Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I want to efficiently solve the following optimization problem: \begin{align} \min &\quad \left\|\mathbf{x}-\mathbf{x}_0\right\|_2^2 + \lambda\left\|\mathbf{x}\right\|_1\\ \text{Subject to}& \quad A\mathbf{x} \leq c, \end{align}

where $\mathbf{x}\in \mathbb{R}^{n\times 1}$, for big values of $n$. I tried coordinate descend, but it doesn't work. I also don't want to formulate it as QP and use interior-point methods because they are slow. Any idea?

share|improve this question
    
Apparently, the problem has been solved recently: Peng Zeng, Yu Zhu, and Tianhong He, Linearly Constrained Lasso with Application in Glioblastoma Data. I haven't read it yet, but it looks like they have designed a LARS type algorithm. –  Taha Jul 31 '12 at 23:26
    
Also any Lasso algorithm with projecting to constraint set after each iteration also should work. –  mirror2image Aug 1 '12 at 7:42

1 Answer 1

"The Constrained LASSO" by Gareth M James, Courtney Paulson, and Paat Rusmevichientong which is under review and linked below

http://www-bcf.usc.edu/~gareth/research/CLassoFinal.pdf

seems to have solved this specific problem.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.