ํ•œ ์–‘์ž ์žก์Œ์œผ๋กœ ์ธํ•ด ์˜์ƒ ํ’ˆ์งˆ์ด ํฌ๊ฒŒ ์ €ํ•˜๋œ๋‹ค. ์ „ํ†ต์  ๋ชจ๋ธ ๊ธฐ๋ฐ˜ ๋ฐฉ๋ฒ•(BM3D, TV, NLM ๋“ฑ)์€ ๋ช…์‹œ์ ์ธ ์ด๋ฏธ์ง€ ์‚ฌ์ „(์ž๊ธฐ์œ ์‚ฌ์„ฑ, ์ŠคํŒŒ์Šค์„ฑ ๋“ฑ)์„ ์ด์šฉํ•ด ํ•ด์„ ๊ฐ€๋Šฅํ•˜์ง€๋งŒ ํŒŒ๋ผ๋ฏธํ„ฐ ํŠœ๋‹์— ์˜์กดํ•œ๋‹ค. ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ๋ฐ˜ ๋ฐฉ๋ฒ•(DnCNN, GAN, Uโ€‘Net ๋“ฑ)์€ ๋ฐ์ดํ„ฐ์— ๋งž์ถฐ ์ž๋™์œผ๋กœ ์ตœ์ ํ™”๋˜์ง€๋งŒ โ€œ๋ธ”๋ž™๋ฐ•์Šคโ€ ํŠน์„ฑ, ๋Œ€๊ทœ๋ชจ ๋ผ๋ฒจ๋ง ํ•„์š”, ๊ทธ๋ฆฌ๊ณ  ๋…ธ์ด์ฆˆ ๋ ˆ๋ฒจ๋งˆ๋‹ค ์žฌํ•™์Šต์ด ํ•„์š”ํ•˜๋‹ค๋Š” ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ Deep Unfolding ์ ์šฉ Dee

Network
No Image

Dense and Diverse Goal Coverage in Multi Goal Reinforcement Learning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๊ธฐ์กด RL์˜ ํ•œ๊ณ„ : ๋Œ€๋ถ€๋ถ„์˜ RL ์—ฐ๊ตฌ๋Š” ๋‹จ์ผ ์ตœ์  ๋ชฉํ‘œ(์˜ˆ: ํŠน์ • ๋ณด์ƒ ์ตœ๋Œ€ํ™”)์— ์ง‘์ค‘ํ•œ๋‹ค. ์ด๋Š” ์ •์ฑ…์ด ํŠน์ • ๋ณด์ƒ ์›์ฒœ์— ํŽธํ–ฅ๋˜๋Š” ํ˜„์ƒ์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ๋ชฉํ‘œ ๋‹ค์–‘์„ฑ ํ•„์š”์„ฑ : ๋กœ๋ด‡ ํƒ์‚ฌ, ๊ฒŒ์ž„ ๋ ˆ๋ฒจ ๋””์ž์ธ, ๋ฉ€ํ‹ฐโ€‘์—์ด์ „ํŠธ ํ˜‘์—… ๋“ฑ์—์„œ๋Š” ๋‹ค์–‘ํ•œ ๋ชฉํ‘œ ๋ฅผ ๊ณ ๋ฅด๊ฒŒ ๋‹ฌ์„ฑํ•˜๋Š” ๊ฒƒ์ด ์‹œ์Šคํ…œ์˜ ์•ˆ์ •์„ฑยท์•ˆ์ „์„ฑ์— ๊ธฐ์—ฌํ•œ๋‹ค. Multiโ€‘Goal RL ์ •์˜ : ๋ชฉํ‘œ ์ƒํƒœ๋ฅผ ํŒ๋ณ„ํ•˜๋Š” ์˜ค๋ผํด ๋ถ„๋ฅ˜๊ธฐ (state โ†’ goal / nonโ€‘goal) ์กด์žฌ๋ฅผ ๊ฐ€์ •ํ•˜๊ณ , โ€œ๋ชฉํ‘œ ์ƒํƒœ๋ฅผ ๊ท ๋“ฑํžˆ ๋ฐฉ๋ฌธํ•˜๋ฉด์„œ ๊ธฐ๋Œ€ ๋ณด์ƒ์„ ์ตœ๋Œ€ํ™”โ€ํ•˜๋Š” ๋ฌธ์ œ

Learning
No Image

DensiCrafter: Physically-Constrained Generation and Fabrication of Self-Supporting Hollow Structures

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ 3D ์ƒ์„ฑ ๋ชจ๋ธ์˜ ํ•œ๊ณ„ : ํ˜„์žฌ ํ…์ŠคํŠธยท์ด๋ฏธ์ง€ โ†’ 3D ๋ณ€ํ™˜ ํŒŒ์ดํ”„๋ผ์ธ์€ ์‹œ๊ฐ์  ํ’ˆ์งˆ์— ์ง‘์ค‘ํ•˜์ง€๋งŒ, ๋ฌผ๋ฆฌ์  ๊ฐ•๋„ยท์ œ์กฐ ๊ฐ€๋Šฅ์„ฑ์€ ๋ฌด์‹œ๋˜๋Š” ๊ฒฝ์šฐ๊ฐ€ ๋งŽ๋‹ค. ๊ฒฝ๋Ÿ‰ยท์ž์ฒด ์ง€์ง€ ๊ตฌ์กฐ : ํ•ญ๊ณตยท์ž๋™์ฐจยท์†Œํ˜• ๋กœ๋ด‡ ๋“ฑ ๊ฒฝ๋Ÿ‰ํ™”๊ฐ€ ํ•ต์‹ฌ์ธ ๋ถ„์•ผ์—์„œ, ๋‚ด๋ถ€๊ฐ€ ๋น„์–ด ์žˆ์œผ๋ฉด์„œ๋„ ํ”„๋ฆฐํŒ… ์ค‘ ์ž์ฒด ์ง€์ง€๋ฅผ ์œ ์ง€ํ•˜๋Š” ์„ค๊ณ„๋Š” ๋น„์šฉยท์‹œ๊ฐ„ ์ ˆ๊ฐ์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์นœ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ๋ฐ€๋„ ํ•„๋“œ ์ตœ์ ํ™” : ๊ธฐ์กด Trellis๊ฐ€ ์ถœ๋ ฅํ•˜๋Š” coarse voxel์„ โ€œ์—ฐ์† ๋ฐ€๋„ ํ•„๋“œโ€๋กœ ํ•ด์„ํ•˜๊ณ , ์ด๋ฅผ ๋ฏธ๋ถ„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ๋‹ค๋ฃจ์–ด ์ตœ์ ํ™”ํ•œ๋‹ค. ์„ธ ๊ฐ€์ง€

