From Xception to NEXcepTion: New Design Decisions and Neural Architecture Search
Hadar Shavit, Filip Jatelnicki, Pol Mor-Puigvent\'os, Wojtek, Kowalczyk

TL;DR
This paper introduces NEXcepTion, a modified Xception architecture enhanced through new design choices and Neural Architecture Search, achieving higher accuracy and throughput on ImageNet compared to the original model.
Contribution
The paper presents a novel architecture, NEXcepTion, with improved performance and efficiency, achieved by combining design modifications, training procedures, and NAS.
Findings
NEXcepTion achieves 81.5% top-1 accuracy on ImageNet.
NEXcepTion-TP reaches 81.8% accuracy with higher throughput.
The models outperform the original Xception architecture.
Abstract
In this paper, we present a modified Xception architecture, the NEXcepTion network. Our network has significantly better performance than the original Xception, achieving top-1 accuracy of 81.5% on the ImageNet validation dataset (an improvement of 2.5%) as well as a 28% higher throughput. Another variant of our model, NEXcepTion-TP, reaches 81.8% top-1 accuracy, similar to ConvNeXt (82.1%), while having a 27% higher throughput. Our model is the result of applying improved training procedures and new design decisions combined with an application of Neural Architecture Search (NAS) on a smaller dataset. These findings call for revisiting older architectures and reassessing their potential when combined with the latest enhancements.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Neural Network Applications · Domain Adaptation and Few-Shot Learning · COVID-19 diagnosis using AI
MethodsConvNeXt · Pointwise Convolution · Average Pooling · Residual Connection · Max Pooling · Depthwise Convolution · Softmax · 1x1 Convolution · Convolution · Depthwise Separable Convolution
