ADMM SVM training with Nystrom approximation reduces kernel matrix dimension 32x with 2% accuracy loss and yields extreme efficiency gains in an edge seizure detector chip.
A 12nW always-on acoustic sensing and object recognition microsys- tem using frequency-domain feature extraction and SVM classification
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.SP 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Hardware-Efficient ADMM-Based SVM Training Algorithm for Edge Computing
ADMM SVM training with Nystrom approximation reduces kernel matrix dimension 32x with 2% accuracy loss and yields extreme efficiency gains in an edge seizure detector chip.