Lower Bounds on the Depth of Integral ReLU Neural Networks via Lattice Polytopes.
Christian HaaseChristoph HertrichGeorg LohoPublished in: ICLR (2023)
Keyphrases
- lower bound
- neural network
- lattice points
- upper bound
- convex hull
- branch and bound
- integer points
- branch and bound algorithm
- pattern recognition
- artificial neural networks
- back propagation
- np hard
- neural network model
- lower and upper bounds
- fuzzy systems
- neural nets
- genetic algorithm
- objective function
- worst case
- lattice structure
- multilayer perceptron
- upper and lower bounds
- geometric properties
- lower bounding
- online algorithms
- vc dimension
- optimal solution
- depth information
- fuzzy logic
- feed forward
- depth map
- training process
- three dimensional
- quadratic assignment problem
- learning algorithm
- sufficient conditions