`

Timezone: »

 
Poster
Do Vision Transformers See Like Convolutional Neural Networks?
Maithra Raghu · Tom Unterthiner · Simon Kornblith · Chiyuan Zhang · Alexey Dosovitskiy

Tue Dec 07 08:30 AM -- 10:00 AM (PST) @ None #None

Convolutional neural networks (CNNs) have so far been the de-facto model for visual data. Recent work has shown that (Vision) Transformer models (ViT) can achieve comparable or even superior performance on image classification tasks. This raises a central question: how are Vision Transformers solving these tasks? Are they acting like convolutional networks, or learning entirely different visual representations? Analyzing the internal representation structure of ViTs and CNNs on image classification benchmarks, we find striking differences between the two architectures, such as ViT having more uniform representations across all layers. We explore how these differences arise, finding crucial roles played by self-attention, which enables early aggregation of global information, and ViT residual connections, which strongly propagate features from lower to higher layers. We study the ramifications for spatial localization, demonstrating ViTs successfully preserve input spatial information, with noticeable effects from different classification methods. Finally, we study the effect of (pretraining) dataset scale on intermediate features and transfer learning, and conclude with a discussion on connections to new architectures such as the MLP-Mixer.

Author Information

Maithra Raghu (Google Brain)
Tom Unterthiner (Google Research)
Simon Kornblith (Google Brain)
Chiyuan Zhang (Google Research)
Alexey Dosovitskiy (Inceptive)

More from the Same Authors