KOINEU Logo
No Image

BiHDTrans: binary hyperdimensional transformer for efficient multivariate time series classification

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ IoT์™€ MTS ๋ฐ์ดํ„ฐ ํญ์ฆ : ์„ผ์„œ ๋„คํŠธ์›Œํฌ, ์Šค๋งˆํŠธ ํŒฉํ† ๋ฆฌ ๋“ฑ์—์„œ ์‹ค์‹œ๊ฐ„์œผ๋กœ ๋‹ค๋ณ€๋Ÿ‰ ์‹œ๊ณ„์—ด์ด ์ƒ์„ฑ๋ผ ์—ฃ์ง€ ๋””๋ฐ”์ด์Šค์—์„œ ๋น ๋ฅธ ๋ถ„๋ฅ˜๊ฐ€ ํ•„์ˆ˜. HD ์ปดํ“จํŒ…์˜ ์žฅ์  : ์ดˆ๊ณ ์ฐจ์› ์ด์ง„ ๋ฒกํ„ฐ๋ฅผ ์ด์šฉํ•ด ์—ฐ์‚ฐ์ด ๊ฐ„๋‹จํ•˜๊ณ  ํ•˜๋“œ์›จ์–ด ์นœํ™”์ ์ด๋ฉฐ, ๋ฉ”๋ชจ๋ฆฌ ๋Œ€์—ญํญ ์š”๊ตฌ๊ฐ€ ๋‚ฎ์Œ. ํŠธ๋žœ์Šคํฌ๋จธ์˜ ํ•œ๊ณ„ : ์…€ํ”„โ€‘์–ดํ…์…˜์€ ๋ณต์žกํ•œ ์‹œ๊ณ„์—ด ํŒจํ„ด์„ ์ž˜ ํฌ์ฐฉํ•˜์ง€๋งŒ, QยทKยทV ๋งคํŠธ๋ฆญ์Šค ์—ฐ์‚ฐ๊ณผ ์ •๊ทœํ™” ๊ณผ์ •์ด ๋น„์šฉ์ด ํฌ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ BiHDTrans | ์š”์†Œ | ๊ธฐ์กด ์ ‘๊ทผ | BiHDTrans์—์„œ์˜ ๊ตฌํ˜„ | | | | | | ํ‘œํ˜„

No Image

Block Empirical Likelihood Inference for Longitudinal Generalized Partially Linear Single-Index Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์žฅ๊ธฐ ๋ฐ์ดํ„ฐ๋Š” ํ”ผํ—˜์ž ๋‚ด ์ƒ๊ด€์ด ์กด์žฌํ•˜๋ฏ€๋กœ GEEยทQIF์™€ ๊ฐ™์€ ๋ฐ˜๋ชจ์ˆ˜์  ์ ‘๊ทผ์ด ์ผ๋ฐ˜์ ์ด๋‹ค. ํ•˜์ง€๋งŒ ์ƒŒ๋“œ์œ„์น˜ ๊ณต๋ถ„์‚ฐ์— ์˜์กดํ•˜๋Š” Wald ๊ฒ€์ •์€ ๊ตฐ์ง‘ ์ˆ˜๊ฐ€ ์ค‘๊ฐ„ ์ˆ˜์ค€์ผ ๋•Œ ๋ถˆ์•ˆ์ •ํ•˜๊ณ , ์ž‘์—… ์ƒ๊ด€๊ตฌ์กฐ๊ฐ€ ์ž˜๋ชป ์ง€์ •๋˜๋ฉด ํšจ์œจ์„ฑ์ด ํฌ๊ฒŒ ์ €ํ•˜๋œ๋‹ค. GPLSIM์€ ์„ ํ˜• ํšจ๊ณผ์™€ ๋น„์„ ํ˜• ์ธ๋ฑ์Šค ํšจ๊ณผ๋ฅผ ๋™์‹œ์— ํฌ์ฐฉํ•  ์ˆ˜ ์žˆ์–ด ํ˜„๋Œ€์˜ ๋ณต์žกํ•œ ์ƒ๋ฌผยท์‚ฌํšŒ๊ณผํ•™ ๋ฐ์ดํ„ฐ์— ์ ํ•ฉํ•˜์ง€๋งŒ, ๋น„์„ ํ˜• ๋งํฌ ํ•จ์ˆ˜๊ฐ€ โ€˜๊ท€์ฐฎ์€โ€™ ๋ฌดํ•œ ์ฐจ์› ๋งค๊ฐœ๋ณ€์ˆ˜์ด๊ธฐ ๋•Œ๋ฌธ์— ํŒŒ๋ผ๋ฉ”ํŠธ๋ฆญ ๋ถ€๋ถ„์— ๋Œ€ํ•œ ์ •ํ™•ํ•œ ์ถ”๋ก ์ด ์–ด๋ ค์› ๋‹ค. 2. ํ•ต์‹ฌ ๋ฐฉ๋ฒ•๋ก  ์Šคํ”Œ๋ผ์ธ ์‚ฌ์ด๋ธŒ : ๋ฏธ์ง€

Statistics Model
No Image

Bottleneck Transformer-Based Approach for Improved Automatic STOI Score Prediction

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ STOI ๋Š” ๊ฐ๊ด€์ ์ธ ์Œ์„ฑ ๊ฐ€๋…์„ฑ ์ง€ํ‘œ์ด์ง€๋งŒ, ๋ ˆํผ๋Ÿฐ์Šค ์‹ ํ˜ธ๊ฐ€ ํ•„์š”ํ•ด ๋น„์นจ์ž…ํ˜• ์‹œ๋‚˜๋ฆฌ์˜ค์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ์ตœ๊ทผ ๋น„์นจ์ž…ํ˜• ํ‰๊ฐ€๋ฅผ ์œ„ํ•ด Qualityโ€‘Net , STOIโ€‘Net , MOSAโ€‘Net , MTIโ€‘Net ๋“ฑ ๋‹ค์–‘ํ•œ ๋”ฅ๋Ÿฌ๋‹ ๊ธฐ๋ฐ˜ ๋ชจ๋ธ์ด ์ œ์•ˆ๋์ง€๋งŒ, ํŠน์ง• ์ถ”์ถœยท์‹œ๊ฐ„โ€‘์ปจํ…์ŠคํŠธ ๋ชจ๋ธ๋ง ์—์„œ ํ•œ๊ณ„๊ฐ€ ๋‚จ์•„ ์žˆ๋‹ค. ํŠนํžˆ, Selfโ€‘Supervised Learning(SSL) ๊ธฐ๋ฐ˜ ์ž„๋ฒ ๋”ฉ์€ ๊ฐ•๋ ฅํ•˜์ง€๋งŒ, ์—ฐ์‚ฐ๋Ÿ‰ยทํŒŒ๋ผ๋ฏธํ„ฐ ์ˆ˜ ๊ฐ€ ์ปค์ง€๋Š” ๊ฒฝํ–ฅ์ด ์žˆ๋‹ค. 2. ์ œ์•ˆ ๋ชจ๋ธ์˜ ํ•ต์‹ฌ ์•„์ด๋””์–ด | ๊ตฌ์„ฑ ์š”์†Œ | ์—ญํ•  | ์ฃผ์š”

Audio Processing Electrical Engineering and Systems Science
No Image

Budget-aware Test-time Scaling via Discriminative Verification

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๋Œ€ํ˜• ์–ธ์–ด ๋ชจ๋ธ(LLM)์€ ๋ณต์žกํ•œ ์ˆ˜ํ•™ยท๋…ผ๋ฆฌ ๋ฌธ์ œ์—์„œ ๋›ฐ์–ด๋‚œ ์„ฑ๋Šฅ์„ ๋ณด์ด์ง€๋งŒ, ๋‹จ์ผ ์ถ”๋ก  ๊ฒฐ๊ณผ์— ์˜์กดํ•˜๋ฉด ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•˜๊ธฐ ์‰ฝ๋‹ค. ํ…Œ์ŠคํŠธโ€‘ํƒ€์ž„ ์Šค์ผ€์ผ๋ง์€ ์—ฌ๋Ÿฌ ํ›„๋ณด ๋‹ต์•ˆ์„ ์ƒ์„ฑํ•˜๊ณ , ๊ทธ ์ค‘ ์ตœ์ ์„ ์„ ํƒํ•จ์œผ๋กœ์จ ์ •ํ™•๋„๋ฅผ ๋†’์ด๋Š” ๋ฐฉ๋ฒ•์ด๋‹ค. ๊ธฐ์กด์—๋Š” ์ƒ์„ฑํ˜• ๊ฒ€์ฆ๊ธฐ ๊ฐ€ ํ›„๋ณด ๋‹ต์•ˆ์„ ๋‹ค์‹œ ์ƒ์„ฑยท์žฌ๊ฒ€์ฆํ•˜๋Š” ๋ฐฉ์‹์œผ๋กœ ์‚ฌ์šฉ๋ผ, ํ›„๋ณด ์ˆ˜๊ฐ€ ๋Š˜์–ด๋‚ ์ˆ˜๋ก ์—ฐ์‚ฐ๋Ÿ‰์ด ๊ธฐํ•˜๊ธ‰์ˆ˜์ ์œผ๋กœ ์ฆ๊ฐ€ํ•œ๋‹ค. 2. ํŒ๋ณ„ํ˜• ๊ฒ€์ฆ๊ธฐ์˜ ํ•ต์‹ฌ ์•„์ด๋””์–ด ํŒ๋ณ„ํ˜• ๊ฒ€์ฆ๊ธฐ๋Š” โ€œ์ด ๋‹ต์•ˆ์ด ์ •๋‹ต์ธ์ง€ ์•„๋‹Œ์ง€โ€๋ฅผ ์ด์ง„(๋˜๋Š” ๋‹ค์ค‘) ๋ถ„๋ฅ˜ ํ˜•ํƒœ๋กœ ํŒ๋‹จํ•œ๋‹ค. ์ƒ์„ฑํ˜• ๊ฒ€์ฆ๊ธฐ

No Image

Can Generative Artificial Intelligence Survive Data Contamination? Theoretical Guarantees under Contaminated Recursive Training

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฐ์ดํ„ฐ ์˜ค์—ผ์˜ ์‹ค์ฆ์  ์ฆ๊ฑฐ : ์œ„ํ‚คํ”ผ๋””์•„, ๋‰ด์Šค ์‚ฌ์ดํŠธ ๋“ฑ์—์„œ AIโ€‘์ƒ์„ฑ ํ…์ŠคํŠธยท์ด๋ฏธ์ง€ยท์ฝ”๋“œ๊ฐ€ ๊ธ‰์ฆํ•˜๊ณ  ์žˆ๋‹ค๋Š” ์–ธ๋ก  ๋ณด๋„๊ฐ€ ๋Š˜์–ด๋‚˜๊ณ  ์žˆ๋‹ค. ์ด๋Š” ํ–ฅํ›„ ๋ชจ๋ธ ํ•™์Šต ์‹œ โ€œ์ง„์งœโ€์™€ โ€œ๊ฐ€์งœโ€๋ฅผ ๊ตฌ๋ถ„ํ•˜๊ธฐ ์–ด๋ ค์šด ์ƒํ™ฉ์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ๊ธฐ์กด ์ด๋ก ์˜ ํ•œ๊ณ„ : ๊ธฐ์กด ์—ฐ๊ตฌ(Shumailov 2024, Suresh 2024 ๋“ฑ)๋Š” ์ด์‚ฐยท๊ฐ€์šฐ์‹œ์•ˆ ๊ฐ€์ • ํ•˜์— ๋ชจ๋ธ ๋ถ•๊ดด ๋ฅผ ๋ณด์˜€์ง€๋งŒ, ํ˜„๋Œ€ ๋”ฅ ์ƒ์„ฑ ๋ชจ๋ธ(Transformer, Diffusion, GAN ๋“ฑ)์€ ์ด๋Ÿฌํ•œ ๊ฐ€์ •์„ ๋งŒ์กฑํ•˜์ง€ ์•Š๋Š”๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์—ฌ | ๊ตฌ๋ถ„ | ๊ธฐ์กด ์—ฐ๊ตฌ |

Data Computer Science Machine Learning
No Image

CarAT: Carbon Atom Tracing across Industrial Chemical Value Chains via Chemistry Language Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ทœ์ œ ์••๋ ฅ : TfS์˜ 2026๋…„ BCC ๋ณด๊ณ  ์˜๋ฌด๋Š” ๊ธฐ์กด์˜ ์ •์„ฑ์ ยท๋ถ€๋ถ„์  ํƒ„์†Œ ํšŒ๊ณ„ ๋ฐฉ์‹์„ ์™„์ „ ์ž๋™ํ™”ยท์ •๋Ÿ‰ํ™”๋œ ์‹œ์Šคํ…œ์œผ๋กœ ์ „ํ™˜ํ•ด์•ผ ํ•จ์„ ์˜๋ฏธํ•œ๋‹ค. ๊ธฐ์กด ๋ฐฉ๋ฒ•์˜ ํ•œ๊ณ„ : ํƒ„์†Œโ€‘14 ๋ถ„์„์€ ์ •ํ™•ํ•˜์ง€๋งŒ ๋น„์šฉยท์‹œ๊ฐ„ยท์‹œ๋ฃŒ ์š”๊ตฌ๋Ÿ‰์ด ํฌ๊ณ , ์—ฐ์†์ ์ธ ๊ณต์ • ๋ชจ๋‹ˆํ„ฐ๋ง์— ๋ถ€์ ํ•ฉํ•˜๋‹ค. 2. CarAT์˜ ํ•ต์‹ฌ ๊ธฐ์ˆ  ๊ตฌ์„ฑ | ๋‹จ๊ณ„ | ํ•ต์‹ฌ ๊ธฐ์ˆ  | ์—ญํ•  | ๊ธฐ๋Œ€ ํšจ๊ณผ | | | | | | | ๋ฐ์ดํ„ฐ ์ „์ฒ˜๋ฆฌ | ERP ๋ฐ์ดํ„ฐ ํ†ตํ•ฉยท์ •๊ทœํ™” | ์›๋ฃŒยท์ œํ’ˆยท๊ณต์ • ํ๋ฆ„์„ ํ‘œ์ค€ํ™”๋œ ๊ทธ๋ž˜ํ”„ ํ˜•ํƒœ๋กœ ๋ณ€ํ™˜ | ๋‹ค์–‘ํ•œ ์‹œ์Šคํ…œ ๊ฐ„ ๋ฐ์ดํ„ฐ ์ด์งˆ์„ฑ

