TropNNC: Structured Neural Network Compression Using Tropical Geometry
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866909974020489216 |
|---|---|
| author | Fotopoulos, Konstantinos Maragos, Petros Misiakos, Panagiotis |
| author_facet | Fotopoulos, Konstantinos Maragos, Petros Misiakos, Panagiotis |
| contents | We present TropNNC, a framework for compressing neural networks with linear and convolutional layers and ReLU activations using tropical geometry. By representing a network's output as a tropical rational function, TropNNC enables structured compression via reduction of the corresponding tropical polynomials. Our method refines the geometric approximation of previous work by adaptively selecting the weights of retained neurons. Key contributions include the first application of tropical geometry to convolutional layers and the tightest known theoretical compression bound. TropNNC requires only access to network weights - no training data - and achieves competitive performance on MNIST, CIFAR, and ImageNet, matching strong baselines such as ThiNet and CUP. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2409_03945 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | TropNNC: Structured Neural Network Compression Using Tropical Geometry Fotopoulos, Konstantinos Maragos, Petros Misiakos, Panagiotis Computer Vision and Pattern Recognition We present TropNNC, a framework for compressing neural networks with linear and convolutional layers and ReLU activations using tropical geometry. By representing a network's output as a tropical rational function, TropNNC enables structured compression via reduction of the corresponding tropical polynomials. Our method refines the geometric approximation of previous work by adaptively selecting the weights of retained neurons. Key contributions include the first application of tropical geometry to convolutional layers and the tightest known theoretical compression bound. TropNNC requires only access to network weights - no training data - and achieves competitive performance on MNIST, CIFAR, and ImageNet, matching strong baselines such as ThiNet and CUP. |
| title | TropNNC: Structured Neural Network Compression Using Tropical Geometry |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2409.03945 |