Efficient gradient computation for dynamical models.
Efficient gradient computation for dynamical models.
复制标题
动态模型的有效梯度计算。
DOI:
10.1016/j.neuroimage.2014.04.040
复制
发表时间:
2014-09
期刊:
影响因子:
5.7
通讯作者:
Penny WD
中科院分区:
文献类型:
--
作者:
Sengupta B;Friston KJ;Penny WD
Data assimilation is a fundamental issue that arises across many scales in neuroscience — ranging from the study of single neurons using single electrode recordings to the interaction of thousands of neurons using fMRI. Data assimilation involves inverting a generative model that can not only explain observed data but also generate predictions. Typically, the model is inverted or fitted using conventional tools of (convex) optimization that invariably extremise some functional — norms, minimum descriptive length, variational free energy, etc. Generally, optimisation rests on evaluating the local gradients of the functional to be optimized. In this paper, we compare three different gradient estimation techniques that could be used for extremising any functional in time — (i) finite differences, (ii) forward sensitivities and a method based on (iii) the adjoint of the dynamical system. We demonstrate that the first-order gradients of a dynamical system, linear or non-linear, can be computed most efficiently using the adjoint method. This is particularly true for systems where the number of parameters is greater than the number of states. For such systems, integrating several sensitivity equations – as required with forward sensitivities – proves to be most expensive, while finite-difference approximations have an intermediate efficiency. In the context of neuroimaging, adjoint based inversion of dynamical causal models (DCMs) can, in principle, enable the study of models with large numbers of nodes and parameters. We compare three methods to compute gradients in dynamical systems. The methods are finite-differences, forward sensitivity and reverse adjoints. The adjoint method is 50–70 folds more efficient.
登录
查看更多内容
影响因子:
5.7
作者:
Friston, Karl J.;Mattout, Jeremie;Penny, Will
通讯作者:
Penny, Will
影响因子:
5.8
作者:
Vyshemirsky, Vladislav;Girolami, Mark A.
通讯作者:
Girolami, Mark A.
影响因子:
4.3
作者:
Deco, Gustavo;Jirsa, Viktor K.;Robinson, Peter A.;Breakspear, Michael;Friston, Karl J.
通讯作者:
Friston, Karl J.
影响因子:
19.9
作者:
Gazzaniga, Michael S.
通讯作者:
Gazzaniga, Michael S.
影响因子:
5.7
作者:
David, Olivier;Kiebel, Stefan J.;Friston, Karl J.
通讯作者:
Friston, Karl J.