Model
No Image

CellMaster: Collaborative Cell Type Annotation in Single-Cell Analysis

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‹จ์ผ์„ธํฌ ๋ฐ์ดํ„ฐ ๊ทœ๋ชจ : ํ˜„์žฌ ๋‹จ์ผ์„ธํฌ ์•„ํ‹€๋ผ์Šค๋Š” ์ˆ˜์ฒœ๋งŒ ๊ฐœ ์…€ยท์ˆ˜๋ฐฑ ์กฐ์ง์„ ํฌ๊ด„ํ•˜๊ณ  ์žˆ์–ด, ์ „ํ†ต์ ์ธ ์ „๋ฌธ๊ฐ€ ์ฃผ์„ ๋ฐฉ์‹์€ ์‹œ๊ฐ„ยท๋…ธ๋ ฅ ๋ฉด์—์„œ ๋น„ํ˜„์‹ค์ ์ด๋‹ค. ๋งˆ์ปค ์˜์กด์„ฑ ํ•œ๊ณ„ : ์กฐ์งยท์ƒํƒœ๋งˆ๋‹ค ๋งˆ์ปค ์œ ์ „์ž์˜ ๋ฐœํ˜„ ํŒจํ„ด์ด ํฌ๊ฒŒ ๋‹ฌ๋ผ์ง€๋ฉฐ, ์‹ ๊ทœ ํ˜น์€ ์ „์ด ์ƒํƒœ๋Š” ๊ธฐ์กด ๋งˆ์ปค ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค์— ์กด์žฌํ•˜์ง€ ์•Š๋Š”๋‹ค. LLM ํ™œ์šฉ ๊ฐ€๋Šฅ์„ฑ : GPTโ€‘4o์™€ ๊ฐ™์€ ์ตœ์‹  LLM์€ ๋ฐฉ๋Œ€ํ•œ ์ƒ๋ฌผํ•™ ํ…์ŠคํŠธ๋ฅผ ๋‚ด์žฌํ™”ํ•˜๊ณ  ์žˆ์–ด, โ€œ์ง€์‹ ๊ธฐ๋ฐ˜โ€ ์ฃผ์„์— ์ ํ•ฉํ•˜๋‹ค๋Š” ์ ์„ ๋ณธ ์—ฐ๊ตฌ๊ฐ€ ์ฐฉ์•ˆํ•˜์˜€๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ๋ฐ ๋ฐฉ๋ฒ•๋ก  | ๋‹จ๊ณ„ | ์ฃผ์š” ์ž‘์—…

Quantitative Biology Analysis
No Image

Central Dogma Transformer II: An AI Microscope for Understanding Cellular Regulatory Mechanisms

| ๋ถ„์„ ํ•ญ๋ชฉ | ์ฃผ์š” ๋‚ด์šฉ | ํ‰๊ฐ€ยท์‹œ์‚ฌ์  | | | | | | 1. ๋ชจ๋ธ ์„ค๊ณ„์™€ ์ƒ๋ฌผํ•™์  ์ •ํ•ฉ์„ฑ | ์ค‘์•™ ์›๋ฆฌ ํ๋ฆ„์„ ๊ทธ๋Œ€๋กœ ์•„ํ‚คํ…์ฒ˜์— ๋ฐ˜์˜ (DNAโ€‘self, RNAโ€‘self, DNAโ†’RNA cross). <br> ๊ฐ ์–ดํ…์…˜ ํ—ค๋“œ๊ฐ€ 8๊ฐœ, ์ฐจ์› ์œ ์ง€ โ†’ ์–ดํ…์…˜ ๋งต์„ ์ง์ ‘ ํ•ด์„ ๊ฐ€๋Šฅํ•˜๊ฒŒ ํ•จ. | ๊ฐ•์  : ๊ตฌ์กฐ์  inductive bias๊ฐ€ ํ•™์Šต ํšจ์œจ์„ ๋†’์ด๊ณ , โ€œ์–ดํ…์…˜ ๊ทœ์ œ ๊ด€๊ณ„โ€๋ผ๋Š” ์ง๊ด€์„ ์ œ๊ณตํ•œ๋‹ค. <br> ํ•œ๊ณ„ : ๋‹จ๋ฐฑ์งˆ ๋ ˆ์ด์–ด๊ฐ€ ์—†์œผ๋ฏ€๋กœ ํ›„์† ๋ฒˆ์—ญยทํ›„๋ฒˆ์—ญ ์กฐ์ ˆ์€ ๊ฐ„์ ‘์ ์œผ๋กœ๋งŒ ๋ฐ˜์˜๋œ๋‹ค. | | 2. ์ž…๋ ฅ ๋ฐ์ดํ„ฐ์™€ ์ „์ฒ˜๋ฆฌ |

Computer Science Machine Learning
No Image

Central limit theorem for random walk in degenerate divergence-free random environment: $mathcal H_{-1}$ reloaded with relaxed ellipticity

| ๊ตฌ๋ถ„ | ๋‚ด์šฉ | | | | | ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ | Kozmaโ€‘Tรณth(2017)์—์„œ๋Š” ์ด์ค‘ ํ™•๋ฅ ์„ฑ(biโ€‘stochastic) ๊ณผ ๊ฐ•ํ•œ ํƒ€์›์„ฑ ์„ ์ „์ œ๋กœ, ๋น„๊ฐ€์—ญ์  ํ๋ฆ„์ด ์กด์žฌํ•˜๋Š” ๊ฒฝ์šฐ์—๋„ Kipnisโ€‘Varadhan ๋ฐฉ์‹์˜ ๋งˆํŒ…๊ฒŒ์ผ ๊ทผ์‚ฌ์™€ Nash ๋ถ€๋“ฑ์‹ ๊ธฐ๋ฐ˜ ์—ดํ•ต์‹ฌ ์ถ”์ •์œผ๋กœ CLT๋ฅผ ์ฆ๋ช…ํ•˜์˜€๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ๊ฐ•ํ•œ ํƒ€์›์„ฑ์€ ์‹ค์ œ ๋ฌผ๋ฆฌยท์ƒ๋ฌผ ๋ชจ๋ธ(์˜ˆ: ์ „๋„์„ฑ ๋ถˆ๊ท ์ผ ๋งค์งˆ, ๋น„๋Œ€์นญ ์ „์ด์œจ์„ ๊ฐ–๋Š” ์ž…์ž ์‹œ์Šคํ…œ)์—์„œ ์ œํ•œ์ ์ด์—ˆ๋‹ค. | | ํ•ต์‹ฌ ๊ธฐ์—ฌ | 1. ํƒ€์›์„ฑ ์™„ํ™” : ๋Œ€์นญ ์ ํ”„์œจ (s k) ์˜ ์—ญ์ˆ˜ ((s k)^{ 1}) ๊ฐ€ (

Mathematics
No Image

Certified Per-Instance Unlearning Using Individual Sensitivity Bounds

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๊ทœ์ œยท๊ณ„์•ฝ์  ์š”๊ตฌ (์˜ˆ: โ€œ์žŠํ˜€์งˆ ๊ถŒ๋ฆฌโ€)์— ๋”ฐ๋ผ ํ•™์Šต ํ›„ ๊ฐœ๋ณ„ ์ƒ˜ํ”Œ์„ ์‚ญ์ œํ•ด์•ผ ํ•˜๋Š” ์ƒํ™ฉ์ด ๋Š˜์–ด๋‚˜๊ณ  ์žˆ๋‹ค. ์ „ํ†ต์  ์ ‘๊ทผ ์€ ์ „์ฒด ๋ฐ์ดํ„ฐ๋ฅผ ๋‹ค์‹œ ํ•™์Šต(retraining)ํ•˜๋Š” ๊ฒƒ์ด์ง€๋งŒ, ๋น„์šฉ์ด prohibitive ํ•˜๋‹ค. ๋จธ์‹  ์–ธ๋Ÿฌ๋‹ ์€ ์žฌํ•™์Šต์„ ๊ทผ์‚ฌํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ์ œ๊ณตํ•œ๋‹ค. ์—ฌ๊ธฐ์„œ ์ธ์ฆ ์–ธ๋Ÿฌ๋‹ ์€ ์ฐจ๋“ฑ ํ”„๋ผ์ด๋ฒ„์‹œ์™€ ์œ ์‚ฌํ•œ ํ˜•ํƒœ๋กœ ์ •๋Ÿ‰์  ๋ณด์ฆ ์„ ์ œ๊ณตํ•œ๋‹ค. ๊ธฐ์กด ์ธ์ฆ ์–ธ๋Ÿฌ๋‹์€ ์ „์ฒด ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ์ตœ์•…โ€‘์‚ฌ๋ก€ ๋ฏผ๊ฐ๋„ ๋ฅผ ์‚ฌ์šฉํ•ด ๋…ธ์ด์ฆˆ๋ฅผ ๊ณผ๋„ํ•˜๊ฒŒ ์‚ฝ์ž…, ์‹ค์šฉ์„ฑ์— ์ œ์•ฝ์ด ์žˆ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด 1. Perโ€‘I

Computer Science Learning Machine Learning
No Image

Certifying Hamilton-Jacobi Reachability Learned via Reinforcement Learning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ HJ ๋„๋‹ฌ๊ฐ€๋Šฅ์„ฑ ์€ ์•ˆ์ „ ๊ฒ€์ฆ์—์„œ โ€œ์–ด๋–ค ์ดˆ๊ธฐ ์ƒํƒœ๊ฐ€ ์ฃผ์–ด์ง„ ์‹œ๊ฐ„ ์•ˆ์— ์œ„ํ—˜ ์˜์—ญ์— ๋„๋‹ฌํ•  ์ˆ˜ ์žˆ๋Š”๊ฐ€โ€๋ฅผ ํŒ๋‹จํ•˜๋Š” ํ•ต์‹ฌ ๋„๊ตฌ์ด๋ฉฐ, ๊ฐ’ ํ•จ์ˆ˜์˜ ๋ถ€ํ˜ธ๊ฐ€ ๋ฐ”๋กœ ์•ˆ์ „/์œ„ํ—˜์„ ๊ตฌ๋ถ„ํ•œ๋‹ค. ์ „ํ†ต์ ์ธ ๊ฒฉ์ž ๊ธฐ๋ฐ˜ ํ•ด๋ฒ•์€ ์ฐจ์› ์ €์ฃผ(curse of dimensionality) ๋•Œ๋ฌธ์— ์‹ค์‹œ๊ฐ„ยท๊ณ ์ฐจ์› ์‹œ์Šคํ…œ์— ์ ์šฉํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ์ตœ๊ทผ ํ•™์Šต ๊ธฐ๋ฐ˜ ์ ‘๊ทผ๋ฒ•์ด ๋‘ ๊ฐˆ๋ž˜๋กœ ์ „๊ฐœ๋œ๋‹ค. 1. Selfโ€‘supervised ๋ฐฉ์‹ (DeepReach, CARe ๋“ฑ) โ€“ ๋™์—ญํ•™ ๋ชจ๋ธ์ด ํ•„์š”ํ•˜๊ฑฐ๋‚˜, SMTยทCEGIS์™€ ๊ฒฐํ•ฉํ•ด ๋ณด์ฆ์„ ์ œ๊ณตํ•˜์ง€๋งŒ, ๋ชจ๋ธ

Learning Electrical Engineering and Systems Science
No Image

Certifying Robustness via Topological Representations

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์œ„์ƒ ๋ฐ์ดํ„ฐ ๋ถ„์„(TDA) ์€ ๋ฐ์ดํ„ฐ์˜ ์ „์—ญ์ ์ธ ํ˜•ํƒœ์™€ ์—ฐ๊ฒฐ์„ฑ์„ ํฌ์ฐฉํ•˜๋Š” ๊ฐ•๋ ฅํ•œ ๋„๊ตฌ์ด๋ฉฐ, ํผ์‹œ์Šคํ„ด์Šค ๋‹ค์ด์–ด๊ทธ๋žจ์€ ์ด๋Ÿฌํ•œ ์œ„์ƒ ์ •๋ณด๋ฅผ ์••์ถ•ํ•œ ๋Œ€ํ‘œ์ ์ธ ์š”์•ฝ ํ˜•ํƒœ์ด๋‹ค. ๊ธฐ์กด ์—ฐ๊ตฌ์—์„œ๋Š” ํผ์‹œ์Šคํ„ด์Šค ๋‹ค์ด์–ด๊ทธ๋žจ์„ ์ง์ ‘ ๋ถ„๋ฅ˜๊ธฐ์— ์ž…๋ ฅํ•˜๊ฑฐ๋‚˜, ์ปค๋„/๊ฑฐ๋ฆฌ ๊ธฐ๋ฐ˜ ๋ฐฉ๋ฒ•์œผ๋กœ ํ™œ์šฉํ–ˆ์ง€๋งŒ, ์‹ ๊ฒฝ๋ง๊ณผ์˜ ํ†ตํ•ฉ ์€ ์•„์ง ์ดˆ๊ธฐ ๋‹จ๊ณ„์— ์žˆ๋‹ค. ๋˜ํ•œ ์ ๋Œ€์  ๊ณต๊ฒฉ ์— ๋Œ€ํ•œ ๊ฒฌ๊ณ ์„ฑ ์ธ์ฆ์€ ์ฃผ๋กœ ์ž…๋ ฅ ๊ณต๊ฐ„์—์„œ์˜ ๋ฆฌํ”„์‹œ์ธ  ์—ฐ์†์„ฑ์„ ์ด์šฉํ•ด ์ด๋ฃจ์–ด์กŒ์œผ๋ฉฐ, ์œ„์ƒ์  ํŠน์„ฑ์„ ํ™œ์šฉํ•œ ์ ‘๊ทผ์€ ๋“œ๋ฌผ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๋‚ด์šฉ | ์˜์˜ | | | |

