【알고리즘】 27강. 기타 알고리즘

27강. 기타 알고리즘

 

추천글 : 【알고리즘】 알고리즘 목차

---

 

○ ADMM(alternating direction method of multipliers) algorithm

○ BRISQUE : 이미지 별 품질 평가 척도 제공

○ BTB

○ Creates n-grams and binary decision rules for feature tuples.

○ Selects predicates using L1-regularized logistic regression.

○ Predicates embedded into 1024-dimension vector and fed into neural network.

○ CHOLMOD
○ dense Cholesky factorization

○ 다익스트라(Dijkstra) 및 Floyd 알고리즘

○ fuzzy logic : 0, 1로 clear하게 구분되는 로직이 아니라 불확실성을 반영. fuzzy neural network는 global minimum에 도달할 가능성이 높음

○ greedy bin-packing algorithm
○ HTL(human-in-the-loop) : AI 시스템의 학습을 가이드하는 사람

○ ICM algorithm 

○ Levenberg-Marquardt (LM) algorithm

○ LSTM 

○ Embeddings combined using learned weighted averaging.

○ Reduced embedding fed into LSTM.

○ Extends recurrent neural entwork by adding three gates.

○ Minimizes log-loss function for predictions.

○ maxstat algorithm 

○ Nussinov algorithm : 핵산 구조 결정 알고리즘

○ SIFT registration : ST Visium-Xenium paired data에서 사용할 수 있는 registration 알고리즘 

○ sparse bundle adjustment (SBA)

○ stable diffusion : 자연어로부터 디지털 이미지를 생성하는 AI 알고리즘

○ stable marriage matching : 두 그룹 간의 안정적이고 preference 기반의 매칭을 찾는 알고리즘

○ structure from motion (SfM) : SIFT, vocabulary tree, bundle adjustment 등

○ TANN

○ Sequence of embeddings used with prior embdding

○ Attribution logits converted to weights using softmax

○ Weighted sum calculated with dimension d ranges

○ UPGMA algorithm 

○ Voronoi tessellation

 

입력: 2021.12.12 11:35

수정: 2024.07.17 09:32