DeepSwarm: Optimising Convolutional Neural Networks using Swarm Intelligence
This is an incremental improvement for deep learning researchers and practitioners seeking efficient neural architecture search methods.
The authors tackled neural architecture search by proposing DeepSwarm, a method using Ant Colony Optimization to generate architectures, achieving competitive performance on MNIST, Fashion-MNIST, and CIFAR-10 datasets compared to existing systems.
In this paper we propose DeepSwarm, a novel neural architecture search (NAS) method based on Swarm Intelligence principles. At its core DeepSwarm uses Ant Colony Optimization (ACO) to generate ant population which uses the pheromone information to collectively search for the best neural architecture. Furthermore, by using local and global pheromone update rules our method ensures the balance between exploitation and exploration. On top of this, to make our method more efficient we combine progressive neural architecture search with weight reusability. Furthermore, due to the nature of ACO our method can incorporate heuristic information which can further speed up the search process. After systematic and extensive evaluation, we discover that on three different datasets (MNIST, Fashion-MNIST, and CIFAR-10) when compared to existing systems our proposed method demonstrates competitive performance. Finally, we open source DeepSwarm as a NAS library and hope it can be used by more deep learning researchers and practitioners.