Probability-one homotopy maps for tracking constrained clustering solutions
نویسندگان
چکیده
Modern machine learning problems typically have multiple criteria, but there is currently no systematic mathematical theory to guide the design of formulations and exploration of alternatives. Homotopy methods are a promising approach to characterize solution spaces by smoothly tracking solutions from one formulation (typically an “easy” problem) to another (typically a “hard” problem). New results in constructing homotopy maps for constrained clustering problems are here presented, which combine quadratic loss functions with discrete evaluations of constraint violations are presented. These maps balance requirements of locality in clusters as well as those of discrete must-link and must-notlink constraints. Experimental results demonstrate advantages in tracking solutions compared to stateof-the-art constrained clustering algorithms.
منابع مشابه
Probability-one Homotopy Maps for Constrained Clustering Problems
Many algorithms for constrained clustering have been developed in the literature that aim to balance vector quantization requirements of cluster prototypes against the discrete satisfaction requirements of constraint (must-link or cannot-link) sets. A significant amount of research has been devoted to designing new algorithms for constrained clustering and understanding when constraints help cl...
متن کاملStudying Constrained Clustering Problems Using Homotopy Maps
Many algorithms for constrained clustering have been developed in the literature that aim to balance vector quantization requirements of cluster prototypes against the discrete satisfaction requirements of constraint (must-link or must-not-link) sets. Significant research has been devoted to designing new algorithms for constrained clustering and understanding when constraints help clustering. ...
متن کاملTheory of Globally Convergent Probability-One Homotopies for Nonlinear Programming
For many years globally convergent probability-one homotopy methods have been remarkably successful on difficult realistic engineering optimization problems, most of which were attacked by homotopy methods because other optimization algorithms failed or were ineffective. Convergence theory has been derived for a few particular problems, and considerable fixed point theory exists, but generally ...
متن کاملProbability-one homotopy algorithms for solving the coupled Lyapunov equations arising in reduced-order H2/H∞ modeling, estimation, and control
Optimal reduced-order modeling, estimation, and control with respect to combined H 2 =H I criteria give rise to coupled Lyapunov and Riccati equations. To develop reliable numerical algorithms for these problems this paper focuses on the coupled Lyapunov equations which appear as a subset of the synthesis equations. In particular, this paper systematically examines the requirements of probabili...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2013