No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ชฉํ‘œ ์šฐ์ฃผ์„  ๋ฎค์˜จ ์€ ๋Œ€๊ธฐ ์ƒ์ธต์—์„œ ์ƒ์„ฑ๋œ 2์ฐจ ์ž…์ž๋กœ, ๋†’์€ ๊ด€ํ†ต๋ ฅ๊ณผ ์ƒ๋Œ€์ ์œผ๋กœ ๊ธด ์ˆ˜๋ช…(์‹œ๊ฐ„ ํŒฝ์ฐฝ์— ์˜ํ•ด ์—ฐ์žฅ) ๋•Œ๋ฌธ์— ์ง€ํ‘œ๋ฉด์—์„œ ๊ฐ€์žฅ ๋งŽ์ด ๊ด€์ธก๋˜๋Š” ์ž…์ž์ด๋‹ค. ์ „ํ†ต์ ์ธ ๊ด‘์ „๊ด€(PMT) ๋Œ€๋น„ MPPC(SiPM) ๋Š” ์ €์ „์•• ๊ตฌ๋™, ์†Œํ˜•ยท๊ฒฝ๋Ÿ‰, ๋†’์€ ์–‘์ž ํšจ์œจ(QE) ๋“ฑ ์‹คํ—˜์‹ค ๋ฐ ํ˜„์žฅ ์ ์šฉ์— ์žฅ์ ์ด ์žˆ๋‹ค. ๋ณธ ์—ฐ๊ตฌ๋Š” (โ‘  MPPCโ€‘์„ฌ๊ด‘ ๊ฒฐํ•ฉ ํŠน์„ฑํ™”, โ‘ก 3์ค‘ ์ฝ”์ธ์‹œ๋˜์Šค ๋ง์›๊ฒฝ ๊ตฌํ˜„, โ‘ข ๋ฎค์˜จ ๊ฐ๋„ ๋ถ„ํฌ ์ •๋ฐ€ ์ธก์ •)์ด๋ผ๋Š” 3๋‹จ๊ณ„ ๋ชฉํ‘œ๋ฅผ ์„ค์ •ํ•˜์˜€๋‹ค. 2. ์‹คํ—˜ ์„ค๊ณ„ ๋ฐ ๋ฐฉ๋ฒ•๋ก  | ํ•ญ๋ชฉ | ๋‚ด์šฉ | ์˜์˜ | | |

Physics
No Image

Chem-SIM: Super-resolution Chemical Imaging via Photothermal Modulation of Structured-Illumination Fluorescence

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ตฌ์กฐ์กฐ๋ช… ํ˜„๋ฏธ๊ฒฝ(SIM) ์€ ๊ด‘ํ•™ ํšŒ์ ˆ ํ•œ๊ณ„๋ฅผ ์•ฝ 2๋ฐฐ ์ดˆ๊ณผํ•˜๋Š” ํ•ด์ƒ๋„๋ฅผ ์ œ๊ณตํ•˜์ง€๋งŒ, ํ˜•๊ด‘ ์‹ ํ˜ธ๋งŒ์„ ๊ธฐ๋ฐ˜์œผ๋กœ ํ•˜๋ฏ€๋กœ ๋ฌผ์งˆ์˜ ํ™”ํ•™์  ํŠน์„ฑ์„ ์ง์ ‘์ ์œผ๋กœ ํŒŒ์•…ํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ๊ธฐ์กด ๊ด‘์—ด ํ˜„๋ฏธ๊ฒฝ(MIP/Fโ€‘MIP) ์€ ์ ์™ธ์„  ํก์ˆ˜์— ๋”ฐ๋ฅธ ์˜จ๋„ ๋ณ€ํ™”๋ฅผ ๊ฒ€์ถœํ•ด ํ™”ํ•™ ์ •๋ณด๋ฅผ ์–ป์ง€๋งŒ, ๋ฌผ ํก์ˆ˜์™€ ๋‚ฎ์€ ํ˜•๊ด‘ ๊ด‘์ž๋Ÿ‰์œผ๋กœ ์ธํ•ด ์ˆ˜์šฉ์•ก ํ™˜๊ฒฝ์—์„œ์˜ ์ดˆ๊ณ ํ•ด์ƒ๋„ ๊ตฌํ˜„์ด ์ œํ•œ์ ์ด์—ˆ๋‹ค. 2. ํ•ต์‹ฌ ๊ธฐ์ˆ  ํ˜์‹  | ๊ธฐ์ˆ  ์š”์†Œ | ๊ธฐ์กด ๋ฐฉ์‹ | Chemโ€‘SIM ํ˜์‹ ์  | | | | | | ๊ด‘์—ด ์ž๊ทน | ๋‹จ์ผ ํŽ„์Šคยท์ ยท์Šค์บ” ๋ฐฉ์‹ | 50 kHz

Physics
No Image

ChemRecon: a Consolidated Meta-Database Platform for Biochemical Data Integration

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค ๋‹ค๋ณ€ํ™” : ํ˜„์žฌ ์ƒํ™”ํ•™ ๋ถ„์•ผ๋Š” ์ˆ˜์‹ญ ๊ฐœ์˜ ์ „๋ฌธ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค๊ฐ€ ์กด์žฌํ•˜์ง€๋งŒ, ํฌ๋งทยท์ ‘๊ทผ ๋ฐฉ์‹ยท์˜จํ†จ๋กœ์ง€๊ฐ€ ์„œ๋กœ ๋‹ฌ๋ผ ํ†ตํ•ฉ์ด ์–ด๋ ค์›€. ๊ธฐ์กด ํ•œ๊ณ„ : ๊ฐ DB๋งˆ๋‹ค ๋ณ„๋„ ํŒŒ์„œยทAPI ๊ตฌํ˜„์ด ํ•„์š”ํ•˜๊ณ , ๊ต์ฐจ ์ฐธ์กฐ๋Š” ์กด์žฌํ•˜์ง€๋งŒ ์‹ค์ œ ํ™œ์šฉ์„ ์œ„ํ•ด์„œ๋Š” ๋ณต์žกํ•œ ๋งคํ•‘ ์ž‘์—…์ด ์„ ํ–‰๋ผ์•ผ ํ•จ. ๋˜ํ•œ, ๋™์ผ ํ™”ํ•ฉ๋ฌผ์— ๋Œ€ํ•œ ๋ถˆ์ผ์น˜(tautomer, ๊ตฌ์กฐ, ID) ๊ฐ€ ๋นˆ๋ฒˆํžˆ ๋ฐœ์ƒํ•œ๋‹ค. 2. ChemRecon์˜ ํ•ต์‹ฌ ์„ค๊ณ„ | ์š”์†Œ | ์„ค๋ช… | ์žฅ์  | | | | | | ๋ฉ”ํƒ€โ€‘๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค | ๋ชจ๋“  ์›๋ณธ DB๋ฅผ ์™„์ „ ๋‹ค์šด๋กœ๋“œ ํ›„

Data Quantitative Biology
No Image

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

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ BNN์˜ ์ž ์žฌ๋ ฅ : ์ „๋ ฅยท์ƒ˜ํ”Œ ํšจ์œจ์„ฑ, ํ’๋ถ€ํ•œ ๋ณ‘๋ ฌ ๋™์—ญํ•™, ์‹œ๋ƒ…์Šค ๊ฐ€์†Œ์„ฑ ๋“ฑ์€ ๊ธฐ์กด ์‹ค๋ฆฌ์ฝ˜ ๊ธฐ๋ฐ˜ ์ปดํ“จํŒ…์ด ์ œ๊ณตํ•˜์ง€ ๋ชปํ•˜๋Š” ํŠน์„ฑ์ด๋‹ค. ๋””์ง€ํ„ธโ€‘์ƒ๋ฌผํ•™ ์ธํ„ฐํŽ˜์ด์Šค์˜ ๋ณ‘๋ชฉ : ์ „๊ธฐ์  ์ž๊ทนยท๊ธฐ๋ก์„ ์‹ค์‹œ๊ฐ„์œผ๋กœ ์ˆ˜ํ–‰ํ•˜๋ ค๋ฉด ยตsโ€‘๋‹จ์œ„ ํƒ€์ด๋ฐ ์ •ํ™•๋„ ์™€ ๋‹ค์ฑ„๋„ ๋™๊ธฐํ™” ๊ฐ€ ํ•„์ˆ˜์ธ๋ฐ, ๊ธฐ์กด ์†”๋ฃจ์…˜์€ ํ•˜๋“œ์›จ์–ด ์ข…์†์„ฑ(๋ณต์žก๋„) ํ˜น์€ ๋†’์€ ์ง€์—ฐ(>60 ms) ์‚ฌ์ด์—์„œ ์„ ํƒ์„ ๊ฐ•์š”ํ•œ๋‹ค. ์†Œํ”„ํŠธ์›จ์–ด ๊ณ„์ธต์˜ ๊ณต๋ฐฑ : FPGAโ€‘์ง์ ‘ ์ œ์–ด๋Š” ๋น ๋ฅด์ง€๋งŒ ๊ฐœ๋ฐœ ์ฃผ๊ธฐ๊ฐ€ ๊ธธ๊ณ  ํˆฌ๋ช…์„ฑ์ด ๋‚ฎ์œผ๋ฉฐ, ์ˆœ์ˆ˜ ํŒŒ์ด์ฌ ๊ตฌํ˜„์€ ์‚ฌ์šฉ์ด ์‰ฌ์šฐ๋‚˜ ์‹œ๊ฐ„ ์ •ํ™•๋„๊ฐ€ ๋–จ์–ด

Network Quantitative Biology
No Image

Collaboration drives phase transitions towards cooperation in prisoner's dilemma

| ํ•ญ๋ชฉ | ๋‚ด์šฉ ๋ฐ ์˜์˜ | | | | | ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ | ๊ธฐ์กด ํ˜‘๋ ฅ ์ด๋ก (์นœ์กฑ ์„ ํƒ, ์ง์ ‘ยท๊ฐ„์ ‘ ์ƒํ˜ธ ํ˜ธํ˜œ, ๋„คํŠธ์›Œํฌ ์ƒํ˜ธ ํ˜ธํ˜œ, ์ง‘๋‹จ ์„ ํƒ) ์™ธ์— ์ธ๊ฐ„ ๊ณ ์œ ์˜ ๊ณต์œ  ์˜๋„(shared intentionality) ๋ฅผ ๋„์ž…ํ•ด ํ˜‘๋™(coalition) ๋ฉ”์ปค๋‹ˆ์ฆ˜์„ ์ •๋Ÿ‰ํ™”ํ•œ๋‹ค. ์ด๋Š” โ€˜ํ˜‘์—…์ด ์ง์ ‘ ์ƒํ˜ธ ํ˜ธํ˜œ๋ฅผ ๋Œ€์ฒดํ•œ๋‹คโ€™๋Š” ์ƒˆ๋กœ์šด ์‹œ๊ฐ์„ ์ œ๊ณตํ•œ๋‹ค. | | ๋ชจ๋ธ ์„ค๊ณ„ | ๊ณ ๋ฆฌ ๋„คํŠธ์›Œํฌ (์ •๊ทœ 2โ€‘์ฐจ์› ๊ทธ๋ž˜ํ”„) โ†’ ๋ถ„์„์ด ์šฉ์ดํ•˜๋ฉด์„œ๋„ ๊ทผ์ ‘ ์ด์›ƒ ๊ฐ„ ์ƒํ˜ธ์ž‘์šฉ์„ ๋ช…ํ™•ํžˆ ํ•จ.<br> ๋‘ ๋‹จ๊ณ„ ์—…๋ฐ์ดํŠธ : (1) ํ˜‘์—… ํ™•๋ฅ  p ๋กœ ์ด์›ƒ๊ณผ ๊ณต๋™ ์ตœ์ 

Physics
No Image

Collaborative Safe Bayesian Optimization

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ 5Gยท6G ๋„คํŠธ์›Œํฌ์˜ ๋ณต์žก์„ฑ : ์•ˆํ…Œ๋‚˜ ๊ธฐ์šธ๊ธฐ, ๋น”ํญ ๋“ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๊ฐ€ ์‚ฌ์šฉ์ž ์ปค๋ฒ„๋ฆฌ์ง€์™€ ์Šค๋ฃจํ’‹์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋ฉฐ, ํ™˜๊ฒฝ ๋ณ€ํ™”์— ๋”ฐ๋ผ ์‹ค์‹œ๊ฐ„ ์žฌ์กฐ์ •์ด ํ•„์ˆ˜์ ์ด๋‹ค. ์•ˆ์ „ ์ œ์•ฝ : ์ตœ์†Œ ์ปค๋ฒ„๋ฆฌ์ง€, ์ตœ์†Œ ์‹ ํ˜ธ ํ’ˆ์งˆ ๋“ฑ ์„œ๋น„์Šค ํ’ˆ์งˆ์„ ๋ณด์žฅํ•ด์•ผ ํ•˜๋ฏ€๋กœ, ํŒŒ๋ผ๋ฏธํ„ฐ ํƒ์ƒ‰ ๊ณผ์ •์—์„œ โ€œ๋ถˆ์•ˆ์ „โ€ํ•œ ์„ค์ •์„ ์ ˆ๋Œ€๋กœ ์ ์šฉํ•  ์ˆ˜ ์—†๋‹ค. ๊ธฐ์กด ๋ฐฉ๋ฒ•์˜ ํ•œ๊ณ„ ๊ทœ์น™ ๊ธฐ๋ฐ˜ : ๋ณด์ˆ˜์ ์ด๋ฉฐ ์ƒˆ๋กœ์šด ์ƒํ™ฉ์— ์ ์‘์ด ์–ด๋ ค์›€. ๊ฐ•ํ™”ํ•™์Šต(RL) : ๋Œ€๊ทœ๋ชจ ์˜คํ”„๋ผ์ธ ํ•™์Šตยท์‹œ๋ฎฌ๋ ˆ์ดํ„ฐ ํ•„์š”, ์•ˆ์ „ ๋ณด์žฅ์ด ๋ฏธํก. ์ „ํ†ต์  Safe BO : ์ดˆ๊ธฐ ์•ˆ์ „ ์ง‘ํ•ฉ์ด