No Image

Design of low-energy transfers in cislunar space using sequences of lobe dynamics

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์‹œ์Šฌ๋ฃจ๋‚˜ ๊ณต๊ฐ„ ์€ Artemis ํ”„๋กœ๊ทธ๋žจ๊ณผ ์†Œํ˜• ์œ„์„ฑ ์ž„๋ฌด ์ฆ๊ฐ€๋กœ ๊ธ‰๊ฒฉํžˆ ๊ด€์‹ฌ์ด ๋†’์•„์ง€๊ณ  ์žˆ๋‹ค. ๊ธฐ์กด ๊ณ ์† ์ „์ด๋Š” ์—ฐ๋ฃŒ ์†Œ๋ชจ๊ฐ€ ํฌ๊ณ , ์ €์—๋„ˆ์ง€ ์ „์ด๋Š” ๋น„ํ–‰ ์‹œ๊ฐ„์ด ๊ธธ์–ด์ง€๋Š” tradeโ€‘off๊ฐ€ ์กด์žฌํ•œ๋‹ค. ๋‹ค์ฒด ํ™˜๊ฒฝ์€ ์ดˆ๊ธฐ ์กฐ๊ฑด์— ๋งค์šฐ ๋ฏผ๊ฐํ•˜๋ฏ€๋กœ, ํ˜ผ๋ˆ ๋™์—ญํ•™ ์„ ํ™œ์šฉํ•œ ์„ค๊ณ„๊ฐ€ ํ•„์ˆ˜์ ์ด๋‹ค. 2. ๊ธฐ์กด ์ ‘๊ทผ๋ฒ•๊ณผ ํ•œ๊ณ„ | ์ ‘๊ทผ๋ฒ• | ํŠน์ง• | ์žฅ์  | ๋‹จ์  | | | | | | | ์ˆ˜์น˜ ์ตœ์ ํ™” (์˜ˆ: ์ „์—ญ ํƒ์ƒ‰, ๋‹ค์ค‘ ์‚ฌ๊ฒฉ) | ์ดˆ๊ธฐ ์กฐ๊ฑด์„ ๋งŽ์ด ์ƒ˜ํ”Œ๋งํ•ด ํŒŒ๋ ˆํ†  ์ตœ์  ํ•ด ๋„์ถœ | ๋‹ค์–‘ํ•œ ์—ฐ๋ฃŒโ€‘์‹œ๊ฐ„ ์กฐํ•ฉ ์ œ๊ณต | ์ดˆ

Nonlinear Sciences
No Image

Device-Centric ISAC for Exposure Control via Opportunistic Virtual Aperture Sensing

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ MPE ๊ทœ์ œ์™€ ํ˜„์žฌ ์†”๋ฃจ์…˜ : FCCยทETSI ๋“ฑ์€ 6 GHz ์ด์ƒ์—์„œ ์ „ํŒŒ ์ „๋ ฅ ๋ฐ€๋„๋ฅผ ์ œํ•œํ•œ๋‹ค. ํ˜„์žฌ ์Šค๋งˆํŠธํฐ์€ 25 mm ์ดํ•˜๊ฐ€ ๋˜๋ฉด ์ „๋ ฅ์„ ๊ธ‰๊ฒฉํžˆ ๋‚ฎ์ถ”๋Š” ์ด์ง„ ๊ทผ์ ‘ ์„ผ์„œ๋งŒ ํƒ‘์žฌํ•˜๊ณ  ์žˆ์–ด, ์‹ค์ œ ๊ฑฐ๋ฆฌ์™€ ๋ฌด๊ด€ํ•˜๊ฒŒ ์ตœ์•… ์ƒํ™ฉ์„ ๊ฐ€์ •ํ•œ๋‹ค. ์ด๋Š” mmWave 5G/6G์—์„œ ๋งํฌ ๋งˆ์ง„์„ ํฌ๊ฒŒ ๊ฐ์†Œ ์‹œํ‚จ๋‹ค. ISAC์˜ ๋“ฑ์žฅ : 6G ์‹œ๋Œ€์— ํ†ต์‹ ยท์„ผ์‹ฑ ํŒŒํ˜•ยท์ŠคํŽ™ํŠธ๋Ÿผยทํ•˜๋“œ์›จ์–ด๋ฅผ ๊ณต์œ ํ•˜๋Š” ISAC์ด ํ™œ๋ฐœํžˆ ์—ฐ๊ตฌ๋˜๊ณ  ์žˆ๋‹ค. ๋Œ€๋ถ€๋ถ„์€ ๋„คํŠธ์›Œํฌโ€‘์ค‘์‹ฌ (๊ธฐ์ง€๊ตญ ๊ธฐ๋ฐ˜) ํ˜น์€ ์–‘๋ฐฉํ–ฅ ์ฐจ๋Ÿ‰ ์‹œ๋‚˜๋ฆฌ์˜ค์— ์ดˆ์ ์„ ๋งž์ถ”์—ˆ์œผ๋ฉฐ, ํ•ธ๋“œํ—ฌ๋“œ

