Symmetry-Breaking Bifurcations of the Information Bottleneck and Related Problems

被引:1
|
作者
Parker, Albert E. [1 ]
Dimitrov, Alexander G. [2 ]
机构
[1] Montana State Univ, Ctr Biofilm Engn, Dept Math Sci, Bozeman, MT 59717 USA
[2] Washington State Univ, Dept Math & Stat, Vancouver, WA 98686 USA
关键词
information bottleneck; optimization; annealing; gradient flow; bifurcations; symmetry; CLASSIFICATION;
D O I
10.3390/e24091231
中图分类号
O4 [物理学];
学科分类号
0702 ;
摘要
In this paper, we investigate the bifurcations of solutions to a class of degenerate constrained optimization problems. This study was motivated by the Information Bottleneck and Information Distortion problems, which have been used to successfully cluster data in many different applications. In the problems we discuss in this paper, the distortion function is not a linear function of the quantizer. This leads to a challenging annealing optimization problem, which we recast as a fixed-point dynamics problem of a gradient flow of a related dynamical system. The gradient system possesses an S-N symmetry due to its invariance in relabeling representative classes. Its flow hence passes through a series of bifurcations with specific symmetry breaks. Here, we show that the dynamical system related to the Information Bottleneck problem has an additional spurious symmetry that requires more-challenging analysis of the symmetry-breaking bifurcation. For the Information Bottleneck, we determine that when bifurcations occur, they are only of pitchfork type, and we give conditions that determine the stability of the bifurcating branches. We relate the existence of subcritical bifurcations to the existence of first-order phase transitions in the corresponding distortion function as a function of the annealing parameter, and provide criteria with which to detect such transitions.
引用
收藏
页数:20
相关论文
共 50 条