EMind reports that one masked-autoencoder transformer, pretrained on 81 million heterogeneous IQ samples, transfers to seven electromagnetic signal tasks with strong accuracy, but post-hoc checkpoint selection and missing baselines temper the claim.
Self-supervised visual feature learning with deep neural networks: A survey
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SP 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
EMind: A Foundation Model for Multi-task Electromagnetic Signals Understanding
EMind reports that one masked-autoencoder transformer, pretrained on 81 million heterogeneous IQ samples, transfers to seven electromagnetic signal tasks with strong accuracy, but post-hoc checkpoint selection and missing baselines temper the claim.