Electrical Engineering and Systems Science
No Image

Diffolio: A Diffusion Model for Multivariate Probabilistic Financial Time-Series Forecasting and Portfolio Construction

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‹ค๋ณ€๋Ÿ‰ ๊ธˆ์œต ์‹œ๊ณ„์—ด ์€ ๊ฐœ๋ณ„ ์ž์‚ฐ ๊ฐ„์˜ ์ƒ๊ด€๊ด€๊ณ„๊ฐ€ ํฌํŠธํด๋ฆฌ์˜ค ์œ„ํ—˜ยท์ˆ˜์ต์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋ฏ€๋กœ, ๋‹จ์ผ ์‹œ๊ณ„์—ด ์˜ˆ์ธก๋ณด๋‹ค ๋ณต์žก๋„๊ฐ€ ๋†’๋‹ค. ๊ธฐ์กด ํ™•๋ฅ  ์˜ˆ์ธก ๋ชจ๋ธ(์˜ˆ: ๋ฒ ์ด์ง€์•ˆ VAR, GARCH, ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ๋ฐ˜ ์‹œ๊ณ„์—ด ๋ชจ๋ธ)์€ ๊ต์ฐจ์„น์…˜ ์ƒ๊ด€๊ตฌ์กฐ ๋ฅผ ์ถฉ๋ถ„ํžˆ ํฌ์ฐฉํ•˜์ง€ ๋ชปํ•˜๊ฑฐ๋‚˜, ํ™•๋ฅ  ๋ถ„ํฌ๋ฅผ ์ •๋ฐ€ํžˆ ๋ชจ๋ธ๋งํ•˜๋Š” ๋ฐ ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค. ๋”ฐ๋ผ์„œ ๋””ํ“จ์ „ ๋ชจ๋ธ ์„ ํ™œ์šฉํ•ด ์‹œ๊ณ„์—ด ์ „์ฒด์˜ ํ™•๋ฅ ์  ํ๋ฆ„์„ ๋‹จ๊ณ„๋ณ„๋กœ ๋ณต์›ํ•˜๋Š” ์ ‘๊ทผ๋ฒ•์€ ์ƒˆ๋กœ์šด ๊ฐ€๋Šฅ์„ฑ์„ ์ œ์‹œํ•œ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ๋ฒ• | ์š”์†Œ | ์„ค๋ช… | ๊ธฐ๋Œ€ ํšจ๊ณผ | | | | | | ๋””๋…ธ์ด์ง•

Model
No Image

Distributionally Robust Scheduling of Electrified Heating Under Heat Demand Forecast Uncertainty

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ 1. ์ „๊ธฐํ™” ๋‚œ๋ฐฉ์˜ ์‹œ์žฅ์  ์˜๋ฏธ ์ „๊ธฐ ๋ณด์ผ๋ŸฌยทํžˆํŠธํŽŒํ”„๋Š” ์ „๋ ฅ ์ˆ˜์š”๋ฅผ ์‹œ๊ฐ„๋Œ€๋ณ„๋กœ ์ด๋™์‹œํ‚ฌ ์ˆ˜ ์žˆ์–ด ์žฌ์ƒ์—๋„ˆ์ง€์™€์˜ ์—ฐ๊ณ„๊ฐ€ ์šฉ์ดํ•˜๊ณ , ์ „๋ ฅ ์‹œ์Šคํ…œ์˜ ์•ˆ์ •์„ฑ์— ๊ธฐ์—ฌํ•œ๋‹ค. 2. ๊ธฐ์กด ์Šค์ผ€์ค„๋ง์˜ ํ•œ๊ณ„ ๋Œ€๋ถ€๋ถ„์˜ ์—ฐ๊ตฌ๋Š” ์ ์˜ˆ์ธก ์— ์˜์กดํ•˜๊ฑฐ๋‚˜, ๊ณผ๊ฑฐ ์˜ค์ฐจ๋ฅผ ์ •๊ทœ๋ถ„ํฌ ๋“ฑ ๊ณ ์ •๋œ ํ˜•ํƒœ๋กœ ๊ฐ€์ •ํ•œ๋‹ค. ์‹ค์ œ ํ˜„์žฅ์—์„œ๋Š” ๋ฐ์ดํ„ฐ ๋ถ€์กฑ , ๊ณ„์ ˆยท๊ธฐํ›„ ๋ณ€๋™ , ์˜ˆ์ธก ๋ชจ๋ธ ์™ธ๋ถ€ ์ œ๊ณต ๋“ฑ์œผ๋กœ ๋ถ„ํฌ๊ฐ€ ๋ณ€ํ•  ๊ฐ€๋Šฅ์„ฑ์ด ํฌ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ๋ฒ• ๋ฐ ๋ชจ๋ธ๋ง | ์š”์†Œ | ์„ค๋ช… | ์žฅ์  | | | | | | ์›Œํ„ฐ์Šคํ…Œ์ธ ๊ตฌํ˜• ๋ชจํ˜ธ์„ฑ ์ง‘ํ•ฉ | ๊ฒฝํ—˜์  ๋ถ„ํฌ๋ฅผ

