Data can be represented as a vector in a continuous space, but often it is concentrated in a narrow area. Can it be improved?
Computer vision, image recognition and classification, model architectures such as ResNets, MNIST and more.
Augmenting images training data to add variability doesn't always end well since the augmentation changes the semantic content of the image. Here's how to solve it.
This paper suggests a new computationally efficient method for constructing low-dimensional representation of unlabeled data.
A review and clear explanation of the NeRF method, which can be used to synthesize 3D scenes out of an input image. This method is the base for many other research methods that followed.
Yam Peleg examines a kaggle solution using convolutional neural networks which can process tabular data while being columns order agnostic.