Reproduce “Semi-Supervised Learning with Ladder Networks” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Antti Rasmus, Harri Valpola, Mikko Honkala, Mathias Berglund, Tapani Raiko

The Ladder network extends unsupervised denoising to supervised and semi-supervised classification by training a classifier alongside a decoder that recovers clean internal features from noisy representations, using lateral encoder information and higher-layer context. On permutation-invariant MNIST with 100 labels, the full model reports 1.06% test error, versus 21.74% for a supervised MLP; some runs failed to converge.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.