Electrical Engineering and Systems Science
No Image

Dual-Process Scaffold Reasoning for Enhancing LLM Code Debugging

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ LLM ์ถ”๋ก ์˜ ์ด์ค‘ ์‹œ์Šคํ…œ ๋ชจ๋ธ : ์ธ๊ฐ„ ์ธ์ง€ ์ด๋ก ์—์„œ System 1(์ง๊ด€ยท๋น ๋ฅธ ํŒ๋‹จ)๊ณผ System 2(๋…ผ๋ฆฌยท๋А๋ฆฐ ํŒ๋‹จ)๋ฅผ ์ฐจ์šฉํ•ด LLM์˜ ์ถœ๋ ฅ๊ณผ ์ค‘๊ฐ„ ๊ณผ์ •์— ์ ์šฉ. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ System 1์— ์ดˆ์ ์„ ๋งž์ถ”์–ด โ€œํ•œ ๋ฒˆ์— ๋‹ต์„ ๋‚ด๋Š”โ€ ๋ฐฉ์‹์— ๋จธ๋ฌผ๋ €์Œ. System 2(์‹ฌ์ธต ์ถ”๋ก )์˜ ๋ถ€์žฌ : ๋ณต์žกํ•œ ์ฝ”๋“œ ๋””๋ฒ„๊น…์€ ๋‹จ์ˆœํžˆ ์ •๋‹ต์„ ์ œ์‹œํ•˜๋Š” ๊ฒƒ๋ณด๋‹ค ๋ฌธ์ œ ๊ตฌ์กฐ ํŒŒ์•…ยท์›์ธ ๋ถ„์„ยทํ•ด๊ฒฐ ์ „๋žต ์„ค๊ณ„ ๊ฐ€ ํ•„์š”ํ•˜๋ฏ€๋กœ, System 2 ์ˆ˜์ค€์˜ ๋‹จ๊ณ„์  ์‚ฌ๊ณ ๊ฐ€ ํ•„์ˆ˜์ ์ž„์„ ๊ฐ•์กฐ. 2. ์ œ์•ˆ ํ”„๋ ˆ์ž„์›Œํฌ ๊ตฌ์กฐ | ํ๋ฆ„ | ์—ญํ• 

No Image

e1: Learning Adaptive Control of Reasoning Effort

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ถ”๋ก  ๋น„์šฉ๊ณผ ์ •ํ™•๋„์˜ ํŠธ๋ ˆ์ด๋“œ์˜คํ”„ ๋Š” ํ˜„์žฌ LLM ํ™œ์šฉ์—์„œ ํ•ต์‹ฌ ์ด์Šˆ๋‹ค. ํŠนํžˆ ์‹ค์‹œ๊ฐ„ ์„œ๋น„์Šค๋‚˜ ๋น„์šฉ ์ œํ•œ์ด ์žˆ๋Š” ํ™˜๊ฒฝ์—์„œ๋Š” ๋ถˆํ•„์š”ํ•˜๊ฒŒ ๊ธด CoT๋ฅผ ์ƒ์„ฑํ•˜๋Š” ๊ฒƒ์ด ๋น„ํšจ์œจ์ ์ด๋‹ค. ๊ธฐ์กด โ€œํ† ํฐ ์ˆ˜ ๊ณ ์ •โ€ ๋ฐฉ์‹์€ ๋ฌธ์ œ ๋‚œ์ด๋„ ์˜ˆ์ธก ์ด๋ผ๋Š” ์ „์ œ์— ์˜์กดํ•œ๋‹ค. ์‚ฌ์šฉ์ž๋Š” ์‚ฌ์ „ ์ง€์‹์ด ์—†์„ ๊ฒฝ์šฐ ๊ณผ์†Œยท๊ณผ๋‹ค ํ• ๋‹น ์œ„ํ—˜์ด ํฌ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด Adaptive Effort Control (AEC) : โ€œํ˜„์žฌ ํ‰๊ท  CoT ๊ธธ์ด ๋Œ€๋น„ ๋น„์œจโ€์ด๋ผ๋Š” ์ƒ๋Œ€์  ํ† ํฐ ์˜ˆ์‚ฐ์„ ์‚ฌ์šฉํ•œ๋‹ค. ๊ฐ•ํ™”ํ•™์Šต(RL) ํ”„๋ ˆ์ž„์›Œํฌ ๋‚ด์—์„œ effort ํŒŒ

Learning
No Image

Efficient Automated Diagnosis of Retinopathy of Prematurity by Customize CNN Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ROP ์€ ์กฐ์‚ฐ์•„์—์„œ ์‹œ๋ ฅ ์†์‹ค์„ ์ดˆ๋ž˜ํ•  ์ˆ˜ ์žˆ๋Š” ์‹ฌ๊ฐํ•œ ์•ˆ๊ณผ ์งˆํ™˜์ด๋ฉฐ, ์กฐ๊ธฐ ์ง„๋‹จ์ด ์น˜๋ฃŒ ์„ฑ๊ณต๋ฅ ์„ ํฌ๊ฒŒ ์ขŒ์šฐํ•œ๋‹ค. ๊ธฐ์กด ์ง„๋‹จ์€ ์•ˆ๊ณผ ์ „๋ฌธ์˜์˜ ์ฃผ๊ด€์  ํŒ๋…์— ์˜์กดํ•ด ์‹œ๊ฐ„ยท์ธ๋ ฅยท์˜ค์ฐจ ๋ฌธ์ œ๊ฐ€ ์กด์žฌํ•œ๋‹ค. ๋”ฅ๋Ÿฌ๋‹, ํŠนํžˆ CNN ์€ ์ด๋ฏธ์ง€ ๊ธฐ๋ฐ˜ ์งˆ๋ณ‘ ์ง„๋‹จ์—์„œ ๋†’์€ ์„ฑ๋Šฅ์„ ๋ณด์—ฌ ์™”์ง€๋งŒ, ์˜๋ฃŒ ํ˜„์žฅ ์ ์šฉ ์‹œ ์—ฐ์‚ฐ ๋น„์šฉยท์‹ค์‹œ๊ฐ„์„ฑ ์ด ํฐ ์žฅ์• ๋ฌผ์ด๋‹ค. 2. ๋ฐ์ดํ„ฐ์…‹ ๋ฐ ์ „์ฒ˜๋ฆฌ | ํ•ญ๋ชฉ | ๋‚ด์šฉ | | | | | ๋ฐ์ดํ„ฐ ์ถœ์ฒ˜ | ๋…ผ๋ฌธ์— ๊ตฌ์ฒด์  ์ถœ์ฒ˜๊ฐ€ ๋ช…์‹œ๋˜์ง€ ์•Š์•„ ์ถ”์ •: ๊ณต๊ฐœ ROP ์ด๋ฏธ์ง€ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค(์˜ˆ: ROPBase ,

