# Practical Block-wise Neural Network Architecture Generation

arXiv: https://arxiv.org/abs/1708.05552 · HTML: https://mlsanity.com/paper/1708.05552/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Aug 2017.

## Built on by (most cited first, 5 of 5)

- [Discovering Physics-Informed Neural Networks Model for Solving Partial Differential Equations through Evolutionary Computation](https://mlsanity.com/paper/2405.11208/) (2024)
- [Efficient Visual Recognition with Deep Neural Networks: A Survey on Recent Advances and New Directions](https://mlsanity.com/paper/2108.13055/) (2021)
- [Neural Architecture Searching for Facial Attributes-based Depression Recognition](https://mlsanity.com/paper/2201.09799/) (2022)
- [Warm-starting DARTS using meta-learning](https://mlsanity.com/paper/2205.06355/) (2022)
- [Evolution under Length Constraints for CNN Architecture design](https://mlsanity.com/paper/2305.03416/) (2023)

## Abstract (arXiv)

Convolutional neural networks have gained a remarkable success in computer vision. However, most usable network architectures are hand-crafted and usually require expertise and elaborate design. In this paper, we provide a block-wise network generation pipeline called BlockQNN which automatically builds high-performance networks using the Q-Learning paradigm with epsilon-greedy exploration strategy. The optimal network block is constructed by the learning agent which is trained sequentially to choose component layers. We stack the block to construct the whole auto-generated network. To accelerate the generation process, we also propose a distributed asynchronous framework and an early stop strategy. The block-wise generation brings unique advantages: (1) it performs competitive results in comparison to the hand-crafted state-of-the-art networks on image classification, additionally, the best network generated by BlockQNN achieves 3.54% top-1 error rate on CIFAR-10 which beats all existing auto-generate networks. (2) in the meanwhile, it offers tremendous reduction of the search space in designing networks which only spends 3 days with 32 GPUs, and (3) moreover, it has strong generalizability that the network built on CIFAR also performs well on a larger-scale ImageNet dataset.