Electrical Engineering and Systems Science
No Image

Color-based Emotion Representation for Speech Emotion Recognition

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๊ธฐ์กด SER์˜ ํ•œ๊ณ„ : ๋ฒ”์ฃผํ˜• ์ ‘๊ทผ์€ ๊ฐ์ • ํ˜ผํ•ฉ์ด๋‚˜ ๋ฏธ๋ฌ˜ํ•œ ์ฐจ์ด๋ฅผ ํฌ์ฐฉํ•˜์ง€ ๋ชปํ•˜๊ณ , ์ฐจ์›ํ˜• ์ ‘๊ทผ์€ ํ•ด์„์ด ์–ด๋ ค์›Œ ์‹ค๋ฌด ์ ์šฉ์— ์ œ์•ฝ์ด ์žˆ๋‹ค. ์ƒ‰ ์†์„ฑ์˜ ์žฅ์  : ์ƒ‰์€ ์ˆ˜์น˜ํ™”๋œ hueโ€‘saturationโ€‘value(HSV) ํ˜•ํƒœ๋กœ ์—ฐ์†์ ์ธ ๊ฐ’์„ ๊ฐ€์ง€๋ฉด์„œ๋„ ์‹œ๊ฐ์ ์œผ๋กœ ์ง๊ด€์ ์ด๋‹ค. ๋”ฐ๋ผ์„œ ๊ฐ์ • ์ •๋ณด๋ฅผ โ€œ๋ณด๋Š”โ€ ๋ฐฉ์‹์œผ๋กœ ์ „๋‹ฌํ•  ์ˆ˜ ์žˆ๋‹ค. 2. ๋ฐ์ดํ„ฐ ๊ตฌ์ถ• ๋ฐ์ดํ„ฐ์…‹ : ์ผ๋ณธ์–ด ์—ฐ๊ธฐ ์Œ์„ฑ ์ฝ”ํผ์Šค JVNV (1,615 ๋ฐœํ™”, 6๊ฐ€์ง€ ๊ฐ์ •) ์‚ฌ์šฉ. ํฌ๋ผ์šฐ๋“œ์†Œ์‹ฑ ๋ผ๋ฒจ๋ง : Lancers ํ”Œ๋žซํผ์„ ํ†ตํ•ด 10๋ช…์—๊ฒŒ ๊ฐ ๋ฐœํ™”๋งˆ๋‹ค

Audio Processing Electrical Engineering and Systems Science
No Image

Combined dynamic-kinematic validation of droplet-wall impact modeling

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋‹ค์–‘ํ•œ ์‚ฐ์—… ์ ์šฉ (๊ณ ์˜จ ๋ƒ‰๊ฐ, ์Šคํ”„๋ ˆ์ด ์ฝ”ํŒ…, 3D ๋ฐ”์ด์˜คํ”„๋ฆฐํŒ… ๋“ฑ)์—์„œ ๋“œ๋กญโ€‘์›” ์ถฉ๋Œ์€ ํ•ต์‹ฌ ํ˜„์ƒ์ด์ง€๋งŒ, ๊ธฐ์กด CFD ๊ฒ€์ฆ์€ ์ฃผ๋กœ ์ „๊ฐœ ์ง๊ฒฝ ์—๋งŒ ์˜์กดํ•ด ์™”๋‹ค. ๋‚ด๋ถ€ ์œ ๋™(๋ฐฉ์‚ฌํ˜• ์†๋„, ์ „๊ฐœ ์†๋„ ๋“ฑ)์€ ์—ด์ „๋‹ฌ, ๋ฌผ์งˆ ์ „๋‹ฌ, ์ ‘์ด‰์„  ๊ฑฐ๋™ ์— ์ง์ ‘์ ์ธ ์˜ํ–ฅ์„ ๋ฏธ์น˜๋ฏ€๋กœ, ์ด๋ฅผ ๋ฌด์‹œํ•˜๋ฉด ์„ค๊ณ„ ์ตœ์ ํ™”์— ํฐ ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•œ๋‹ค. 2. ๋ฐฉ๋ฒ•๋ก  | ๋‹จ๊ณ„ | ๋‚ด์šฉ | ํ•ต์‹ฌ ํฌ์ธํŠธ | | | | | | ์‹คํ—˜ | ๋ฌผโ€‘๊ธ€๋ฆฌ์„ธ๋ฆฐ(60 wt % ๊ธ€๋ฆฌ์„ธ๋ฆฐ)ยทTiOโ‚‚ ์ž…์ž ํ˜ผํ•ฉ์•ก, ์‚ฌํŒŒ์ด์–ด ์œ ๋ฆฌ ํ‘œ๋ฉด, We 20, 80, 250 |

Physics Model
No Image

Comments on `Comment on Aurรฉlien Drezet's defense of relational quantum mechanics' by Jay Lawrence, Marcin Markiewicz and Marek ลนukowski

1. ๋…ผ๋ฌธ์˜ ๋ฐฐ๊ฒฝ ๋ฐ ๋ชฉ์  ๊ด€๊ณ„์–‘์ž์—ญํ•™(RQM) : ๋ฌผ๋ฆฌ๋Ÿ‰์ด ๊ด€์ฐฐ์ž(๊ด€๊ณ„)๋งˆ๋‹ค ๋‹ค๋ฅด๊ฒŒ ์ •์˜๋œ๋‹ค๋Š” ์ž…์žฅ์„ ์ทจํ•˜๋Š” ํ•ด์„์œผ๋กœ, ์ „ํ†ต์  ์‹ค์žฌ๋ก ์  ํ•ด์„๊ณผ๋Š” ์ฐจ์ด๋ฅผ ๋ณด์ธ๋‹ค. ๋…ผ์Ÿ์˜ ์ „๊ฐœ : Drezet์€ ์ด์ „ ๋…ผ๋ฌธ์—์„œ RQM์ด ๊ฐ–๋Š” ์ผ๊ด€์„ฑ๊ณผ ์‹คํ—˜์  ํƒ€๋‹น์„ฑ์„ ์˜นํ˜ธํ–ˆ์œผ๋ฉฐ, ๋กœ๋ Œ์Šคยท๋งˆ๋ฅดํ‚ค์—๋น„์น˜ยท์ฃผ์ฝ”ํ”„์Šคํ‚ค๋Š” ๊ทธ ๋ฐฉ์–ด์— ๋Œ€ํ•ด โ€œ์ธก์ • ๋ฌธ์ œ์™€ ๋น„๊ตญ์†Œ์„ฑโ€์„ ์ง€์ ํ•˜๋ฉฐ ๋น„ํŒํ•˜์˜€๋‹ค. ๋ณธ ๋…ผ๋ฌธ์˜ ๋ชฉํ‘œ : ๋น„ํŒ์— ๋Œ€ํ•œ โ€œ์งง์€โ€ ๋ฐ˜๋ฐ•์„ ์ œ๊ณตํ•˜๊ณ , ๊ธฐ์กด ๋ฐ˜๋ฐ•์—์„œ ๋…์ž๋“ค์ด ํ˜ผ๋™ํ•  ์ˆ˜ ์žˆ๋˜ ๋ถ€๋ถ„์„ ๋ช…ํ™•ํžˆ ํ•˜๋Š” ๊ฒƒ์ด๋‹ค. 2. ์ฃผ์š” ์ฃผ์žฅ ๋ฐ ๋ฐ˜๋ฐ• ๊ตฌ์กฐ | ๊ตฌ๋ถ„ | ๋กœ๋ Œ์Šคยท๋งˆ๋ฅด

No Image

Competing Risk Analysis in Cardiovascular Outcome Trials: A Simulation Comparison of Cox and Fine-Gray Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๊ฒฝ์Ÿ์œ„ํ—˜์˜ ์‹ค๋ฌด์  ์ค‘์š”์„ฑ : ๋น„์‹ฌํ˜ˆ๊ด€ ์‚ฌ๋ง์ด MACE ๊ด€์ฐฐ์„ ์ฐจ๋‹จํ•˜๋ฉด, โ€œ๊ด€์ฐฐ๋˜์ง€ ์•Š์€โ€ ์‚ฌ๊ฑด์ด ์‹ค์ œ ์น˜๋ฃŒ ํšจ๊ณผ๋ฅผ ์™œ๊ณกํ•  ์œ„ํ—˜์ด ์žˆ๋‹ค. ํ˜„ํ–‰ ๊ฐ€์ด๋“œ๋ผ์ธ๊ณผ ์‹ค๋ฌด ๊ฒฉ์ฐจ : NEJM ๋“ฑ ๊ณ ์œ„ํ—˜ ์ €๋„์€ ๊ฒฝ์Ÿ์œ„ํ—˜์„ โ€œ์‹ค์งˆ์ ์ธ ๋น„์œจ์ด ํด ๊ฒฝ์šฐโ€ Fineโ€‘Gray ๋ชจ๋ธ์„ ๊ณ ๋ คํ•˜๋„๋ก ๊ถŒ๊ณ ํ•˜์ง€๋งŒ, โ€œ์–ผ๋งˆ๋‚˜ ํฐ ๋น„์œจโ€์ธ์ง€ ์ •๋Ÿ‰์  ๊ธฐ์ค€์„ ์ œ์‹œํ•˜์ง€ ์•Š๋Š”๋‹ค. ์ถ”์ •๋Ÿ‰(estimand) ์ฐจ์ด : Cox๋Š” ์กฐ๊ฑด๋ถ€ ์œ„ํ—˜๋น„ (causeโ€‘specific hazard) โ†’ โ€œ์‚ฌ๊ฑด์ด ์•„์ง ๋ฐœ์ƒํ•˜์ง€ ์•Š์€ ํ™˜์ž๊ตฐโ€์— ๋Œ€ํ•œ ํšจ๊ณผ๋ฅผ ์ถ”์ •ํ•˜๊ณ , Fin

Statistics Analysis Model
No Image

Concept Influence: Leveraging Interpretability to Improve Performance and Efficiency in Training Data Attribution

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ๊ธฐ์กด TDA์˜ ํ•œ๊ณ„ ๋ฐ์ดํ„ฐ ํฌ์ธํŠธ ์ˆ˜์ค€ ๊ท€์†์€ ์˜๋ฏธ์  ์›์ธ์„ ํŒŒ์•…ํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ๋‹จ์ผ ํ…Œ์ŠคํŠธ ์˜ˆ์‹œ ์— ์˜์กดํ•˜๋ฉด ๊ตฌ๋ฌธยทํ‘œ๋ฉด์  ์œ ์‚ฌ๋„ ์— ๊ณผ๋„ํ•˜๊ฒŒ ํŽธํ–ฅ๋œ๋‹ค(์˜ˆ: BM25๊ฐ€ ์˜ํ–ฅ ํ•จ์ˆ˜๋ณด๋‹ค ์šฐ์ˆ˜ํ•œ ๊ฒฝ์šฐ). ๋Œ€๊ทœ๋ชจ LLM์— ์ ์šฉํ•˜๋ ค๋ฉด Hessian ์—ญํ–‰๋ ฌ ๊ณ„์‚ฐ ๋“ฑ ๊ณ ๋น„์šฉ ์—ฐ์‚ฐ์ด ํ•„์ˆ˜์ ์ด๋‹ค. ๊ฐœ๋… ๊ธฐ๋ฐ˜ ํ•ด์„์˜ ํ•„์š”์„ฑ ์ด๋ฏธ์ง€ยท์–ธ์–ด ๋ถ„์•ผ์—์„œ Concept Activation Vectors(CAV) , Persona Vectors ๋“ฑ ์˜๋ฏธ์  ๋ฐฉํ–ฅ์ด ํ–‰๋™์„ ์ œ์–ดํ•œ๋‹ค๋Š” ์—ฐ๊ตฌ๊ฐ€ ๋Š˜์–ด๋‚˜๊ณ  ์žˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋ฐฉํ–ฅ์„ TDA์— ์ง์ ‘ ์—ฐ

Data Computer Science Artificial Intelligence
No Image

Conditions for Bacterial Selection and Extinction Driven by Growth-Kill Trade-Off in Cyclic Antimicrobial Treatments

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํ˜„์‹ค์  ๋ฌธ์ œ : ํ•ญ๊ท ์ œ ์‚ฌ์šฉ์€ ์ผ์ƒ์ ์ด์ง€๋งŒ, ๋™์ผ ์ง‘๋‹จ ๋‚ด์—์„œ๋„ ์ €ํ•ญ์„ฑยท๋‚ด์„ฑ ๋ณ€์ด๊ฐ€ ๋ฐœ์ƒํ•ด ์น˜๋ฃŒ ์‹คํŒจ๋ฅผ ์ดˆ๋ž˜ํ•œ๋‹ค. ๊ธฐ์กด ๋ชจ๋ธ์˜ ํ•œ๊ณ„ : ๋Œ€๋ถ€๋ถ„์˜ ๊ธฐ์กด ๋ชจ๋ธ์€ (i) ์ผ์ •ํ•œ ํ•ญ๊ท ์ œ ๋†๋„ ๊ฐ€์ •, (ii) ๋ณต์žกํ•œ stochastic birthโ€‘death ๊ณผ์ •, (iii) ๋‹ค์ˆ˜ ํŒŒ๋ผ๋ฏธํ„ฐ ํ•„์š” ๋“ฑ์œผ๋กœ ์‹คํ—˜ ์ ์šฉ์ด ์–ด๋ ค์› ๋‹ค. ๋ณธ ์—ฐ๊ตฌ์˜ ์ฐจ๋ณ„์  : ์ˆœํ™˜(์ฃผ๊ธฐ์ ) ์ฒ˜๋ฆฌ๋ผ๋Š” ์‹ค์ œ ํ”„๋กœํ† ์ฝœ์— ์ดˆ์ ์„ ๋งž์ถ”๊ณ , ๋‘ ๊ฐœ์˜ ํ•ต์‹ฌ ํŠน์„ฑ(์„ฑ์žฅ๋ฅ , ์‚ดํ•ด์œจ) ๋งŒ์œผ๋กœ ํ˜„์ƒ์„ ์„ค๋ช…ํ•œ๋‹ค๋Š” ์ ์—์„œ ๋‹จ์ˆœํ•˜๋ฉด์„œ๋„ ์‹ค์šฉ์ ์ด๋‹ค. 2. ๋ชจ๋ธ ๊ตฌ์กฐ ๋ฐ ํ•ต

Quantitative Biology
No Image

Conference Proceedings of the Inaugural Conference of the International Society for Tractography (IST 2025 Bordeaux)

