곰곰의 일지 곰곰의 일지

전체 글

• [Stat][검정] 순열 검정(Permutation test)

  Paper Review(논문이야기)/관련 개념 정리 2025. 3. 13. 20:12

  비모수 검정 관련 방법들을 보다가 Below Permutation null ~, 이라는 말이 너무 많이 나와서 한번 정리해본다. 본 내용은 https://angeloyeo.github.io/2021/10/28/permutation_test.html 공돌이의 수학노트의 내용을 기반으로 작성되었습니다. 대부분의 그림자료 또한 해당 블로그의 자료임을 밝힙니다.  우선 Permutation test에 관해 알아보자.  Permutaion test순열검정법(permuation test)은 두 개 이상의 표본을 함께 결합하여 관측값들을 무작위로 재표본으로 추출하고 이를 이용하여 가설 검정을 진행하는 방법을 말한다. 이때 재표본 추출(resampling)은 비복원 방식으로 진행한다....? 조금 더 알아보자.  두 ..

  Read More
• [Two sample test][Graph] Collaborative non-parametric two-sample testing(by Alejandro de la Concha, Nicolas Vayatis, Argyris Kalogeratos)

  Paper Review(논문이야기) 2025. 3. 10. 15:54

  https://arxiv.org/abs/2402.05715 Collaborative non-parametric two-sample testingThis paper addresses the multiple two-sample test problem in a graph-structured setting, which is a common scenario in fields such as Spatial Statistics and Neuroscience. Each node $v$ in fixed graph deals with a two-sample testing problem between two nodearxiv.orgAbstract Graph 구조에서의 multiple two sample test에 관해서 다룬..

  Read More
• [CPD][AUC] Model-free Change-point Detection Using ModernClassifiers(by Rohit Kanrar, Feiyu Jiang, and Zhanrui Cai)

  Paper Review(논문이야기) 2025. 3. 7. 18:07

  https://arxiv.org/abs/2404.06995 Model-free Change-point Detection Using Modern ClassifiersIn contemporary data analysis, it is increasingly common to work with non-stationary complex datasets. These datasets typically extend beyond the classical low-dimensional Euclidean space, making it challenging to detect shifts in their distribution withouarxiv.orgKeypointAUC(Area Under the Curve) 를 변화점 탐지..

  Read More
• [Out of Distribution Detection(CPD)][Parameter] Neural Mean Discrepancy for Efficient Out-of-Distribution Detection (CVPR 2022)

  Paper Review(논문이야기) 2025. 3. 6. 13:14

  https://arxiv.org/abs/2104.11408 Neural Mean Discrepancy for Efficient Out-of-Distribution DetectionVarious approaches have been proposed for out-of-distribution (OOD) detection by augmenting models, input examples, training sets, and optimization objectives. Deviating from existing work, we have a simple hypothesis that standard off-the-shelf models mayarxiv.orgKeypointNMD가 Integral probability..

  Read More
• [Change point detection][Clustering] Unsupervised Change Point Detection in Multivariate Time Series (Wu, D., Gundimeda, S., Mou, S. & Quinn, C. AISTATS 2024)

  Paper Review(논문이야기) 2025. 3. 4. 10:52

  https://proceedings.mlr.press/v238/wu24g.html Unsupervised Change Point Detection in Multivariate Time SeriesWe consider the challenging problem of unsupervised change point detection in multivariate time series when the number of change points is unknown. Our method eliminates the user’s need for careful...proceedings.mlr.press IntroductionNon-Stationary를 모델링하는 것은 다양한 분야에 사용되나 일반적으로 어려운 문제이다. 대..

  Read More
• [Time series][CPD] Neural network-based CUSUM for online change-point detection (by Tingnan Gong, Junghwan Lee, Xiuyuan Cheng, and Yao Xie)

  Paper Review(논문이야기) 2025. 2. 24. 22:02

  https://arxiv.org/abs/2210.17312 Neural network-based CUSUM for online change-point detectionChange-point detection, detecting an abrupt change in the data distribution from sequential data, is a fundamental problem in statistics and machine learning. CUSUM is a popular statistical method for online change-point detection due to its efficiency froarxiv.org이전 논문 https://par.nsf.gov/servlets/purl/..

  Read More
• [Finance][Time series] Unsupervised Change Point Detection and Trend Prediction for Financial Time-Series Using a New CUSUM-Based Approach(IEEE 2022)

  Paper Review(논문이야기) 2025. 2. 17. 16:42

  https://ieeexplore.ieee.org/document/9741807/ Unsupervised Change Point Detection and Trend Prediction for Financial Time-Series Using a New CUSUM-Based ApproachThe aim of this research is to propose a binary segmentation algorithm to detect the change points in financial time-series based on the Iterative Cumulative Sum of Squares (ICSS). The proposed algorithm, entitled KW-ICSS, utilizes the n..

  Read More
• [Time Series][Change point Detection] CUSUM method

  Paper Review(논문이야기)/관련 개념 정리 2025. 2. 9. 23:52

  CUSUM에 관한 내용을 최근 읽었던 논문에서 기초하고 있는 방법인만큼 해당 내용을 보다 상세히 작성하여보았다. 대부분은 https://zephyrus1111.tistory.com/400 에서 기초하여 작성함을 밝힌다   - 문제정의 CUSUM 알고리즘이 풀고자하는 문제는 시계열내에 급격한 변경점이 없다는 귀무가설(H0)과 하나의 변경점이 있다는 대립가설 Ha을 세우고 이 중 어떤것을 선택해야하는지에 관한 문제이다.  먼저 시계열 데이터 Xt가 있다고 하자(t는 1부터 k까지 존재), 각가은 IID 조건을 따르고 해당 확률분포의 확률밀도함수를 p(x)라 하자. Change point dection은 시계열 데이터에서 변화가 생겼을 경우에 해당하는 시점 tc를 찾는 알고리즘이기에 θ">θ(확률..

  Read More