Model
No Image

Efficient Test-Time Retrieval Augmented Generation

| ํ•ญ๋ชฉ | ๋‚ด์šฉ ๋ฐ ํ‰๊ฐ€ | | | | | ํ•ต์‹ฌ ์•„์ด๋””์–ด | ํ…Œ์ŠคํŠธ ์‹œ์ ์—๋งŒ ์ž‘๋™ํ•˜๋Š” RAG ํ”„๋ ˆ์ž„์›Œํฌ๋กœ, ์‚ฌ์ „ ํ•™์ŠตยทํŒŒ์ธํŠœ๋‹ ์—†์ด ๋ฐ”๋กœ ์ ์šฉ ๊ฐ€๋Šฅ.<br> โ€œ๋ถ€๋ถ„ ์ƒ์„ฑ + ๋‹ค์ˆ˜๊ฒฐโ€์ด๋ผ๋Š” ์ƒˆ๋กœ์šด ๋น„์šฉโ€‘ํšจ์œจ ํŠธ๋ฆญ์„ ๋„์ž…ํ•ด ์ „์ฒด ์ƒ์„ฑ ๋น„์šฉ์„ ํฌ๊ฒŒ ์ ˆ๊ฐ. | | ๋ฐฉ๋ฒ•๋ก  | 1. ๊ฒ€์ƒ‰ ๋‹จ๊ณ„ : ๊ธฐ์กด BM25/FAISS ๋“ฑ dense retriever ์‚ฌ์šฉ, ๊ฐ€์žฅ ๊ด€๋ จ์„ฑ ๋†’์€ k ๋ฌธ์„œ ์„ ํƒ.<br>2. ๋‹ค์–‘ํ•œ ํ›„๋ณด ์ƒ์„ฑ : LLM์— ๋™์ผ ์งˆ์˜์™€ ๊ฒ€์ƒ‰ ๋ฌธ์„œ๋ฅผ ์ œ๊ณตํ•˜๋˜, ์‘๋‹ต ๊ธธ์ด ์ œํ•œ (์˜ˆ: 20~30 ํ† ํฐ)์œผ๋กœ ์—ฌ๋Ÿฌ ๊ฐœ์˜ ์งง์€ ํ›„๋ณด๋ฅผ ๋น ๋ฅด

No Image

Elastic Architecture Search for Efficient Language Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํ˜„์žฌ ๊ฐ€์žฅ ๊ฐ•๋ ฅํ•œ NLU ์„ฑ๋Šฅ์„ ๋‚ด๋Š” ๋ชจ๋ธ๋“ค์€ ์ˆ˜์–ต~์ˆ˜์‹ญ์–ต ํŒŒ๋ผ๋ฏธํ„ฐ ๊ทœ๋ชจ์ด๋ฉฐ, ํ•™์Šตยท์ถ”๋ก  ์‹œ ์ „๋ ฅ ์†Œ๋น„์™€ ๋น„์šฉ์ด ๊ธ‰์ฆํ•œ๋‹ค. ๊ธฐ์กด NAS ๊ธฐ๋ฐ˜ ๊ฒฝ๋Ÿ‰ํ™” ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ ์ด๋ฏธ์ง€ ๋ถ„์•ผ์— ๊ตญํ•œ๋˜์—ˆ๊ณ , ์–ธ์–ด ๋ชจ๋ธ์— ์ ์šฉํ•  ๊ฒฝ์šฐ Transformer ํŠน์œ ์˜ ๊ตฌ์กฐ์  ์ œ์•ฝ (์˜ˆ: ๋‹ค์ค‘ ํ—ค๋“œ ์–ดํ…์…˜, ๋ ˆ์ด์–ด ์ •๊ทœํ™” ๋“ฑ)์„ ์ถฉ๋ถ„ํžˆ ๋ฐ˜์˜ํ•˜์ง€ ๋ชปํ•œ๋‹ค๋Š” ํ•œ๊ณ„๊ฐ€ ์žˆ๋‹ค. 2. ELM์˜ ํ•ต์‹ฌ ๊ธฐ์—ฌ ํƒ„๋ ฅ์  ๊ฒ€์ƒ‰ ๊ณต๊ฐ„ (Elastic Search Space) Efficient Transformer Blocks : ๊ธฐ์กด Transformer ๋ธ”

Model
No Image

Enabling Vibration-Based Gesture Recognition on Everyday Furniture via Energy-Efficient FPGA Implementation of 1D Convolutional Networks

