Learning Transferable Architectures for Scalable Image Recognition
read the original abstract
Developing neural network image classification models often requires significant architecture engineering. In this paper, we study a method to learn the model architectures directly on the dataset of interest. As this approach is expensive when the dataset is large, we propose to search for an architectural building block on a small dataset and then transfer the block to a larger dataset. The key contribution of this work is the design of a new search space (the "NASNet search space") which enables transferability. In our experiments, we search for the best convolutional layer (or "cell") on the CIFAR-10 dataset and then apply this cell to the ImageNet dataset by stacking together more copies of this cell, each with their own parameters to design a convolutional architecture, named "NASNet architecture". We also introduce a new regularization technique called ScheduledDropPath that significantly improves generalization in the NASNet models. On CIFAR-10 itself, NASNet achieves 2.4% error rate, which is state-of-the-art. On ImageNet, NASNet achieves, among the published works, state-of-the-art accuracy of 82.7% top-1 and 96.2% top-5 on ImageNet. Our model is 1.2% better in top-1 accuracy than the best human-invented architectures while having 9 billion fewer FLOPS - a reduction of 28% in computational demand from the previous state-of-the-art model. When evaluated at different levels of computational cost, accuracies of NASNets exceed those of the state-of-the-art human-designed models. For instance, a small version of NASNet also achieves 74% top-1 accuracy, which is 3.1% better than equivalently-sized, state-of-the-art models for mobile platforms. Finally, the learned features by NASNet used with the Faster-RCNN framework surpass state-of-the-art by 4.0% achieving 43.1% mAP on the COCO dataset.
This paper has not been read by Pith yet.
Forward citations
Cited by 7 Pith papers
-
AGAN: Towards Automated Design of Generative Adversarial Networks
AGAN is the first neural architecture search method for GANs that discovers architectures outperforming state-of-the-art on CIFAR-10 unsupervised image generation and competitive on supervised tasks.
-
Auto-FP: An Experimental Study of Automated Feature Preprocessing for Tabular Data
Experimental comparison of 15 HPO and NAS algorithms for automated feature preprocessing on 45 tabular datasets finds evolution-based methods and random search as top performers.
-
NetTailor: Tuning the Architecture, Not Just the Weights
NetTailor adapts CNN architecture for new tasks by assembling pre-trained universal blocks with task-specific layers, trained via activation mimicry and complexity penalties to match accuracy while reducing size for s...
-
Placeto: Learning Generalizable Device Placement Algorithms for Distributed Machine Learning
Placeto learns generalizable RL policies for device placement via iterative improvements and graph embeddings, needing up to 6.1x fewer steps than prior methods and applying to unseen graphs without retraining.
-
Searching for Activation Functions
Automated search discovers Swish activation f(x) = x * sigmoid(βx) that improves top-1 ImageNet accuracy over ReLU by 0.9% on Mobile NASNet-A and 0.6% on Inception-ResNet-v2.
-
MLFriend: Interactive Prediction Task Recommendation for Event-Driven Time-Series Data
MLFriend enumerates prediction tasks for event-driven time-series data and interactively recommends useful ones, with evaluation on three datasets yielding 2885 tasks of which 722 were deemed useful by experts.
-
Clustering and Classification Networks
Three-level FC architecture with softmax-based clustering and mask reclassification reaches 11.56% error on CIFAR-100.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.