# Interflow: Aggregating Multi-layer Feature Mappings with Attention Mechanism

arXiv: https://arxiv.org/abs/2106.14073 · HTML: https://mlsanity.com/paper/2106.14073/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 34 references that are in the map.
- First posted on arXiv 26 Jun 2021.

## Builds on (most cited first, 24 of 24)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://mlsanity.com/paper/1505.04597/) (2015)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation](https://mlsanity.com/paper/1612.00593/) (2016)
- [Fully Convolutional Networks for Semantic Segmentation](https://mlsanity.com/paper/1411.4038/) (2014)
- [You Only Look Once: Unified, Real-Time Object Detection](https://mlsanity.com/paper/1506.02640/) (2015)
- [Squeeze-and-Excitation Networks](https://mlsanity.com/paper/1709.01507/) (2017)
- [Going Deeper with Convolutions](https://mlsanity.com/paper/1409.4842/) (2014)
- [SSD: Single Shot MultiBox Detector](https://mlsanity.com/paper/1512.02325/) (2015)
- [CBAM: Convolutional Block Attention Module](https://mlsanity.com/paper/1807.06521/) (2018)
- [Rethinking Atrous Convolution for Semantic Image Segmentation](https://mlsanity.com/paper/1706.05587/) (2017)
- [Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks](https://mlsanity.com/paper/1511.06434/) (2015)
- [Two-Stream Convolutional Networks for Action Recognition in Videos](https://mlsanity.com/paper/1406.2199/) (2014)
- [Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields](https://mlsanity.com/paper/1611.08050/) (2016)
- [StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks](https://mlsanity.com/paper/1612.03242/) (2016)
- [DensePose: Dense Human Pose Estimation In The Wild](https://mlsanity.com/paper/1802.00434/) (2018)
- [RMPE: Regional Multi-person Pose Estimation](https://mlsanity.com/paper/1612.00137/) (2016)
- [DSSD : Deconvolutional Single Shot Detector](https://mlsanity.com/paper/1701.06659/) (2017)
- [HyperNet: Towards Accurate Region Proposal Generation and Joint Object Detection](https://mlsanity.com/paper/1604.00600/) (2016)
- [PolyNet: A Pursuit of Structural Diversity in Very Deep Networks](https://mlsanity.com/paper/1611.05725/) (2016)
- [Parameter-Free Spatial Attention Network for Person Re-Identification](https://mlsanity.com/paper/1811.12150/) (2018)
- [Fusion of convolution neural network, support vector machine and Sobel filter for accurate detection of COVID-19 patients using X-ray images](https://mlsanity.com/paper/2102.06883/) (2021)

## Built on by (most cited first, 2 of 2)

- [NSTR: Neural Spectral Transport Representation for Space-Varying Frequency Fields](https://mlsanity.com/paper/2511.18384/) (2025)
- [Scaling Implicit Fields via Hypernetwork-Driven Multiscale Coordinate Transformations](https://mlsanity.com/paper/2511.18387/) (2025)

## Abstract (arXiv)

Traditionally, CNN models possess hierarchical structures and utilize the feature mapping of the last layer to obtain the prediction output. However, it can be difficulty to settle the optimal network depth and make the middle layers learn distinguished features. This paper proposes the Interflow algorithm specially for traditional CNN models. Interflow divides CNNs into several stages according to the depth and makes predictions by the feature mappings in each stage. Subsequently, we input these prediction branches into a well-designed attention module, which learns the weights of these prediction branches, aggregates them and obtains the final output. Interflow weights and fuses the features learned in both shallower and deeper layers, making the feature information at each stage processed reasonably and effectively, enabling the middle layers to learn more distinguished features, and enhancing the model representation ability. In addition, Interflow can alleviate gradient vanishing problem, lower the difficulty of network depth selection, and lighten possible over-fitting problem by introducing attention mechanism. Besides, it can avoid network degradation as a byproduct. Compared with the original model, the CNN model with Interflow achieves higher test accuracy on multiple benchmark datasets.