paper-with-me

홈 › Papers

An Algorithmic Theory of Dependent Regularizers, Part 1: Submodular Structure

2013-12-06 · Hoyt Koepke, Marina Meila

We present an exploration of the rich theoretical connections between several classes of regularized models, network flows, and recent results in submodular function theory. This work unifies key aspects of these problems under a common theory, leading to novel methods for working with several important models of interest in statistics, machine learning and computer vision. In Part 1, we review the concepts of network flows and submodular function optimization theory foundational to our results. We then examine the connections between network flows and the minimum-norm algorithm from submodular optimization, extending and improving several current results. This leads to a concise representation of the structure of a large class of pairwise regularized models important in machine learning, statistics and computer vision. In Part 2, we describe the full regularization path of a class of penalized regression problems with dependent variables that includes the graph-guided LASSO and total variation constrained models. This description also motivates a practical algorithm. This allows us to efficiently find the regularization path of the discretized version of TV penalized models. Ultimately, our new algorithms scale up to high-dimensional problems with millions of variables.

📄 PDF Abstract BibTeX arXiv:1312.1970

Code (0)

등록된 구현이 없습니다.

Tasks

BIG-bench Machine Learning

Similar Papers 제목 키워드 기반

On the Convergence Rate of Decomposable Submodular Function Minimization

2014-06-25 · NeurIPS 2014 12 · Robert Nishihara, Stefanie Jegelka, Michael. I. Jordan

Submodular functions describe a variety of discrete problems in machine learning, signal processing, and computer vision. However, minimizing submodular functions poses a number of algorithmic challenges. Recent work int…

BIG-bench Machine Learning

Joint Continuous and Discrete Model Selection via Submodularity

2021-02-17 · Jonathan Bunton, Paulo Tabuada

In model selection problems for machine learning, the desire for a well-performing model with meaningful structure is typically expressed through a regularized optimization problem. In many scenarios, however, the meanin…

modelModel Selection

Minimax Optimization: The Case of Convex-Submodular

2021-11-01 · Arman Adibi, Aryan Mokhtari, Hamed Hassani

Minimax optimization has been central in addressing various applications in machine learning, game theory, and control theory. Prior literature has thus far mainly focused on studying such problems in the continuous doma…

Selecting Diverse Features via Spectral Regularization

2012-12-01 · NeurIPS 2012 12 · Abhimanyu Das, Anirban Dasgupta, Ravi Kumar

We study the problem of diverse feature selection in linear regression: selecting a small subset of diverse features that can predict a given objective. Diversity is useful for several reasons such as interpretability, r…

Diversityfeature selectionregression

Distributionally Robust Submodular Maximization

2018-02-14 · Matthew Staib, Bryan Wilder, Stefanie Jegelka

Submodular functions have applications throughout machine learning, but in many settings, we do not have direct access to the underlying function $f$. We focus on stochastic functions that are given as an expectation of …