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ง„๋™ ๊ธฐ๋ฐ˜ ์ œ์Šค์ฒ˜ ์ธ์‹์€ ์นด๋ฉ”๋ผยท๋งˆ์ดํฌ์™€ ๋‹ฌ๋ฆฌ ํ”„๋ผ์ด๋ฒ„์‹œ ์นจํ•ด ์œ„ํ—˜์ด ๋‚ฎ๊ณ , ๊ฐ€๊ตฌ์— ์ž์—ฐ์Šค๋Ÿฝ๊ฒŒ ํ†ตํ•ฉ๋  ์ˆ˜ ์žˆ๋‹ค๋Š” ์žฅ์ ์ด ์žˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ๊ธฐ์กด ์‹œ์Šคํ…œ์€ ๊ณ ์„ฑ๋Šฅ DSP/CPU ํ˜น์€ GPU๋ฅผ ํ•„์š”๋กœ ํ•˜์—ฌ ์‹ค์ œ ๊ฐ€์ •์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ค์› ๋‹ค. ์ €์ „๋ ฅ FPGA๋Š” ๋†’์€ ์—ฐ์‚ฐ ํšจ์œจ๊ณผ ์žฌ๊ตฌ์„ฑ ๊ฐ€๋Šฅ์„ฑ์„ ์ œ๊ณตํ•˜์ง€๋งŒ, ๋ฉ”๋ชจ๋ฆฌ์™€ ๋…ผ๋ฆฌ ์ž์›์ด ์ œํ•œ์ ์ด๋ฏ€๋กœ ๋ชจ๋ธ ๊ฒฝ๋Ÿ‰ํ™”๊ฐ€ ํ•„์ˆ˜์ ์ด๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์ˆ ์  ๊ธฐ์—ฌ Raw Waveform ์ž…๋ ฅ : ์ „ํ†ต์ ์ธ ์ŠคํŽ™ํŠธ๋Ÿผ ๋ณ€ํ™˜(FFT, ๋ฉœโ€‘์ŠคํŽ™ํŠธ๋กœ๊ทธ๋žจ ๋“ฑ)์„ ๋ฐฐ์ œํ•˜๊ณ  ์›์‹œ ์ง„๋™ ํŒŒํ˜•์„ ๊ทธ๋Œ€๋กœ ์‚ฌ์šฉํ•จ์œผ๋กœ์จ

Network
No Image

Environment-Aware Network-Level Design of Generalized Pinching-Antenna Systems--Part I: Traffic-Aware Case

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋„คํŠธ์›Œํฌโ€‘๋ ˆ๋ฒจ ๊ด€์  ์ „ํ™˜ : 5Gยท6G ์‹œ๋Œ€์— ์„œ๋น„์Šค ํ’ˆ์งˆ์€ ๋‹จ์ผ ๋งํฌ๊ฐ€ ์•„๋‹Œ ์ง€์—ญ ์ „์ฒด ์˜ ํ‰๊ท ยท์ตœ์•…โ€‘์ผ€์ด์Šค ์„ฑ๋Šฅ์œผ๋กœ ํ‰๊ฐ€๋œ๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ๊ณ ์ •๋œ ์ธํ”„๋ผ(์•ˆํ…Œ๋‚˜ ๊ธฐ์šธ๊ธฐยท์ „๋ ฅยท๋น”ํŒจํ„ด)๋งŒ์„ ์กฐ์ •ํ•ด ์™”์œผ๋ฉฐ, ์ด๋Š” ๊ณต๊ฐ„์  ์ž์œ ๋„ ๊ฐ€ ์ œํ•œ๋ผ ํŠธ๋ž˜ํ”ฝยทํ™˜๊ฒฝ ๋น„๊ท ์ผ์„ฑ์„ ์ถฉ๋ถ„ํžˆ ๋ณด์ •ํ•˜์ง€ ๋ชปํ•œ๋‹ค. ํ•€์นญ ์•ˆํ…Œ๋‚˜์˜ ์ฐจ๋ณ„์  : ํŒŒ๋™์„ ์ „๋„ ๋งค์ฒด(์œ ์ „์ฒด ํŒŒ์ดํ”„, LCX ๋“ฑ)๋กœ ์ „๋‹ฌํ•˜๊ณ , ์›ํ•˜๋Š” ์œ„์น˜์—์„œ ์„ ํƒ์ ์œผ๋กœ ๋ฐฉ์‚ฌํ•จ์œผ๋กœ์จ โ€œ์—๋„ˆ์ง€ ์ฃผ์ž…์ โ€ ์ž์ฒด๋ฅผ ์žฌ๋ฐฐ์น˜ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ๊ธฐ์กด ์ธํ”„๋ผ์˜ ๋ฌผ๋ฆฌ์  ๊ณ ์ •์„ฑ์„ ๋„˜์–ด์„œ๋Š” ๊ณต๊ฐ„ ์žฌ๊ตฌ์„ฑ

Network System Electrical Engineering and Systems Science
No Image

Environment-Aware Network-Level Design of Generalized Pinching-Antenna Systems--Part II: Geometry-Aware Case

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๋„คํŠธ์›Œํฌโ€‘๋ ˆ๋ฒจ ๊ด€์  : ๊ธฐ์กด ๋ฌด์„  ๋„คํŠธ์›Œํฌ ์„ค๊ณ„๋Š” โ€œ๊ฐ€์žฅ ๋น ๋ฅธ ๋งํฌโ€์— ์ดˆ์ ์„ ๋งž์ถ”์—ˆ์ง€๋งŒ, ์‹ค์ œ ์šด์˜์ž๋Š” ์ง€์—ญ ์ „์ฒด์— ๊ณ ๋ฅธ ์„œ๋น„์Šค ๋ฅผ ์š”๊ตฌํ•œ๋‹ค. ํŠนํžˆ ์‹ค๋‚ดยท๊ณต์žฅยท๋„์‹œ ํ˜‘๊ณก ๋“ฑ ์žฅ์• ๋ฌผ์ด ๋งŽ์€ ํ™˜๊ฒฝ์—์„œ๋Š” ๊ณ ์ •๋œ ๊ธฐ์ง€๊ตญ ์œ„์น˜๋งŒ์œผ๋กœ๋Š” ์˜๊ตฌ์ ์ธ ์ปค๋ฒ„๋ฆฌ์ง€ ๊ตฌ๋ฉ์„ ๋ฉ”์šฐ๊ธฐ ์–ด๋ ต๋‹ค. ํ•€์นญ ์•ˆํ…Œ๋‚˜์˜ ์ฐจ๋ณ„์  : ํŒŒ๋™๊ฐ€์ด๋“œ(์œ ์ „์ฒด, ๋ˆ„์„ค๋™์ถ•์ผ€์ด๋ธ” ๋“ฑ) ์œ„์— ์—ฌ๋Ÿฌ ํ›„๋ณด ์ง€์ ์„ ๋‘๊ณ , ์ „ํŒŒ ๋ฐฉ์ถœ/์ˆ˜์‹  ์œ„์น˜๋ฅผ ๋™์ ์œผ๋กœ ์žฌ๋ฐฐ์น˜ ํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ์ „ํ†ต์ ์ธ ๋น”ํฌ๋ฐ(๊ฐ๋„)๊ณผ๋Š” ๋ณ„๊ฐœ์˜ ๊ธฐํ•˜ํ•™์  ์ž์œ ๋„ ๋ฅผ ์ œ๊ณตํ•œ๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ

