The Convex Relaxation Barrier, Revisited: Tightened Single-Neuron Relaxations for Neural Network Verification.
Christian TjandraatmadjaRoss AndersonJoey HuchetteWill MaKrunal PatelJuan Pablo VielmaPublished in: NeurIPS (2020)
Keyphrases
- convex relaxation
- single neuron
- neural network
- convex optimization
- globally optimal
- control algorithm
- multistage
- multi label
- semidefinite
- activation function
- spike trains
- multiple kernel learning
- optimization methods
- pid control
- graph cuts
- pid controller
- network architecture
- multilayer perceptron
- radial basis function
- hidden layer
- artificial neural networks
- control strategy
- machine learning algorithms
- support vector
- kernel function
- linear combination
- fuzzy logic
- learning algorithm