from Initialization [Vaishnavh Nagarajan+, NIPSW2017] http://www.cs.cmu.edu/~vaishnan/papers/nips17_dltp.pdf • Towards Understanding the Role of Over-Parametrization in Generalization of Neural Networks [Behnam Neyshabur+, arXiv2018] https://arxiv.org/abs/1805.12076 ランダム初期化時と学習後の重みの値の距離に基づいて汎化誤差を分析 • DropBack: Continuous Pruning During Training [Maximilian Golub+, arXiv2018] https://arxiv.org/abs/1806.06949 • Intriguing Properties of Randomly Weighted Networks: Generalizing While Learning Next to Nothing [Amir Rosenfeld+,arXiv2018] https://arxiv.org/abs/1802.00844 重みの大半をランダム初期値で固定し、一部の重みのみを更新 • Insights on representational similarity in neural networks with canonical correlation [Ari S. Morcos+, arXiv2018] https://arxiv.org/abs/1806.05759 大きなネットワークほど似た解に収束する