1. ์—ฐ๊ตฌ์˜ ์ „์ฒด์  ์˜์˜ ํ†ตํ•ฉ์  ํ”Œ๋žซํผ ๊ตฌ์ถ• : ๋‹ค์–‘ํ•œ ์—ฐ๋ นยท์ข…(species)ยท์ž„์ƒ ์ƒํ™ฉ์„ ์•„์šฐ๋ฅด๋Š” ํŠธ๋ž™ํ† ๊ทธ๋ž˜ํ”ผ ์—ฐ๊ตฌ๋ฅผ ํ•˜๋‚˜์˜ ํ•™์ˆ  ํ–‰์‚ฌ์— ๋ชจ์•„, ๋ถ„์•ผ ๊ฐ„ ์‹œ๋„ˆ์ง€์™€ ๋ฐ์ดํ„ฐยท๋ฐฉ๋ฒ•๋ก  ํ‘œ์ค€ํ™”๋ฅผ ์ด‰์ง„ํ•œ๋‹ค. ์ž„์ƒยท๊ธฐ์ดˆยท๊ธฐ์ˆ  ์œตํ•ฉ : ์‹ ์ƒ์•„ ๋‡Œ ๋ฐœ๋‹ฌ, ์ •์‹ ยท์‹ ๊ฒฝ์งˆํ™˜, ๋‡Œโ€‘์ฒ™์ˆ˜ ์ž๊ทน(Deep Brain Stimulation) ํƒ€๊ฒŒํŒ…, ๊ทธ๋ฆฌ๊ณ  ์˜์žฅ๋ฅ˜ ๋น„๊ตํ•ด๋ถ€ํ•™๊นŒ์ง€, ํŠธ๋ž™ํ† ๊ทธ๋ž˜ํ”ผ๊ฐ€ ๋‹จ์ผ ๊ธฐ์ˆ ์„ ๋„˜์–ด ๋‹คํ•™์ œ์  ์—ฐ๊ฒฐ ๊ณ ๋ฆฌ ์—ญํ• ์„ ํ•จ์„ ๋ณด์—ฌ์ค€๋‹ค. 2. ํ•ต์‹ฌ ์—ฐ๊ตฌ๋ณ„ ์ƒ์„ธ ๋ถ„์„ | | ์—ฐ๊ตฌ ์ฃผ์ œ | ์ฃผ์š” ๋ฐฉ๋ฒ•ยท๊ธฐ์ˆ  | ํ•ต์‹ฌ ๊ฒฐ๊ณผ | ์ž„์ƒยทํ•™๋ฌธ์  ํŒŒ๊ธ‰ํšจ๊ณผ

Image Processing Electrical Engineering and Systems Science
No Image

Confidence as Forecast: A Decision-Theoretic Interpretation of Confidence Intervals

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ œ๊ธฐ Neymanโ€‘Frequentist ์ „ํ†ต : ฮธ๋Š” ๊ณ ์ •๋œ ๋ฏธ์ง€์ˆ˜์ด๋ฉฐ, ๊ตฌ๊ฐ„์ด ๋งŒ๋“ค์–ด์ง„ ๋’ค์—๋Š” ๋ฎ์Œ ์—ฌ๋ถ€๊ฐ€ ํ™•์ •์ (0 ๋˜๋Š” 1)์ด๋ผ๋Š” โ€œ๋ฎ์Œ์€ ํ™•๋ฅ ์ด ์•„๋‹ˆ๋‹คโ€ ์ž…์žฅ. ์‹ค์ œ ํ˜ผ๋ž€ : ํ•™์ƒยท์‹ค๋ฌด์ž ๋ชจ๋‘ โ€œ๊ตฌ๊ฐ„์ด 95 % ์‹ ๋ขฐ ์ˆ˜์ค€์ด๋ฉด ์‚ฌํ›„์—๋„ 95 % ํ™•๋ฅ ๋กœ ฮธ๋ฅผ ํฌํ•จํ•œ๋‹คโ€๋Š” ์ง๊ด€๊ณผ, โ€œ๋ฎ์Œ์€ 0/1์ด๋ฏ€๋กœ ํ™•๋ฅ ์„ ๋งํ•  ์ˆ˜ ์—†๋‹คโ€๋Š” ๊ต๊ณผ์„œ์  ์„ค๋ช… ์‚ฌ์ด์—์„œ ๊ฐˆ๋“ฑ์„ ๊ฒช์Œ. ๋ณธ ๋…ผ๋ฌธ์˜ ํ•ต์‹ฌ ์งˆ๋ฌธ : โ€œ๋นˆ๋„์ฃผ์˜์  ๋ชจ๋ธ ์•ˆ์—์„œ, ์‚ฌํ›„์—๋„ ์˜๋ฏธ ์žˆ๋Š” ํ™•๋ฅ  ์˜ˆ๋ณด๋ฅผ ์ œ๊ณตํ•  ์ˆ˜ ์žˆ๋Š”๊ฐ€?โ€ 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ํ™•๋ฅ  ์˜ˆ๋ณด๋กœ์„œ์˜

Statistics
No Image

Confidence Distributions for FIC scores

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ œ๊ธฐ | ๊ธฐ์กด FIC ์ ‘๊ทผ | ํ•œ๊ณ„ | | | | | FIC plot์— ์  ์ถ”์ •๊ฐ’(p) ๊ณผ rootโ€‘FIC (์ถ”์ • ์œ„ํ—˜)๋งŒ ํ‘œ์‹œ | ๊ฐ ์ ์ด ์ถ”์ • ์˜ค์ฐจ ์™€ ํŽธํ–ฅ ์„ ํฌํ•จํ•œ๋‹ค๋Š” ์‚ฌ์‹ค์„ ๋ฌด์‹œ โ†’ โ€œ์ตœ์  ๋ชจ๋ธโ€์ด ์‹ค์ œ๋กœ ์œ ์˜๋ฏธํ•œ ์ฐจ์ด๋ฅผ ๋ณด์ด๋Š”์ง€ ํŒ๋‹จ ์–ด๋ ค์›€ | | ๋ชจ๋ธ ์„ ํƒ ์‹œ ํŽธํ–ฅ ๋ณด์ • ์€ ์ž„์˜์ (์˜ˆ: 0์œผ๋กœ ์ ˆ๋‹จ) | ํŽธํ–ฅ ์ถ”์ •์ด ๋ถˆ์•ˆ์ •ํ•˜๋ฉด ์„ ํƒ ๊ฒฐ๊ณผ๊ฐ€ ํฌ๊ฒŒ ๋ณ€๋™ | 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ Confidence Distribution (CD) ์ ์šฉ CD ๋Š” ํŒŒ๋ผ๋ฏธํ„ฐ์— ๋Œ€ํ•œ ์ „์ฒด ๋ˆ„์  ํ™•๋ฅ  ๋ถ„ํฌ๋ฅผ ์ œ๊ณต, ์ฆ‰ โ€œ์ „

Statistics
No Image

Conformal Prediction-Driven Adaptive Sampling for Digital Water Twins

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋””์ง€ํ„ธ ํŠธ์œˆ๊ณผ ๋ฌผ ๊ณต๊ธ‰๋ง : ๋ฌผ ๊ณต๊ธ‰๋ง์€ ๋ณต์žกํ•œ ๋™์  ์‹œ์Šคํ…œ์œผ๋กœ, ์‹ค์‹œ๊ฐ„ ์ƒํƒœ ์ถ”์ •์ด ์šด์˜ ํšจ์œจ๊ณผ ์‚ฌ๊ณ  ๋ฐฉ์ง€์— ํ•ต์‹ฌ์ ์ž…๋‹ˆ๋‹ค. ํ•˜์ง€๋งŒ ์„ผ์„œ ์„ค์น˜ ๋น„์šฉ๊ณผ ์œ ์ง€๋ณด์ˆ˜ ์ œ์•ฝ์œผ๋กœ ์ „์ฒด ๋„คํŠธ์›Œํฌ์— ๊ณ ๋ฅด๊ฒŒ ์„ผ์„œ๋ฅผ ๋ฐฐ์น˜ํ•˜๊ธฐ ์–ด๋ ต์Šต๋‹ˆ๋‹ค. ๋ถˆํ™•์‹ค์„ฑ ๊ธฐ๋ฐ˜ ์ƒ˜ํ”Œ๋ง์˜ ์˜์˜ : ๋…ธ๋“œ๋งˆ๋‹ค ์ˆ˜์š” ๋ณ€๋™์„ฑ๊ณผ ๋ชจ๋ธ๋ง ์˜ค์ฐจ๊ฐ€ ๋‹ค๋ฅด๊ธฐ ๋•Œ๋ฌธ์—, ๋ถˆํ™•์‹ค์„ฑ์ด ํฐ ๋…ธ๋“œ์— ์„ผ์„œ๋ฅผ ์ง‘์ค‘ํ•˜๋ฉด ์ „์ฒด ์ถ”์ • ์ •ํ™•๋„๋ฅผ ํฌ๊ฒŒ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. 2. ํ•ต์‹ฌ ๋ฐฉ๋ฒ•๋ก  | ๋‹จ๊ณ„ | ์„ค๋ช… | ์ฃผ์š” ๊ธฐ์ˆ  | | | | | | ์˜ˆ์ธก ๋ชจ๋ธ | ๊ฐ ๋…ธ๋“œ์˜ ๋ฏธ๋ž˜ ์ˆ˜์š”๋ฅผ LSTM(

No Image

Conjugate Learning Theory: Uncovering the Mechanisms of Trainability and Generalization in Deep Neural Networks

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ํ•™์Šต ๊ฐ€๋Šฅ์„ฑ(Trainability) : ๊ณผ์ž‰ ํŒŒ๋ผ๋ฏธํ„ฐํ™”๋œ ๋น„๋ณผ๋ก DNN์ด SGD๋กœ๋„ ๋‚ฎ์€ ๊ฒฝํ—˜์  ์œ„ํ—˜์„ ๋‹ฌ์„ฑํ•œ๋‹ค๋Š” ํ˜„์ƒ์€ ๊ธฐ์กด ๋น„๋ณผ๋ก ์ตœ์ ํ™” ์ด๋ก ์œผ๋กœ๋Š” ์„ค๋ช…๋˜์ง€ ์•Š๋Š”๋‹ค. ์ผ๋ฐ˜ํ™”(Generalization) : VCโ€‘dimensionยทRademacher ๋ณต์žก๋„ ๊ธฐ๋ฐ˜ ๊ณ ์ „ ์ด๋ก ์€ ๊ณผ์ž‰ ํŒŒ๋ผ๋ฏธํ„ฐํ™” ์ƒํ™ฉ์—์„œ ์‹ค์ œ ์„ฑ๋Šฅ์„ ๊ณผ์†Œํ‰๊ฐ€ํ•œ๋‹ค(โ€˜์ผ๋ฐ˜ํ™” ์—ญ์„คโ€™). 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ์ปจ์ฃผ๊ฒŒ์ดํŠธ ํ•™์Šต ์ด๋ก  | ๋‹จ๊ณ„ | ๋‚ด์šฉ | ์˜์˜ | | | | | | ์กฐ๊ฑด๋ถ€ ๋ถ„ํฌ ๊ด€์  | ํ•™์Šต ๊ณผ์ œ๋Š” (P(Y|X)) ๋ฅผ ์ถ”์ •ํ•˜๋Š” ๋ฌธ์ œ

Statistics Network Learning Machine Learning
No Image

Consensus Based Task Allocation for Angles-Only Local Catalog Maintenance of Satellite Systems

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ LEO(์ €๊ถค๋„) ํ˜ผ์žก๋„ ์ฆ๊ฐ€ โ†’ ์ถฉ๋Œ ํšŒํ”ผยทํ˜‘๋™ ์ž„๋ฌด(๊ตฌ์กฐ, ๊ฒ€์‚ฌ, ํŽธ๋Œ€ ๋น„ํ–‰ ๋“ฑ) ์ˆ˜ํ–‰์— ์ •ํ™•ํ•œ ์ƒ๋Œ€ ์ƒํƒœ ์ •๋ณด๊ฐ€ ํ•„์ˆ˜. ์ง€์ƒ ๊ธฐ๋ฐ˜ ํŠธ๋ž˜ํ‚น ํ•œ๊ณ„ : ๊ด€์ธก ์ •ํ™•๋„ยท์‹œ๊ฐ„ ์ง€์—ฐ ๋ฌธ์ œ๋กœ ์œ„์„ฑ ์ž์ฒด ์„ผ์„œ์™€ ํ˜‘์—…์ด ์š”๊ตฌ๋จ. ๊ฐ๋„โ€‘์ „์šฉ ์„ผ์„œ ๋Š” ํ•˜๋“œ์›จ์–ด๊ฐ€ ๊ฐ„๋‹จํ•˜๊ณ  ์ „๋ ฅยท๋ฌด๊ฒŒ๊ฐ€ ์ ์ง€๋งŒ, ๊ฑฐ๋ฆฌ ์ •๋ณด๊ฐ€ ์—†์œผ๋ฏ€๋กœ ๊ด€์ธก ๋Œ€์ƒ๊ณผ์˜ ๊ฑฐ๋ฆฌยท๊ณต๋ถ„์‚ฐ ํ˜•ํƒœ ๋ฅผ ๊ณ ๋ คํ•œ ์ž‘์—… ์„ ํƒ์ด ํ•„์š”. 2. ๊ธฐ์กด ์ ‘๊ทผ๋ฒ•๊ณผ ํ•œ๊ณ„ | ์ ‘๊ทผ๋ฒ• | ํŠน์ง• | ํ•œ๊ณ„ | | | | | | ์ค‘์•™์ง‘์ค‘์‹ ์Šค์ผ€์ค„๋ง | ์ „์ฒด ์ •๋ณด๋ฅผ ์ˆ˜์ง‘ ํ›„ ์ตœ์ ํ™” | ํ†ต์‹  ๋ถ€ํ•˜ยท๋‹จ์ผ์  ๊ณ 

Computer Science System Multiagent Systems
No Image

Continual Learning Strategies for 3D Engineering Regression Problems: A Benchmarking Study

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฐ์ดํ„ฐ ์ œํ•œ๊ณผ ๊ณ„์‚ฐ ๋น„์šฉ : ์—”์ง€๋‹ˆ์–ด๋ง ์‹œ๋ฎฌ๋ ˆ์ด์…˜(์˜ˆ: CFD, FEM)์€ ๊ณ ๋น„์šฉ ์—ฐ์‚ฐ์„ ์š”๊ตฌํ•œ๋‹ค. ๋”ฐ๋ผ์„œ ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘์ด ์ œํ•œ์ ์ด๋ฉฐ, ์ƒˆ๋กœ์šด ์„ค๊ณ„๊ฐ€ ์ถ”๊ฐ€๋  ๋•Œ๋งˆ๋‹ค ์ „์ฒด ๋ฐ์ดํ„ฐ๋ฅผ ๋‹ค์‹œ ํ•™์Šต์‹œํ‚ค๋Š” ๊ฒƒ์€ ๋น„ํšจ์œจ์ ์ด๋‹ค. ์—ฐ์†์ ์ธ ์„ค๊ณ„ ๋ณ€ํ™” : ์‹ค์ œ ์ œํ’ˆ ๊ฐœ๋ฐœ ์ฃผ๊ธฐ์—์„œ๋Š” ์„ค๊ณ„ ํŒŒ๋ผ๋ฏธํ„ฐ, ์žฌ๋ฃŒ ํŠน์„ฑ, ํ™˜๊ฒฝ ์ œ์•ฝ ๋“ฑ์ด ์ง€์†์ ์œผ๋กœ ์—…๋ฐ์ดํŠธ๋œ๋‹ค. ๋ชจ๋ธ์ด ์ด๋Ÿฌํ•œ ๋ณ€ํ™”๋ฅผ ๋ฐ˜์˜ํ•˜๋ ค๋ฉด ์ ์ง„์  ํ•™์Šต ์ด ํ•„์ˆ˜์ ์ด๋‹ค. 2. ์—ฐ์†ํ•™์Šต(Continual Learning) ๊ฐœ๋… ์ •๋ฆฌ Catastrophic Forgetting : ์ƒˆ ๋ฐ์ดํ„ฐ

