Hyper-Representations: Self-Supervised Representation Learning on Neural Network Weights for Model Characteristic Prediction
This work addresses the challenge of interpreting neural network weights for researchers and practitioners, though it appears incremental as it builds on existing self-supervised learning methods.
The paper tackles the problem of understanding neural network weight spaces by proposing self-supervised learning to learn hyper-representations from populations of neural networks, showing that these representations outperform prior work in predicting model characteristics like hyper-parameters, test accuracy, and generalization gap.
Self-Supervised Learning (SSL) has been shown to learn useful and information-preserving representations. Neural Networks (NNs) are widely applied, yet their weight space is still not fully understood. Therefore, we propose to use SSL to learn hyper-representations of the weights of populations of NNs. To that end, we introduce domain specific data augmentations and an adapted attention architecture. Our empirical evaluation demonstrates that self-supervised representation learning in this domain is able to recover diverse NN model characteristics. Further, we show that the proposed learned representations outperform prior work for predicting hyper-parameters, test accuracy, and generalization gap as well as transfer to out-of-distribution settings.