3 ms·
Interesting! Relevant work: 1. What's one of the most important hyperparameters? Architecture: https://arxiv.org/abs/1806.09055 https://arxiv.org/abs/1806.090
by axiom92 7y ago
Interesting!
Relevant work:
1. What's one of the most important hyperparameters? Architecture: https://arxiv.org/abs/1806.09055 https://arxiv.org/abs/1806.09055
2. Dropout: https://papers.nips.cc/paper/5032-adaptive-dropout-for-training-deep-neural-networks.pdf https://papers.nips.cc/paper/5032-adaptive-dropout-for-train...
3. Learning activation thresholds: https://datascience.stackexchange.com/questions/18583/what-is-the-difference-between-leakyrelu-and-prelu https://datascience.stackexchange.com/questions/18583/what-i...
- shoeffner 7y agoRegarding 2: I saw a talk yesterday by Alex Hernández-García in which he showed that dropout and weight decay can be replaced with data augmentation in many cases. The relevant paper by him and König is https://arxiv.org/abs/1806.03852v2 https://arxiv.org/abs/1806.03852v2 edit: changed to /abs/ instead of /pdf/ link