Perceiver: General Perception with Iterative Attention
This addresses the need for more general and scalable perception models in AI, reducing reliance on modality-specific architectures, though it is incremental in building upon Transformers.
The paper tackles the problem of building a single deep learning model that can handle diverse high-dimensional inputs across multiple modalities without relying on domain-specific assumptions, and demonstrates that the Perceiver model achieves performance comparable to specialized models like ResNet-50 and ViT on tasks such as ImageNet and AudioSet.
Biological systems perceive the world by simultaneously processing high-dimensional inputs from modalities as diverse as vision, audition, touch, proprioception, etc. The perception models used in deep learning on the other hand are designed for individual modalities, often relying on domain-specific assumptions such as the local grid structures exploited by virtually all existing vision models. These priors introduce helpful inductive biases, but also lock models to individual modalities. In this paper we introduce the Perceiver - a model that builds upon Transformers and hence makes few architectural assumptions about the relationship between its inputs, but that also scales to hundreds of thousands of inputs, like ConvNets. The model leverages an asymmetric attention mechanism to iteratively distill inputs into a tight latent bottleneck, allowing it to scale to handle very large inputs. We show that this architecture is competitive with or outperforms strong, specialized models on classification tasks across various modalities: images, point clouds, audio, video, and video+audio. The Perceiver obtains performance comparable to ResNet-50 and ViT on ImageNet without 2D convolutions by directly attending to 50,000 pixels. It is also competitive in all modalities in AudioSet.