AutoSciDACT LIGO Dataset

This is the LIGO (gravitational waves) dataset from the paper "AutoSciDACT: Automated Scientific Discovery through Contrastive Embedding and Hypothesis Testing", presented at NeurIPS 2025 in San Diego, CA. The npz files contains low-dimensional embeddings of particle jets from the JetClass dataset, obtained via supervised contrastive contrastive learning, as described in the paper. Each file corresponds to a different embedding dimensionality, as indicated in the file name. There are 81781 spread across 9 classes of gravitational wave signals. The files contain the corresponding keys: data - the N-dimensional gravitational wave embeddings, shape N x d (N = number of samples, d = labels - class labels 0 - 8 (0 = Sine-Gaussian, 1 = Binary Black Hole Merger (BBH), 2 = Gaussian, 3 = Cusp, 4 = Kink, 5 = Double Kink, 6 = White Noise Burst, 7 = Background (no signal), 8 = Glitches (detector malfunctions))