Perceiver: One Neural-Network Model for Multiple Input Data Types

Google's DeepMind company has recently released a state-of-the-art deep-learning model called Perceiver that receives and processes multiple input data ranging from audio to images, similarly to how the human brain perceives multimodal data. Perceiver is able to receive and classify input multiple data types, namely point cloud, audio and images. For this purpose, the deep-learning model is based on transformers (a.k.a. Usually the bottleneck of using transformers is the quadratic number of operations needed for algorithms. For instance, processing an image measuring 224 pixels by 224 pixels could lead to 224 2 operations, over 50,000, which is a huge computational overhead.