# SparseMask: Differentiable Connectivity Learning for Dense Image Prediction

## TL;DR

This paper introduces SparseMask, a differentiable method for automatically learning efficient connectivity structures in encoder-decoder networks for dense image prediction, resulting in faster and more parameter-efficient models.

## Contribution

It proposes a novel gradient-based search for optimal, sparse connectivity in dense networks, improving speed and efficiency over existing methods.

## Key findings

- Achieves competitive segmentation results
- Runs over three times faster than state-of-the-art
- Uses less than half the parameters of comparable models

## Abstract

In this paper, we aim at automatically searching an efficient network architecture for dense image prediction. Particularly, we follow the encoder-decoder style and focus on designing a connectivity structure for the decoder. To achieve that, we design a densely connected network with learnable connections, named Fully Dense Network, which contains a large set of possible final connectivity structures. We then employ gradient descent to search the optimal connectivity from the dense connections. The search process is guided by a novel loss function, which pushes the weight of each connection to be binary and the connections to be sparse. The discovered connectivity achieves competitive results on two segmentation datasets, while runs more than three times faster and requires less than half parameters compared to the state-of-the-art methods. An extensive experiment shows that the discovered connectivity is compatible with various backbones and generalizes well to other dense image prediction tasks.