Learnable Heterogeneous Convolution: Learning both topology and strength
This addresses efficiency and accuracy issues in deep learning for computer vision, offering a novel method that unifies existing techniques, though it is incremental in building on biological inspiration.
The paper tackles the high computational complexity of convolution in neural networks by proposing Learnable Heterogeneous Convolution, which jointly learns kernel shape and weights, reducing computation by up to 5x on CIFAR10 and 2x on ImageNet without performance loss or improving accuracy by up to 1.0% on CIFAR10 and 0.5% on ImageNet.
Existing convolution techniques in artificial neural networks suffer from huge computation complexity, while the biological neural network works in a much more powerful yet efficient way. Inspired by the biological plasticity of dendritic topology and synaptic strength, our method, Learnable Heterogeneous Convolution, realizes joint learning of kernel shape and weights, which unifies existing handcrafted convolution techniques in a data-driven way. A model based on our method can converge with structural sparse weights and then be accelerated by devices of high parallelism. In the experiments, our method either reduces VGG16/19 and ResNet34/50 computation by nearly 5x on CIFAR10 and 2x on ImageNet without harming the performance, where the weights are compressed by 10x and 4x respectively; or improves the accuracy by up to 1.0% on CIFAR10 and 0.5% on ImageNet with slightly higher efficiency. The code will be available on www.github.com/Genera1Z/LearnableHeterogeneousConvolution.