Network System Electrical Engineering and Systems Science
No Image

Evaporation of a freely floating droplet in an airstream: effects of temperature, humidity, and shape oscillations

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ธฐํ›„ยท๊ธฐ์ƒํ•™์  ์ค‘์š”์„ฑ : ๋น„๊ฐ€ ๋‚ด๋ฆฌ๋Š” ๊ณผ์ •์—์„œ ๋ฐฉ์šธ์ด ๊ฒช๋Š” ์ฆ๋ฐœ์€ ๊ตฌ๋ฆ„ ๋ฏธ์„ธ๋ฌผ๋ฆฌยท๊ฐ•์ˆ˜ ๊ฐ•๋„ยท์ž ์—ด ์žฌ๋ถ„๋ฐฐ์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์นœ๋‹ค. ๊ธฐ์กด ๋ชจ๋ธ์€ ์ •์ง€ ๊ตฌํ˜• ๋ฐฉ์šธ์„ ์ „์ œ๋กœ ํ•˜์—ฌ ์‹ค์ œ ๋‚™ํ•˜ยท๋Œ€๋ฅ˜ ํ™˜๊ฒฝ์„ ์ถฉ๋ถ„ํžˆ ๋ฐ˜์˜ํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์‚ฐ์—…์  ์‘์šฉ : ์Šคํ”„๋ ˆ์ด ์—ฐ์†Œ, ๊ฑด์กฐยท๋ƒ‰๊ฐ ์‹œ์Šคํ…œ ๋“ฑ์—์„œ๋„ ์ž์œ  ๋ถ€์–‘ ๋ฐฉ์šธ์˜ ์ฆ๋ฐœ ํŠน์„ฑ์ด ํ•ต์‹ฌ์ด์ง€๋งŒ, ๋Œ€๋ถ€๋ถ„์€ ๊ณ ์ •๋œ ๊ธฐํŒ ์œ„์˜ sessile droplet์„ ๋Œ€์ƒ์œผ๋กœ ์—ฐ๊ตฌ๋ผ ์™”๋‹ค. 2. ์‹คํ—˜ ์„ค๊ณ„์˜ ํ˜์‹ ์„ฑ | ์š”์†Œ | ๊ธฐ์กด ์—ฐ๊ตฌ | ๋ณธ ์—ฐ๊ตฌ | | | | | | ๋ถ€์–‘ ๋ฐฉ์‹ | ์Œํ–ฅ ๋ ˆ๋น„

Physics
No Image

Evidence-Bound Autonomous Research (EviBound): A Governance Framework for Eliminating False Claims

1. ์—ฐ๊ตฌ ๋™๊ธฐ์™€ ๊ธฐ์—ฌ ๋™๊ธฐ : ํ˜„์žฌ LLM ๊ธฐ๋ฐ˜ ์ž๋™ ์—ฐ๊ตฌ ์‹œ์Šคํ…œ์€ โ€œ์™„๋ฃŒโ€ ์‹ ํ˜ธ๊ฐ€ ์‹ค์ œ ์‚ฐ์ถœ๋ฌผ๊ณผ ์ผ์น˜ํ•˜์ง€ ์•Š๋Š” ๊ฒฝ์šฐ๊ฐ€ ๋นˆ๋ฒˆํ•ด, ์—ฐ๊ตฌ ์žฌํ˜„์„ฑ๊ณผ ์‹ ๋ขฐ์„ฑ์„ ํฌ๊ฒŒ ์ €ํ•ดํ•œ๋‹ค. ํ•ต์‹ฌ ๊ธฐ์—ฌ : 1. ์ด์ค‘ ๊ฒŒ์ดํŠธ ์„ค๊ณ„ โ€“ ์‚ฌ์ „ยท์‚ฌํ›„ ๊ฒ€์ฆ์„ ๊ฒฐํ•ฉํ•ด ํ—ˆ์œ„ ์ฃผ์žฅ์„ ๊ทผ๋ณธ์ ์œผ๋กœ ์ฐจ๋‹จ. 2. ์ฆ๊ฑฐ ๊ธฐ๋ฐ˜ ์ฃผ์žฅ ์ „ํŒŒ โ€“ Run ID์™€ ์‚ฐ์ถœ๋ฌผ ์กด์žฌ ์—ฌ๋ถ€๋ฅผ ๊ฐ•์ œํ•จ์œผ๋กœ์จ โ€œ์ฆ๊ฑฐ ์—†๋Š” ์ฃผ์žฅโ€์„ ์›์ฒœ ์ฐจ๋‹จ. 3. ๊ฒฝ๋Ÿ‰ํ™”๋œ ์žฌ์‹œ๋„ ๋ฉ”์ปค๋‹ˆ์ฆ˜ โ€“ ์ œํ•œ๋œ ์žฌ์‹œ๋„๋กœ ์ผ์‹œ์  ์˜ค๋ฅ˜๋ฅผ ๋ณต๊ตฌํ•˜๋ฉด์„œ๋„ ๋ฌดํ•œ ๋ฃจํ”„ ์œ„ํ—˜์„ ์ตœ์†Œํ™”. 4. ์‹คํ—˜ ์žฌํ˜„์„ฑ : ์ „์ฒด ์‹คํ–‰ ๊ถค์ ๊ณผ MLflow ๋ฉ”ํƒ€

