Neural Echos: Depthwise Convolutional Filters Replicate Biological Receptive Fields
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_ | 1866916096849739776 |
|---|---|
| author | Babaiee, Zahra Kiasari, Peyman M. Rus, Daniela Grosu, Radu |
| author_facet | Babaiee, Zahra Kiasari, Peyman M. Rus, Daniela Grosu, Radu |
| contents | In this study, we present evidence suggesting that depthwise convolutional kernels are effectively replicating the structural intricacies of the biological receptive fields observed in the mammalian retina. We provide analytics of trained kernels from various state-of-the-art models substantiating this evidence. Inspired by this intriguing discovery, we propose an initialization scheme that draws inspiration from the biological receptive fields. Experimental analysis of the ImageNet dataset with multiple CNN architectures featuring depthwise convolutions reveals a marked enhancement in the accuracy of the learned model when initialized with biologically derived weights. This underlies the potential for biologically inspired computational models to further our understanding of vision processing systems and to improve the efficacy of convolutional networks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_10178 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Neural Echos: Depthwise Convolutional Filters Replicate Biological Receptive Fields Babaiee, Zahra Kiasari, Peyman M. Rus, Daniela Grosu, Radu Computer Vision and Pattern Recognition Artificial Intelligence Neural and Evolutionary Computing In this study, we present evidence suggesting that depthwise convolutional kernels are effectively replicating the structural intricacies of the biological receptive fields observed in the mammalian retina. We provide analytics of trained kernels from various state-of-the-art models substantiating this evidence. Inspired by this intriguing discovery, we propose an initialization scheme that draws inspiration from the biological receptive fields. Experimental analysis of the ImageNet dataset with multiple CNN architectures featuring depthwise convolutions reveals a marked enhancement in the accuracy of the learned model when initialized with biologically derived weights. This underlies the potential for biologically inspired computational models to further our understanding of vision processing systems and to improve the efficacy of convolutional networks. |
| title | Neural Echos: Depthwise Convolutional Filters Replicate Biological Receptive Fields |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2401.10178 |