# Stochastic Iterative Hard Thresholding for Graph-structured Sparsity Optimization

@inproceedings{Zhou2019StochasticIH, title={Stochastic Iterative Hard Thresholding for Graph-structured Sparsity Optimization}, author={Baojian Zhou and Feng Chen and Yiming Ying}, booktitle={ICML}, year={2019} }

Stochastic optimization algorithms update models with cheap per-iteration costs sequentially, which makes them amenable for large-scale data analysis. Such algorithms have been widely studied for structured sparse models where the sparsity information is very specific, e.g., convex sparsity-inducing norms or $\ell^0$-norm. However, these norms cannot be directly applied to the problem of complex (non-convex) graph-structured sparsity models, which have important application in disease outbreak… CONTINUE READING

Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

#### Citations

##### Publications citing this paper.

## Loopless Semi-Stochastic Gradient Descent with Less Hard Thresholding for Sparse Learning

VIEW 1 EXCERPT

CITES METHODS

#### References

##### Publications referenced by this paper.

SHOWING 1-10 OF 60 REFERENCES

## Fast recovery from a union of subspaces

VIEW 26 EXCERPTS

HIGHLY INFLUENTIAL

## Linear Convergence of Stochastic Iterative Greedy Algorithms With Sparse Constraints

VIEW 19 EXCERPTS

HIGHLY INFLUENTIAL

## A Nearly-Linear Time Framework for Graph-Structured Sparsity

VIEW 13 EXCERPTS

HIGHLY INFLUENTIAL

## Train faster, generalize better: Stability of stochastic gradient descent

VIEW 17 EXCERPTS

HIGHLY INFLUENTIAL

## A fast approximation algorithm for tree-sparse recovery

VIEW 15 EXCERPTS

HIGHLY INFLUENTIAL

## Greedy sparsity-constrained optimization

VIEW 6 EXCERPTS

HIGHLY INFLUENTIAL

## Fast global convergence of gradient methods for high-dimensional statistical recovery

VIEW 9 EXCERPTS

HIGHLY INFLUENTIAL

## Group lasso with overlap and graph lasso

VIEW 3 EXCERPTS

HIGHLY INFLUENTIAL

## How to Escape Saddle Points Efficiently

VIEW 3 EXCERPTS

HIGHLY INFLUENTIAL

## CoSaMP: iterative signal recovery from incomplete and inaccurate samples

VIEW 5 EXCERPTS

HIGHLY INFLUENTIAL