Framework
No Image

Exploring the Utility of MALDI-TOF Mass Spectrometry and Antimicrobial Resistance in Hospital Outbreak Detection

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ WGS์˜ ํ•œ๊ณ„ : ๊ธˆ์ „์ ยท์‹œ๊ฐ„์  ๋น„์šฉ์ด ๋†’์•„ ์ผ์ƒ์ ์ธ ๊ฐ์—ผ๊ด€๋ฆฌ ํ˜„์žฅ์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ค์›€. ๋Œ€์•ˆ ํƒ์ƒ‰ : ์ด๋ฏธ ์ž„์ƒ์‹คํ—˜์‹ค์—์„œ ํ™œ์šฉ ์ค‘์ธ MALDIโ€‘TOF์™€ AR ๋ฐ์ดํ„ฐ๋Š” ๋ณ„๋„ ์ถ”๊ฐ€ ๋น„์šฉ ์—†์ด ํ™•๋ณด ๊ฐ€๋Šฅํ•˜๋ฏ€๋กœ, ์ด๋ฅผ ํ™œ์šฉํ•œ ๋น ๋ฅธ ๊ตฐ์ง‘ ํƒ์ง€๊ฐ€ ์‹ค์งˆ์ ์ธ ๊ฐ€์น˜๋ฅผ ๊ฐ€์ง. 2. ๋ฐ์ดํ„ฐ ๋ฐ ์‹คํ—˜ ์„ค๊ณ„ | ํ•ญ๋ชฉ | ๋‚ด์šฉ | | | | | ๋ฐ์ดํ„ฐ ๊ทœ๋ชจ | 4,921๊ฐœ ๊ท ์ฃผ, 17์ข…, 2021โ€‘10 ~ 2024โ€‘10 | | ์ž…๋ ฅ | ์›์‹œ MALDIโ€‘TOF ์ŠคํŽ™ํŠธ๋Ÿผ(2000โ€‘20,000 Da, 2,000 bins) + 71 ํ•ญ์ƒ

Detection Quantitative Biology
No Image

Far-field heat transfer and monochromatic thermal currents in a cylindrical nonreciprocal cavity

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์˜์˜ ํ‚ค๋ฅดํžˆํ˜ธํ”„ ๋ฒ•์น™์˜ ๋น„๊ฐ€์—ญ์„ฑ : ์ „ํ†ต์ ์ธ ํ‚ค๋ฅดํžˆํ˜ธํ”„ ๋ฒ•์น™์€ ๋ณต์‚ฌ ํก์ˆ˜์œจ(ฮฑ)๊ณผ ๋ฐฉ์ถœ์œจ(ฮต)์ด ๋™์ผํ•œ ๋ฐฉํ–ฅ์—์„œ ๋™์ผํ•˜๋‹ค๊ณ  ๊ฐ€์ •ํ•ฉ๋‹ˆ๋‹ค. ์ตœ๊ทผ ๋น„๊ฐ€์—ญ(๋น„์ž๊ธฐ๊ด‘ํ•™ยท์œ„์ƒ) ์žฌ๋ฃŒ์—์„œ ์ด ๊ฐ€์ •์ด ๊นจ์ง„๋‹ค๋Š” ์‹คํ—˜ยท์ด๋ก ์  ์ฆ๊ฑฐ๊ฐ€ ๋Š˜์–ด๋‚˜๋ฉด์„œ, ์—ด์—๋„ˆ์ง€ ํ๋ฆ„์„ โ€˜ํŽธํ–ฅโ€™์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š” ์ƒˆ๋กœ์šด ๋ฌผ๋ฆฌ์  ๋ฉ”์ปค๋‹ˆ์ฆ˜์ด ๋ถ€๊ฐ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ์›ํ†ตํ˜• ๊ณต๋™์˜ ์„ ํƒ ์ด์œ  : ์›ํ†ต์€ 1์ฐจ์› ํšŒ์ „ ๋Œ€์นญ์„ ๊ฐ€์ง€๊ณ  ์žˆ์–ด ์ˆœํ™˜ํ˜• ์—ด์ „๋ฅ˜(โ€˜์—ด ํšŒ์ „โ€™)๋ฅผ ์ง๊ด€์ ์œผ๋กœ ์‹œ๊ฐํ™”ํ•  ์ˆ˜ ์žˆ๋Š” ์ด์ƒ์ ์ธ ํ”Œ๋žซํผ์ด๋ฉฐ, ๋™์‹œ์— ํ‰๋ฉดยท๊ทผ๊ฑฐ๋ฆฌ(Nearโ€‘field) ์—ฐ๊ตฌ์™€ ์ฐจ๋ณ„ํ™”๋œ ์›๊ฑฐ๋ฆฌ

Physics

< Category Statistics (Total: 5056) >

Electrical Engineering and Systems Science
102
General
4147
General Relativity
3
HEP-EX
4
HEP-PH
3
HEP-TH
4
MATH-PH
6
Nonlinear Sciences
2
Quantum Physics
18

Start searching

Enter keywords to search articles

โ†‘โ†“
โ†ต
ESC
โŒ˜K Shortcut