Parameter Prediction for Unseen Deep Architectures

Deep learning has been successful in automating the design of features in machine learning pipelines. However, the algorithms optimizing neural network parameters remain largely hand-designed and computationally inefficient. We study if we can use deep learning to directly predict these parameters by exploiting the past knowledge of training other networks. We introduce a large-scale dataset of diverse computational graphs of neural architectures - DeepNets-1M - and use it to explore parameter prediction on CIFAR-10 and ImageNet. By leveraging advances in graph neural networks, we propose a hypernetwork that can predict performant parameters in a single forward pass taking a fraction of a second, even on a CPU. The proposed model achieves surprisingly good performance on unseen and diverse networks. For example, it is able to predict all 24 million parameters of a ResNet-50 achieving a 60% accuracy on CIFAR-10. On ImageNet, top-5 accuracy of some of our networks approaches 50%. Our task along with the model and results can potentially lead to a new, more computationally efficient paradigm of training networks. Our model also learns a strong representation of neural architectures enabling their analysis.

PDF Abstract NeurIPS 2021 PDF NeurIPS 2021 Abstract


Introduced in the Paper:


Used in the Paper:


Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Parameter Prediction CIFAR10 GHN-2 Classification Accuracy (Deep) 60.5 # 1
Classification Accuracy (ID-test) 66.9 # 1
Classification Accuracy (Wide) 64 # 1
Classification Accuracy (Dense) 65.8 # 1
Classification Accuracy (BN-free) 36.8 # 1
Classification Accuracy (ResNet-50) 58.6 # 1
Classification Accuracy (ViT) 11.4 # 1