Learning
No Image

Continuous Fluid Antenna Sampling for Channel Estimation in Cell-Free Massive MIMO

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ๊ธฐ์กด์˜ ์œ ์ฒด ์•ˆํ…Œ๋‚˜(FAS) ์—ฐ๊ตฌ๋Š” ๋Œ€๋ถ€๋ถ„ ์•ˆํ…Œ๋‚˜ ์œ„์น˜๋ฅผ ์œ ํ•œํ•œ ํฌํŠธ ์ง‘ํ•ฉ์œผ๋กœ ์ด์‚ฐํ™”ํ•˜๊ณ , ํฌํŠธ ์„ ํƒ์„ ์ตœ์ ํ™”ํ•˜๋Š” ๋ฐฉ์‹์— ๋จธ๋ฌผ๋ €๋‹ค. ์ด๋Š” ์‹ค์ œ ์—ฐ์†์ ์ธ ์œ ์ฒด ์›€์ง์ž„์„ ๊ทผ์‚ฌํ™”ํ•จ์œผ๋กœ์จ ์ •๋ณด ์†์‹ค๊ณผ ํฌํŠธ ์ „ํ™˜์— ๋”ฐ๋ฅธ ์ง€์—ฐยท์ „๋ ฅ ์†Œ๋ชจ๋ฅผ ์ดˆ๋ž˜ํ•œ๋‹ค. CFโ€‘mMIMO ํ™˜๊ฒฝ์—์„œ๋Š” ๋‹ค์ˆ˜์˜ AP๊ฐ€ ํ˜‘์—…ํ•ด ์ฑ„๋„ ์ถ”์ •์„ ์ˆ˜ํ–‰ํ•˜๋ฏ€๋กœ, ๊ฐ AP๊ฐ€ ๊ฐ€๋Šฅํ•œ ํ•œ ํ’๋ถ€ํ•œ ๊ณต๊ฐ„ ์ •๋ณด๋ฅผ ์ œ๊ณตํ•˜๋Š” ๊ฒƒ์ด ์ „์ฒด ์‹œ์Šคํ…œ ์„ฑ๋Šฅ์— ํฌ๊ฒŒ ๊ธฐ์—ฌํ•œ๋‹ค. ๋”ฐ๋ผ์„œ ์—ฐ์†์ ์ธ ์œ„์น˜ ์ƒ˜ํ”Œ๋ง์„ ํ†ตํ•œ ์ฑ„๋„ ๊ด€์ธก์ด ํ•„์š”ํ•˜๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ์—ฐ์† FA ๋ชจ๋ธ๋ง : ์•ˆํ…Œ๋‚˜

Computer Science Information Theory
No Image

Controlled oscillation modeling using port-Hamiltonian neural networks

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ | ๋ฌธ์ œ์  | ๊ธฐ์กด ์ ‘๊ทผ๋ฒ• | ํ•œ๊ณ„ | | | | | | ๋ฐ์ดํ„ฐโ€‘๋งŒ ํ•™์Šต | ์ˆœ์ˆ˜ Neural ODE, PINN ๋“ฑ | ๋ฌผ๋ฆฌ ๋ฒ•์น™(์—๋„ˆ์ง€ ๋ณด์กดยท์ „๋ ฅ ๊ท ํ˜•) ๋ฏธ๋ฐ˜์˜ โ†’ ์ผ๋ฐ˜ํ™” ๋ถˆ์•ˆ์ • | | PHNN | ํฌํŠธโ€‘ํ•ด๋ฐ€ํ† ๋‹ˆ์•ˆ ๊ตฌ์กฐ๋ฅผ ํ•˜๋“œ ์ œ์•ฝ์œผ๋กœ ์‚ฝ์ž… | ๋Œ€๋ถ€๋ถ„ RK ๊ธฐ๋ฐ˜ ์ด์‚ฐํ™” โ†’ ์—๋„ˆ์ง€ ์†์‹ค, ๊ตฌ์กฐ ๋ณด์กด ๋ฏธํก | | ์ˆ˜์น˜ ์•ˆ์ •์„ฑ | ๊ณ ์ฐจ RK, ์ž๋™ ๋ฏธ๋ถ„ | ๊ฐ•์ง(stiff) ๋ฌธ์ œ ์‹œ ๋งค์šฐ ์ž‘์€ ์Šคํ… ์š”๊ตฌ, Jacobian ์•…์กฐ๊ฑดํ™” | ์—ฐ๊ตฌ์ž๋Š” ์œ„ ํ•œ๊ณ„๋ฅผ โ€œ์ด์‚ฐ ๊ทธ๋ž˜๋””์–ธํŠธ + ํ•˜๋“œ ํฌํŠธโ€‘ํ•ด๋ฐ€ํ† ๋‹ˆ์•ˆ ์ œ์•ฝโ€ ์ด๋ผ๋Š”

Computer Science Network Machine Learning Model
No Image

Corrected-Inverse-Gaussian First-Hitting-Time Modeling for Molecular Communication Under Time-Varying Drift

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ถ„์ž ํ†ต์‹  ์ฑ„๋„ ์€ ํ™•์‚ฐยท๋Œ€๋ฅ˜์— ์˜ํ•ด ์‹ ํ˜ธ ๋ถ„์ž๊ฐ€ ์ด๋™ํ•˜๋Š” ํ™•๋ฅ  ๊ณผ์ •์œผ๋กœ, ํก์ˆ˜ํ˜• ์ˆ˜์‹ ๊ธฐ์—์„œ๋Š” ์ž…์ž ๋„์ฐฉ ์‹œ์ ์„ ๋‚˜ํƒ€๋‚ด๋Š” FHT ๋ถ„ํฌ๊ฐ€ ํ•ต์‹ฌ ๋ชจ๋ธ๋ง ๋Œ€์ƒ์ด๋‹ค. ์ •์ƒ(์‹œ๊ฐ„ ๋ถˆ๋ณ€) ๋“œ๋ฆฌํ”„ํŠธ ์—์„œ๋Š” FHT๊ฐ€ ์—ญ๊ฐ€์šฐ์‹œ์•ˆ(Inverseโ€‘Gaussian, IG) ๋ถ„ํฌ๋ฅผ ๋”ฐ๋ฅธ๋‹ค๋Š” ๊ฒƒ์ด ์ž˜ ์•Œ๋ ค์ ธ ์žˆ์–ด, ์‹œ์Šคํ…œ ์„ค๊ณ„์™€ ์„ฑ๋Šฅ ๋ถ„์„์— ๋„๋ฆฌ ํ™œ์šฉ๋œ๋‹ค. ์‹ค์ œ ์ƒ๋ฌผํ•™ยท๋งˆ์ดํฌ๋กœํ”Œ๋ฃจ์ด๋”• ํ™˜๊ฒฝ์€ ์‹œ๊ฐ„์— ๋”ฐ๋ผ ๋ณ€ํ•˜๋Š” ํ๋ฆ„ (๋งฅ๋™ํ˜• ํ˜ˆ๋ฅ˜, ์ „๊ธฐ์˜๋™ ๋“ฑ)์œผ๋กœ, ์ด๋Ÿฌํ•œ ๋น„์ •์ƒ์„ฑ์€ ๊ธฐ์กด IG ๊ธฐ๋ฐ˜ ๋ชจ๋ธ๋กœ๋Š” ์„ค๋ช…์ด ๋ถˆ๊ฐ€๋Šฅํ•˜๊ณ , ํ˜„์žฌ๋Š” ์ˆ˜์น˜ PD

Computer Science Information Theory Model
No Image

Covariate Adjustment for Wilcoxon Two Sample Statistic and Test

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์œŒ์ฝ•์Šจโ€‘Mannโ€‘Whitney ๊ฒ€์ • ์€ ๋น„๋ชจ์ˆ˜์  ๋Œ€์•ˆ์ด์ง€๋งŒ, ๊ณต๋ณ€๋Ÿ‰ ์ •๋ณด๋ฅผ ์ „ํ˜€ ํ™œ์šฉํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์ž„์ƒ์‹œํ—˜ ๋“ฑ์—์„œ ๊ณต๋ณ€๋Ÿ‰โ€‘์ ์‘ ๋žœ๋คํ™” ๊ฐ€ ๋„๋ฆฌ ์“ฐ์ด๋ฉด์„œ, ๊ธฐ์กด ๋น„์กฐ์ • ๊ฒ€์ •์€ ๋žœ๋คํ™” ๋ฐฉ์‹์— ๋”ฐ๋ผ ๋‹ค๋ฅธ asymptotic variance๋ฅผ ๊ฐ€์ ธ ํ†ต๊ณ„์‹ ์žฌ๊ตฌ์„ฑ์ด ํ•„์š” ํ•ด์กŒ๋‹ค. ๊ทœ์ œ๊ธฐ๊ด€(ICH E9, EMA, FDA)์€ โ€œ๊ฐ€๋Šฅํ•œ ์ตœ์†Œ ๊ฐ€์ • ํ•˜์— ๊ณต๋ณ€๋Ÿ‰์„ ํ™œ์šฉํ•˜๋ผโ€๋Š” ์ง€์นจ์„ ์ œ์‹œํ•˜๊ณ  ์žˆ์–ด, ๋ชจ๋ธโ€‘ํ”„๋ฆฌ ์ด๋ฉด์„œ ํšจ์œจ์„ฑ์„ ๋ณด์žฅ ํ•˜๋Š” ๋ฐฉ๋ฒ•์ด ์š”๊ตฌ๋œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด โ€“ ๊ณต๋ณ€๋Ÿ‰ ๋ณด์ •(Calibration) ๋ณด์ •์‹ :

Statistics
No Image

Coverage Guarantees for Pseudo-Calibrated Conformal Prediction under Distribution Shift

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์ปจํฌ๋ฉ€ ์˜ˆ์ธก ์€ ๊ต์ • ๋ฐ์ดํ„ฐ์™€ ํ…Œ์ŠคํŠธ ๋ฐ์ดํ„ฐ๊ฐ€ ๊ตํ™˜๊ฐ€๋Šฅ์ผ ๋•Œ, ์œ ํ•œ ํ‘œ๋ณธ์—์„œ๋„ ์ •ํ™•ํžˆ 1โ€‘ฮฑ ์ˆ˜์ค€์˜ ๋งˆ์ง„ ์ปค๋ฒ„๋ฆฌ์ง€๋ฅผ ๋ณด์žฅํ•œ๋‹ค. ๋ถ„ํฌ ์ด๋™ (distribution shift) ์ƒํ™ฉ์—์„œ๋Š” ๊ตํ™˜๊ฐ€๋Šฅ ๊ฐ€์ •์ด ๊นจ์ง€๋ฉฐ, ํŠนํžˆ ๋ผ๋ฒจ์ด ์—†๋Š” ํƒ€๊นƒ ๋„๋ฉ”์ธ์—์„œ๋Š” ๊ธฐ์กด ๊ฐ€์ค‘์น˜ CP๋‚˜ ๋ถˆํ™•์‹ค์„ฑ ๊ธฐ๋ฐ˜ ์Šค์ฝ”์–ด ์žฌ์กฐ์ • ๋ฐฉ๋ฒ•์ด ์ด๋ก ์  ๋ณด์žฅ์„ ์ œ๊ณตํ•˜์ง€ ๋ชปํ•œ๋‹ค. ๋ผ๋ฒจโ€‘์กฐ๊ฑด๋ถ€ ๊ณต๋ณ€๋Ÿ‰ ์ด๋™ ์„ ๊ฐ€์ •ํ•˜๋ฉด, ๊ฐ ๋ผ๋ฒจ๋ณ„ ์ž…๋ ฅ ๋ถ„ํฌ๋งŒ์ด ์ œํ•œ๋œ Wasserstein ๊ฑฐ๋ฆฌ ฯ ์ดํ•˜๋กœ ๋ณ€ํ•œ๋‹ค๋Š” ๊ฐ•๋ ฅํ•œ ๊ตฌ์กฐ์  ๊ฐ€์ •์„ ํ™œ์šฉํ•  ์ˆ˜ ์žˆ๋‹ค. 2. ํ•ต์‹ฌ ์ด๋ก 

Computer Science Machine Learning
No Image

Critical thresholds for semilinear damped wave equations with Riesz potential power nonlinearity and initial data in pseudo-measure spaces

| ๊ตฌ๋ถ„ | ๋‚ด์šฉ | ํ‰๊ฐ€ยท์˜์˜ | | | | | | ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ | ๊ฐ์‡  ํŒŒ๋™ ๋ฐฉ์ •์‹์€ ์—ด ๋ฐฉ์ •์‹๊ณผ์˜ ํ™•์‚ฐโ€‘ํšจ๊ณผ ์—ฐ๊ฒฐ ๋•Œ๋ฌธ์— Fujita ์ง€์ˆ˜๊ฐ€ ์ž„๊ณ„๊ฐ’์œผ๋กœ ์•Œ๋ ค์ ธ ์žˆ๋‹ค. ์ตœ๊ทผ์—๋Š” ์ดˆ๊ธฐ ๋ฐ์ดํ„ฐ์˜ ์ €์ฃผํŒŒ ํŠน์„ฑ์„ ๋ฐ˜์˜ํ•œ pseudoโ€‘measure(๐’ดแต ) ๊ณต๊ฐ„์ด Navierโ€‘Stokesยท๋ถ„์ˆ˜ Navierโ€‘Stokes ๋“ฑ์—์„œ ์œ ์šฉํ•จ์ด ์ž…์ฆ๋˜์—ˆ๋‹ค. | ์ €์ฃผํŒŒ ์ œ์–ด๊ฐ€ ํ•ด์˜ ์žฅ๊ธฐ ๊ฑฐ๋™์— ๋ฏธ์น˜๋Š” ์˜ํ–ฅ์„ ์ •๋Ÿ‰ํ™”ํ•˜๋ ค๋Š” ์‹œ๋„๋Š” ๋งค์šฐ ์‹œ์˜์ ์ ˆํ•˜๋ฉฐ, ๊ธฐ์กด ์—ฐ๊ตฌ์™€ ์ฐจ๋ณ„ํ™”๋œ ์ ‘๊ทผ์ด๋‹ค. | | ํ•ต์‹ฌ ์•„์ด๋””์–ด | โ‘  ์„ ํ˜• ๊ฐ์‡  ํŒŒ๋™์˜ ์ €์ฃผํŒŒโ€‘๊ณ ์ฃผํŒŒ ๋ถ„ํ•ด๋ฅผ Four

