Connectivity Learning in Multi-Branch Networks

By Karim Ahmed, Lorenzo Torresani. 2017

28 citations recorded by the source corpus.

While much of the work in the design of convolutional networks over the last five years has revolved around the empirical investigation of the importance of depth, filter sizes, and number of feature channels, recent studies have shown that branching, i.e., splitting the computation along parallel but distinct threads and then aggregating their outputs, represents a new promising dimension for significant improvements in performance. To combat the complexity of design choices in multi-branch architectures, prior work has adopted simple strategies, such as a fixed branching factor, the same input being fed to all parallel branches, and an additive combination of the outputs produced by all branches at aggregation points. In this work we remove these predefined choices and propose an algorithm to learn the connections between branches in the network. Instead of being chosen a priori by the human designer, the multi-branch connectivity is learned simultaneously with the weights of the network by optimizing a single loss function defined with respect to the end task. We demonstrate our approach on the problem of multi-class image classification using three different datasets where it yields consistently higher accuracy compared to the state-of-the-art "ResNeXt" multi-branch network given the same learning capacity.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Quoc V. Le
2. Song Han
3. Hanxiao Liu
4. Liang-Chieh Chen
5. Chuang Gan
6. Bernard Ghanem
7. Mingxing Tan
8. Jiashi Feng

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Dacheng Tao
3. Quoc V. Le
4. Ming-Hsuan Yang
5. Alan L. Yuille
6. Chunhua Shen
7. Yi Yang
8. Xiaogang Wang

In Model Architectures

1. Liang-Chieh Chen
2. Hanxiao Liu
3. Bernard Ghanem
4. Quoc V. Le
5. Andrew G. Howard
6. Karen Simonyan
7. Mark Sandler
8. Ali K. Thabet

Citations in the corpus

• AssembleNet: Searching for Multi-Stream Neural Connectivity in Video Architectures (2019, neutral): “Ahmed and Torresani [1] used learnable gating to connect multiple residual module branches, which similarly requires training of all possible modules and connections.”
• SGAS: Sequential Greedy Architecture Search (2019, background): “In order to search in a continuous domain [44, 1, 46, 53], DARTS [35] proposes a continuous relaxation of the architecture representation, making architecture search differentiable and hence much more efficient.”
• Inverted Residuals and Linear Bottlenecks: Mobile Networks for Classification, Detection and Segmentation (2018, methodology): “Recently there has been lots of progress in algorithmic architecture exploration included hyper-parameter optimization [8, 9, 10] as well as various methods of network pruning [11, 12, 13, 14, 15, 16] and connectivity learning [17, 18].”
• MobileNetV2: Inverted Residuals and Linear Bottlenecks (2018, methodology): “methods of network pruning [12, 13, 14, 15, 16, 17] and connectivity learning [18, 19].”
• DARTS: Differentiable Architecture Search (2018, background): “The idea of searching architectures within a continuous domain is not new (Saxena and Verbeek, 2016; Ahmed and Torresani, 2017; Shin et al., 2018), but there are several major distinctions.”