How neural networks hide training data in learned weights

Original: From Latent Space to Training Data: Explainable Specialization in Minimal MLPs

Source: arxiv.org ↗