Data Mathematics
Cultural Rights and the Rights to Development in the Age of AI: Implications for Global Human Rights Governance

Cultural Rights and the Rights to Development in the Age of AI: Implications for Global Human Rights Governance

์ด ๋…ผ๋ฌธ์€ ์ธ๊ณต์ง€๋Šฅ(AI)์˜ ๋ฐœ์ „์ด ๋ฌธํ™”์  ๊ถŒ๋ฆฌ์™€ ๋ฐœ์ „์˜ ๊ถŒ๋ฆฌ๋ฅผ ์–ด๋–ป๊ฒŒ ๋ณ€ํ™”์‹œํ‚ค๋Š”์ง€๋ฅผ ํƒ๊ตฌํ•ฉ๋‹ˆ๋‹ค. AI๊ฐ€ ๋ฌธํ™” ์ฝ˜ํ…์ธ ์˜ ์ฐฝ์กฐ์™€ ํ‘œํ˜„์— ์˜ํ–ฅ์„ ๋ฏธ์น˜๋ฉด์„œ, ๊ฐœ์ธ๊ณผ ์ปค๋ฎค๋‹ˆํ‹ฐ์˜ ์ง€์  ์žฌ์‚ฐ ์‚ฌ์šฉ ๋ฐ ๋ถ„๋ฐฐ์—๋„ ๋ณ€ํ™”๋ฅผ ๊ฐ€์ ธ์˜ต๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋ณ€ํ™”๋Š” ๊ฒฝ์ œ์ , ์‚ฌํšŒ์ , ๋””์ง€ํ„ธ ๊ฒฉ์ฐจ๋ฅผ ๋”์šฑ ๋ฒŒ๋ฆฌ๊ณ  ์†Œ์ˆ˜์ž ์ปค๋ฎค๋‹ˆํ‹ฐ์— ๋Œ€ํ•œ ๋ถˆํ‰๋“ฑ์„ ๊ฐ•ํ™”ํ•  ์œ„ํ—˜๋„ ๋‚ดํฌํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค. ๋…ผ๋ฌธ์€ AI์˜ ์•Œ๊ณ ๋ฆฌ์ฆ˜ ์„ค๊ณ„์™€ ๋ฐฐ์น˜์—์„œ ๋ฌธํ™”์  ๋ฐ ๋ฐœ์ „์  ๊ฐ€์ •์ด ์–ด๋–ป๊ฒŒ ๋ฐ˜์˜๋˜๋Š”์ง€ ๋ถ„์„ํ•˜๋ฉฐ, ์ด๋กœ ์ธํ•œ ๊ฐœ์ธ๊ณผ ๊ตฌ์กฐ์ ์ธ ์˜ํ–ฅ์„ ์‚ดํŽด๋ด…๋‹ˆ๋‹ค. ๋˜ํ•œ, ๊ธฐ์กด์˜ AI ํ†ต์ œ ํ”„๋ ˆ์ž„์›Œํฌ ๋‚ด์—์„œ ๋ฌธํ™”์  ๊ถŒ๋ฆฌ

No Image

Daring few, patient many: division of labor in decentralized foraging collectives

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ์งˆ๋ฌธ ํ•ต์‹ฌ ์งˆ๋ฌธ : ๋ฆฌ๋”๊ฐ€ ์—†๋Š” ์ง‘๋‹จ์ด ์ค‘์•™์ง‘์ค‘์‹ ์˜์‚ฌ๊ฒฐ์ •๊ณผ ๋™๋“ฑํ•œ ํšจ์œจ์„ฑ์„ ์–ด๋–ป๊ฒŒ ๋‹ฌ์„ฑํ•  ์ˆ˜ ์žˆ๋Š”๊ฐ€? ๋™๊ธฐ : ํƒ์ƒ‰ ๋น„์šฉ๊ณผ ์ •๋ณด ํš๋“ ์‚ฌ์ด์˜ ํŠธ๋ ˆ์ด๋“œ์˜คํ”„๋Š” ์‚ฌํšŒ๋™๋ฌผยท์ธ๊ฐ„ ์ง‘๋‹จ ๋ชจ๋‘์—๊ฒŒ ์ค‘์š”ํ•œ ๋ฌธ์ œ์ด๋ฉฐ, ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ ๋ฆฌ๋”์‹ญ ํ˜น์€ ๊ณ ์ •๋œ โ€˜์Šค์นด์šฐํŠธโ€™ ์—ญํ• ์— ์ดˆ์ ์„ ๋งž์ท„๋‹ค. 2. ๋ชจ๋ธ ์„ค๊ณ„ | ์š”์†Œ | ์„ค๋ช… | | | | | ์—์ด์ „ํŠธ | N๋ช…์˜ ๊ฐœ์ฒด, ๋งค ์‹œ์  ํƒ์ƒ‰(์™ธ์ถœ) ํ˜น์€ ๋Œ€๊ธฐ ์„ ํƒ | | ํ™˜๊ฒฝ | ๊ณ ๋ณด์ƒยท์ €๋ณด์ƒ ๋‘ ์ƒํƒœ๊ฐ€ ๋งˆ์ฝ”ํ”„ ์ „์ด(๊ธฐ์–ต ์—†์Œ)๋กœ ์ „ํ™˜ | | ๋ณด์ƒ ๊ตฌ์กฐ | ์„ฑ๊ณต ํ™•๋ฅ  ฮณยฑ์— ๋”ฐ๋ผ ๋ณด์ƒ,

Quantitative Biology
No Image

Decision Quality Evaluation Framework at Pinterest

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํ”Œ๋žซํผ ๊ฑฐ๋ฒ„๋„Œ์Šค์˜ ๋ณตํ•ฉ์„ฑ : ์ •์ฑ… ํ•ด์„์˜ ๋ชจํ˜ธ์„ฑ, ํฌ๊ท€ ์ฝ˜ํ…์ธ , ๋ผ๋ฒจ๋ง ๋น„์šฉ ๋“ฑ์€ โ€œ์ง„์‹ค์˜ ํ”ผ๋ผ๋ฏธ๋“œ(Pyramid of Truth)โ€๋ผ๋Š” ๊ฐœ๋…์œผ๋กœ ์ž˜ ์ •๋ฆฌ๋œ๋‹ค. ์ธ๊ฐ„โ€‘LLM ํ˜‘์—… : ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ธ๊ฐ„โ€‘๊ธฐ๊ณ„ ํ˜‘์—…์˜ ์‚ฌํšŒยท๊ธฐ์ˆ ์  ์ด์Šˆ๋ฅผ ๋‹ค๋ฃจ์ง€๋งŒ, ์‹ค์ œ ์šด์˜์—์„œ ํ’ˆ์งˆ ํ‰๊ฐ€ ์ฒด๊ณ„ ๊ฐ€ ๋ถ€์žฌํ•จ์„ ์ง€์ ํ•œ๋‹ค. ๋ณธ ๋…ผ๋ฌธ์€ ์ด ๊ฒฉ์ฐจ๋ฅผ ๋ฉ”์šฐ๋Š” ์‹ค์šฉ์  ์†”๋ฃจ์…˜์„ ์ œ์‹œํ•œ๋‹ค. 2. ํ•ต์‹ฌ ๊ตฌ์„ฑ ์š”์†Œ | ๊ตฌ์„ฑ ์š”์†Œ | ์—ญํ•  | ์ฃผ์š” ๊ธฐ๋ฒ• | | | | | | GDS (Golden Data Set) | ๊ณ ์‹ ๋ขฐ โ€˜groundโ€‘truthโ€™

Statistics Applications Framework
No Image

Deep description of static and dynamic network ties in Honduran villages

1. ์—ฐ๊ตฌ์˜ ์˜์˜์™€ ๋ฐฐ๊ฒฝ ๋ฉ€ํ‹ฐํ”Œ๋ ‰์Šค ๋„คํŠธ์›Œํฌ ์ ‘๊ทผ : ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ์ฃผ๋กœ ๋‹จ์ผ์ธต(์นœ๊ตฌยท๊ฒฝ์ œยท๋ณด๊ฑด ๋“ฑ) ๋„คํŠธ์›Œํฌ์— ๋จธ๋ฌผ๋ €์œผ๋‚˜, ๋ณธ ๋…ผ๋ฌธ์€ ๋„ค ๊ฐ€์ง€ ๊ด€๊ณ„๋ฅผ ๋™์‹œ์— ๊ณ ๋ คํ•ด ํ˜‘๋ ฅยท๊ฐˆ๋“ฑ์ด ์–ด๋–ป๊ฒŒ ์ƒํ˜ธ์ž‘์šฉํ•˜๋Š”์ง€ ํƒ๊ตฌํ•œ๋‹ค. ์ด๋Š” ํŠนํžˆ ์ž์›์ด ์ œํ•œ๋œ ๋†์ดŒ์—์„œ โ€œ์—ฐ๊ฒฐ์€ ๋งŽ์ง€๋งŒ ๋ถ„๋ฆฌ๋œโ€ ๊ตฌ์กฐ๋ฅผ ์ดํ•ดํ•˜๋Š” ๋ฐ ํ•„์ˆ˜์ ์ด๋‹ค. ์‹œ๊ฐ„์  ๋ณ€ํ™” : ๋‘ ์‹œ์ (2016, 2019)์„ ๋น„๊ตํ•จ์œผ๋กœ์จ ๋„คํŠธ์›Œํฌ ๋™ํƒœ๋ฅผ ํŒŒ์•…ํ•˜๊ณ , ์ •์ฑ…ยทํ”„๋กœ๊ทธ๋žจ ์„ค๊ณ„ ์‹œ โ€œ์‹œ์ ์— ๋”ฐ๋ฅธ ๋„คํŠธ์›Œํฌ ๋ณ€๋™์„ฑโ€์„ ๋ฐ˜์˜ํ•  ์ˆ˜ ์žˆ๋‹ค. 2. ๋ฐ์ดํ„ฐ์™€ ์ธก์ • | ํ•ญ๋ชฉ | ๋‚ด์šฉ | | | | | ํ‘œ๋ณธ | 176๊ฐœ

Statistics Applications Network
No Image

Deep Learning for Point Spread Function Modeling in Cosmology

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ PSF์˜ ์—ญํ•  : ์•ฝํ•œ ์ค‘๋ ฅ ๋ Œ์ฆˆ๋ง, ํŠนํžˆ ์ฝ”์Šค๋ฏน ์‹œ์–ด(cosmic shear) ์ธก์ •์€ PSF ๋ณด์ •์— ๊ทน๋„๋กœ ๋ฏผ๊ฐํ•˜๋‹ค. PSF ์˜ค์ฐจ๋Š” ์ง์ ‘์ ์œผ๋กœ ์€ํ•˜ ํ˜•ํƒœ์˜ ์™œ๊ณก์„ ์™œ๊ณก์‹œ์ผœ ์šฐ์ฃผ ํŒฝ์ฐฝ ๋ฐ ์•”ํ‘ ๋ฌผ์งˆยท์—๋„ˆ์ง€ ํŒŒ๋ผ๋ฏธํ„ฐ ์ถ”์ •์— ํŽธํ–ฅ์„ ์ดˆ๋ž˜ํ•œ๋‹ค. PIFF์˜ ํ•œ๊ณ„ : PIFF๋Š” CCD๋ณ„ ๋…๋ฆฝ ๋ชจ๋ธ๋ง์„ ๊ธฐ๋ณธ์œผ๋กœ ํ•˜์—ฌ ์ „์ฒด ์ดˆ์ ๋ฉด์— ๊ฑธ์นœ ์—ฐ์†์„ฑ์„ ์†์‹คํ•œ๋‹ค. ๋Œ€ํ˜• ๊ด‘ํ•™๊ณ„(์˜ˆ: LSST, Subaru/HSC)๋Š” ์ˆ˜๋ฐฑ ๊ฐœ์˜ CCD๊ฐ€ ๋ชจ์ž์ดํฌ ํ˜•ํƒœ๋กœ ๋ฐฐ์น˜๋ผ ์žˆ์–ด, ์ „์—ญ์ ์ธ ๊ด‘ํ•™ ๋ณ€ํ˜•(์˜ˆ: ๊ด‘ํ•™ ์ˆ˜์ฐจ, ๋Œ€๊ธฐ ๋ณ€๋™)์„ ํฌ์ฐฉํ•˜๊ธฐ

Learning Astrophysics Model
No Image

Defining causal mechanism in dual process theory and two types of feedback control

1. ์ด๋ก ์  ๊ธฐ์—ฌ | ๊ตฌ๋ถ„ | ๊ธฐ์กด ์—ฐ๊ตฌ | ๋ณธ ๋…ผ๋ฌธ์˜ ์ƒˆ๋กœ์›€ | | | | | | ์ธ๊ณผ ๋ฉ”์ปค๋‹ˆ์ฆ˜ | ์ •์‹ โ€‘๋ฌผ๋ฆฌ ์ƒ์œ„โ€‘ํ•˜์œ„ ๊ด€๊ณ„๋Š” ์ฃผ๋กœ ์„ค๋ช…์  (supervenience) ์ˆ˜์ค€์— ๋จธ๋ฌผ๋ €์œผ๋ฉฐ, ์‹ค์ œ ์ธ๊ณผ ์ „๋‹ฌ ๋ฉ”์ปค๋‹ˆ์ฆ˜์€ ์ œ์‹œ๋˜์ง€ ์•Š์Œ. | ์ƒ์œ„ ์ˆ˜์ค€์˜ ๋ฐฉ์ •์‹ ์„ ํƒ (์ธ๋ฑ์Šค ์‹œํ€€์Šค) โ†” ํ•˜์œ„ ์ˆ˜์ค€์˜ ์‹ ๊ฒฝยท์‹œ๋ƒ…์Šค ์กฐ์ • ์ด๋ผ๋Š” ๋‘ ๋…๋ฆฝ์ ์ธ ํ”ผ๋“œ๋ฐฑ ๋ฃจํ”„๋ฅผ ๋ช…์‹œ์ ์œผ๋กœ ๋ชจ๋ธ๋ง, โ€œsupervenient causeโ€์™€ โ€œcausal transmission mechanismโ€์„ ์ˆ˜ํ•™์ ์œผ๋กœ ์ •์˜. | | ์ด์ค‘ ๊ณผ์ • ํ†ตํ•ฉ | Type 1ยทType 2๋ฅผ

