Change-point detection in anomalous-diffusion trajectories utilising machine-learning-based uncertainty estimates

被引:4
作者
Seckler, Henrik [1 ]
Metzler, Ralf [1 ,2 ]
机构
[1] Univ Potsdam, Inst Phys & Astron, D-14476 Potsdam Golm, Germany
[2] Asia Pacific Ctr Theoret Phys, Pohang 37673, South Korea
来源
JOURNAL OF PHYSICS-PHOTONICS | 2024年 / 6卷 / 04期
关键词
diffusion; anomalous diffusion; change-point analysis; AnDi; 2; challenge; Bayesian-deep learning analysis; RANDOM-WALK; PLASMA-MEMBRANE; MODELS;
D O I
10.1088/2515-7647/ad884c
中图分类号
O43 [光学];
学科分类号
070207 ; 0803 ;
摘要
When recording the movement of individual animals, cells or molecules one will often observe changes in their diffusive behaviour at certain points in time along their trajectory. In order to capture the different diffusive modes assembled in such heterogeneous trajectories it becomes necessary to segment them by determining these change-points. Such a change-point detection can be challenging for conventional statistical methods, especially when the changes are subtle. We here apply Bayesian Deep Learning to obtain point-wise estimates of not only the anomalous diffusion exponent but also the uncertainties in these predictions from a single anomalous diffusion trajectory generated according to four theoretical models of anomalous diffusion. We show that we are able to achieve an accuracy similar to single-mode (without change-points) predictions as well as a well calibrated uncertainty predictions of this accuracy. Additionally, we find that the predicted uncertainties feature interesting behaviour at the change-points leading us to examine the capabilities of these predictions for change-point detection. While the series of predicted uncertainties on their own are not sufficient to improve change-point detection, they do lead to a performance boost when applied in combination with the predicted anomalous diffusion exponents.
引用
收藏
页数:16
相关论文
共 122 条
[1]   Moses, Noah and Joseph effects in Levy walks [J].
Aghion, Erez ;
Meyer, Philipp G. ;
Adlakha, Vidushi ;
Kantz, Holger ;
Bassler, Kevin E. .
NEW JOURNAL OF PHYSICS, 2021, 23 (02)
[2]   Classification of stochastic processes by convolutional neural networks [J].
AL-hada, Eman A. ;
Tang, Xiangong ;
Deng, Weihua .
JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL, 2022, 55 (27)
[3]  
[Anonymous], 1981, Stochastic Processes in Physics and Chemistry
[4]  
[Anonymous], 1995, Random walks
[5]   Temporal Analysis of Active and Passive Transport in Living Cells [J].
Arcizet, Delphine ;
Meier, Boern ;
Sackmann, Erich ;
Raedler, Joachim O. ;
Heinrich, Doris .
PHYSICAL REVIEW LETTERS, 2008, 101 (24)
[6]   Classification, inference and segmentation of anomalous diffusion with recurrent neural networks [J].
Argun, Aykut ;
Volpe, Giovanni ;
Bo, Stefano .
JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL, 2021, 54 (29)
[7]   Plasma Membrane Organization of Epidermal Growth Factor Receptor in Resting and Ligand-Bound States [J].
Bag, Nirmalya ;
Huang, Shuangru ;
Wohland, Thorsten .
BIOPHYSICAL JOURNAL, 2015, 109 (09) :1925-1936
[8]   STRANGE KINETICS of single molecules in living cells [J].
Barkai, Eli ;
Garini, Yuval ;
Metzler, Ralf .
PHYSICS TODAY, 2012, 65 (08) :29-35
[9]   Animal search strategies: A quantitative. random-walk analysis [J].
Bartumeus, F ;
Da Luz, MGE ;
Viswanathan, GM ;
Catalan, J .
ECOLOGY, 2005, 86 (11) :3078-3087
[10]   Measurement of anomalous diffusion using recurrent neural networks [J].
Bo, Stefano ;
Schmidt, Falko ;
Eichhorn, Ralf ;
Volpe, Giovanni .
PHYSICAL REVIEW E, 2019, 100 (01)