Quantitative Biology
No Image

Design and Analysis Strategies for Pooling in High Throughput Screening: Application to the Search for a New Anti-Microbial

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ํ•ญ์ƒ์ œ ๋‚ด์„ฑ ์ด 2050๋…„๊นŒ์ง€ ์—ฐ๊ฐ„ 1,000๋งŒ ๋ช… ์‚ฌ๋ง์„ ์ดˆ๋ž˜ํ•œ๋‹ค๋Š” UN ์ „๋ง์€ ์ƒˆ๋กœ์šด ์•ฝ๋ฌผ ํƒ์ƒ‰์˜ ๊ธด๊ธ‰์„ฑ์„ ๊ฐ•์กฐํ•œ๋‹ค. ๊ธฐ์กด HTS์˜ OCOW ๋ฐฉ์‹์€ ํ™”ํ•ฉ๋ฌผ๋‹น 1ํšŒ๋งŒ ๊ด€์ฐฐ๋˜๋ฏ€๋กœ, ํฌ์†Œํ•œ ์ง„์งœ ํžˆํŠธ๋ฅผ ์ฐพ์„ ๋•Œ ํ—ˆ์œ„ ์–‘์„ฑ๋ฅ  ์ด ์‹ค์งˆ์ ์ธ ๋น„์šฉ ํญ์ฆ์„ ์ดˆ๋ž˜ํ•œ๋‹ค. ํ’€๋ง์€ ํ†ต๊ณ„์  ํšจ์œจ์„ฑ (ํ•œ ํ™”ํ•ฉ๋ฌผ์„ ์—ฌ๋Ÿฌ ์›ฐ์— ์ค‘๋ณต ๋ฐฐ์น˜)๊ณผ ์‹คํ—˜์  ํšจ์œจ์„ฑ (์›ฐ ์ˆ˜ ๋Œ€๋น„ ํ™”ํ•ฉ๋ฌผ ์ˆ˜ ์ฆ๊ฐ€)์„ ๋™์‹œ์— ์ œ๊ณตํ•œ๋‹ค๋Š” ์ ์—์„œ ๋งค๋ ฅ์ ์ด๋‹ค. 2. ํ’€๋ง ์„ค๊ณ„ ๋ฐฉ๋ฒ• ๋น„๊ต | ์„ค๊ณ„ | ํ•ต์‹ฌ ์•„์ด๋””์–ด | ํ’€ ํฌ๊ธฐ(c) ์ œ์•ฝ | ํ™”ํ•ฉ๋ฌผ ์žฌํ˜„ ํšŸ์ˆ˜(a) |

Statistics Applications Analysis
No Image

Design Support for Multitape Turing Machines

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฉ€ํ‹ฐํ…Œ์ดํ”„ ํŠœ๋ง ๋จธ์‹ ์˜ ๊ต์œก์  ๋‚œ์ด๋„ : ์ „ํ†ต์ ์ธ ๋‹จ์ผ ํ…Œ์ดํ”„ ๋ชจ๋ธ์— ๋น„ํ•ด ์ƒํƒœ ์ „์ด์™€ ํ—ค๋“œ ์ด๋™์ด ๋ณตํ•ฉ์ ์œผ๋กœ ์–ฝํ˜€ ์žˆ์–ด, ํ•™์ƒ๋“ค์ด ์ง๊ด€์ ์œผ๋กœ ์ดํ•ดํ•˜๊ธฐ ์–ด๋ ต๋‹ค. ๊ธฐ์กด ์ง€์› ํ•œ๊ณ„ : FSM ์–ธ์–ด๊ฐ€ ์ •์˜ยท์‹คํ–‰์„ ์ง€์›ํ•˜์ง€๋งŒ, ์™œ ํŠน์ • ์ž…๋ ฅ์ด ์ˆ˜์šฉยท๊ฑฐ๋ถ€๋˜๋Š”์ง€์— ๋Œ€ํ•œ ๋ฉ”ํƒ€ ์ •๋ณด๋ฅผ ์ œ๊ณตํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์ด๋Š” ํ•™์ƒ๋“ค์˜ โ€˜์šด์˜ ์˜๋ฏธ๋ก (operational semantics)โ€™์— ๋Œ€ํ•œ ์ธ์‚ฌ์ดํŠธ ๋ถ€์กฑ์œผ๋กœ ์ด์–ด์ง„๋‹ค. 2. ์ œ์•ˆ๋œ ์‹œ๊ฐํ™” ๋„๊ตฌ | ๋„๊ตฌ | ์œ ํ˜• | ํ•ต์‹ฌ ๊ธฐ๋Šฅ | ๊ต์œก์  ํšจ๊ณผ | | | | | | | Dynamic

No Image

Detecting Silent Failures in Multi-Agentic AI Trajectories

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ๋ฉ€ํ‹ฐ์—์ด์ „ํŠธ AI ๋Š” ์—ฌ๋Ÿฌ LLM์ด ํ˜‘์—…ยท๊ฒฝ์Ÿํ•˜๋ฉฐ ๋ณตํ•ฉ์ ์ธ ์ž‘์—…์„ ์ˆ˜ํ–‰ํ•œ๋‹ค. ๋น„๊ฒฐ์ •์„ฑ ๋•Œ๋ฌธ์— ๋™์ผ ์ž…๋ ฅ์— ๋Œ€ํ•ด ์„œ๋กœ ๋‹ค๋ฅธ ๊ถค์ ์ด ์ƒ์„ฑ๋  ์ˆ˜ ์žˆ๋‹ค. ๊ธฐ์กด LLM ํ‰๊ฐ€ ๋Š” ์ •๋‹ต ์ผ์น˜์œจ, BLEU, ROUGE ๋“ฑ ์ •๋Ÿ‰์  ์ง€ํ‘œ์— ์˜์กดํ•˜์ง€๋งŒ, ์นจ๋ฌต ์˜ค๋ฅ˜ (์˜ˆ: ๋ชฉํ‘œ์—์„œ ์ ์  ๋ฉ€์–ด์ง€๋Š” drift, ๋ฌดํ•œ ๋ฃจํ”„, ์ค‘์š”ํ•œ ์ •๋ณด ๋ˆ„๋ฝ)๋Š” ์ด๋Ÿฌํ•œ ์ง€ํ‘œ์— ์ž˜ ๋“œ๋Ÿฌ๋‚˜์ง€ ์•Š๋Š”๋‹ค. ์‹ค๋ฌด์—์„œ๋Š” ์‚ฌ์šฉ์ž ๊ฒฝํ—˜ ์ €ํ•˜ ์™€ ์‹œ์Šคํ…œ ์‹ ๋ขฐ์„ฑ ์†์‹ค ์„ ์ดˆ๋ž˜ํ•˜๋ฏ€๋กœ, ์ž๋™ํ™”๋œ ๋น„์ •์ƒ ํƒ์ง€ ๋ฉ”์ปค๋‹ˆ์ฆ˜์ด ์ ˆ์‹คํžˆ ํ•„์š”ํ•˜๋‹ค. 2. ์ฃผ์š” ๊ธฐ์—ฌ | ๋ฒˆํ˜ธ | ๋‚ด์šฉ

No Image

DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Models

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋™๊ธฐ ์—ญยท์ •๋ฐฉํ–ฅ ๋ Œ๋”๋ง์˜ ํ†ตํ•ฉ ํ•„์š”์„ฑ : ๊ธฐ์กด ์—ฐ๊ตฌ๋Š” ๋ณดํ†ต ์—ญ(์ด๋ฏธ์ง€ โ†’ ์žฌ์งˆยท์กฐ๋ช…) ํ˜น์€ ์ •(์žฌ์งˆยท์กฐ๋ช… โ†’ ์ด๋ฏธ์ง€) ์ค‘ ํ•˜๋‚˜์— ์ง‘์ค‘ํ•œ๋‹ค. ์‹ค์ œ ์‘์šฉ์—์„œ๋Š” ๋‘ ๊ณผ์ •์ด ์ˆœํ™˜์ ์œผ๋กœ ์‚ฌ์šฉ๋˜๋ฏ€๋กœ ํ•˜๋‚˜์˜ ๋ชจ๋ธ์ด ์–‘์ชฝ์„ ๋ชจ๋‘ ๋‹ค๋ฃจ๋ฉด ํŒŒ์ดํ”„๋ผ์ธ์ด ๋‹จ์ˆœํ™”๋˜๊ณ  ์˜ค๋ฅ˜ ์ „ํŒŒ๊ฐ€ ๊ฐ์†Œํ•œ๋‹ค. ์ „ํ†ต PBR์˜ ํ•œ๊ณ„ : ์ •ํ™•ํ•œ 3D ์”ฌ ์žฌ๊ตฌ์„ฑ์ด ์–ด๋ ค์šด ์‹ค์„ธ๊ณ„ ์˜์ƒ์— ๋Œ€ํ•ด ๋ฌผ๋ฆฌ ๊ธฐ๋ฐ˜ ์‹œ๋ฎฌ๋ ˆ์ด์…˜์„ ์ ์šฉํ•˜๊ธฐ์—” ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘ยท์ „์ฒ˜๋ฆฌ ๋น„์šฉ์ด ๊ณผ๋‹คํ•˜๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ๋น„๋””์˜ค ํ™•์‚ฐ ๋ชจ๋ธ ํ”„๋ผ์ด์–ด ํ™œ์šฉ : ๋Œ€๊ทœ๋ชจ ๋น„๋””์˜ค ๋ฐ์ดํ„ฐ๋กœ ์‚ฌ์ „ ํ•™์Šต๋œ ํ™•์‚ฐ ๋ชจ๋ธ์€

Model
No Image

Discourse-Aware Scientific Paper Recommendation via QA-Style Summarization and Multi-Level Contrastive Learning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ๋ฌธ์ œ ์ •์˜ ์˜คํ”ˆ ์•ก์„ธ์Šค ๊ธ‰์ฆ ๊ณผ ํ”„๋ผ์ด๋ฒ„์‹œ ์ œ์•ฝ ์œผ๋กœ ์‚ฌ์šฉ์žโ€‘๊ธฐ๋ฐ˜ ํ˜‘์—… ํ•„ํ„ฐ๋ง์ด ์–ด๋ ค์›Œ์กŒ์œผ๋ฉฐ, ํ…์ŠคํŠธ ๊ธฐ๋ฐ˜ ์ฝ˜ํ…์ธ  ์ถ”์ฒœ์ด ์ฃผ๋ฅ˜๊ฐ€ ๋˜์—ˆ๋‹ค. ๊ธฐ์กด ์ฝ˜ํ…์ธ  ๊ธฐ๋ฐ˜ ๋ชจ๋ธ์€ ๋ฌธ์„œ ์ „์ฒด๋ฅผ ํ‰๋ฉด ํ…์ŠคํŠธ ๋กœ ์ฒ˜๋ฆฌํ•ด, ๋…ผ๋ฌธ์˜ ๋‹ดํ™” ๊ตฌ์กฐ(OMRC) ๋ฅผ ํ™œ์šฉํ•˜์ง€ ๋ชปํ•œ๋‹ค. ์ด๋Š” (a) ํ•ต์‹ฌ ์ •๋ณด ์†์‹ค, (b) ์˜๋ฏธ์  ๋ถˆ์™„์ „์„ฑ, (c) ๊ฒฐ๊ณผ ํ•ด์„ ์–ด๋ ค์›€์ด๋ผ๋Š” ์„ธ ๊ฐ€์ง€ ์ฃผ์š” ํ•œ๊ณ„๋ฅผ ๋งŒ๋“ ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ๋ฐ ๊ธฐ๋ฒ• | ๊ตฌ์„ฑ ์š”์†Œ | ์—ญํ•  | ์ฃผ์š” ๊ธฐ์ˆ  | | | | | | QAโ€‘Style OMRC Summarization | ๋…ผ๋ฌธ์„ ๋ชฉ

Learning
No Image

Discovering Unknown Inverter Governing Equations via Physics-Informed Sparse Machine Learning

1. ์—ฐ๊ตฌ ๋ฐฐ๊ฒฝ ๋ฐ ํ•„์š”์„ฑ ์ธ๋ฒ„ํ„ฐ ๊ธฐ๋ฐ˜ ์ž์›(IBR)์˜ ๊ธ‰์ฆ ์œผ๋กœ ์ „๋ ฅ๊ณ„ํ†ต์ด ์ „์ž์ œ์–ด ์ค‘์‹ฌ์œผ๋กœ ๋ณ€๋ชจํ•˜๊ณ  ์žˆ๋‹ค. ์ œ์กฐ์‚ฌโ€‘์šด์˜์ž ๊ฐ„ ์ •๋ณด ๋น„๋Œ€์นญ (์ œ์–ด ๋กœ์ง ๋น„๊ณต๊ฐœ) ๋•Œ๋ฌธ์— ์ „ํ†ต์ ์ธ ์ƒํƒœโ€‘๊ณต๊ฐ„ ๋ชจ๋ธ๋ง์ด ๋ถˆ๊ฐ€๋Šฅํ•ด์กŒ๋‹ค. ๊ธฐ์กด ์ž„ํ”ผ๋˜์Šค ์‹๋ณ„ ์€ ๊ตญ์†Œ ์„ ํ˜• ๊ฐ€์ •์— ๋จธ๋ฌผ๋Ÿฌ, ๋Œ€โ€‘์‹ ํ˜ธ(์˜ˆ: ๊ณ ์žฅ, ๊ธ‰๋ณ€) ๋™์ž‘์„ ์„ค๋ช…ํ•˜์ง€ ๋ชปํ•œ๋‹ค. 2. ํ•ต์‹ฌ ์•„์ด๋””์–ด ๋ฐ ๊ตฌ์„ฑ ์š”์†Œ | ๊ตฌ์„ฑ ์š”์†Œ | ์—ญํ•  | ์ฃผ์š” ํŠน์ง• | | | | | | ํฌ์†Œ ์‹ฌ๋ณผ๋ฆญ ๋ฐฑ๋ณธ (Sparse Symbolic Backbone) | ๋ฌผ๋ฆฌ์ ์œผ๋กœ ์˜๋ฏธ ์žˆ๋Š” ๊ธฐ๋ณธ ํ•ญ(์˜ˆ: Kirchhoff ๋ฒ•์น™

Learning Electrical Engineering and Systems Science

< 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