Stabilizing Gans with Prediction Method Must try! See equation (3) in /www.umiacs.umd.edu/~jaiabhay/paper/ICLR2018.pdf (Abhay Kumar Yadav, Sohil Shah, Zheng Xu, David W. Jacobs, and Tom Goldstein.) Inspired by the "prediction&

ConvNet for NLP Papers: Kim 2014, Convolutional Neural Networks for Sentence Classification /arxiv.org/abs/1408.5882

Important Papers Image Recognition He et. al. Deep Residual Learning for Image Recognition /arxiv.org/pdf/1512.03385.pdf Object Recognition He et. al. Mask R-CNN /arxiv.org/pdf/1703.06870.pdf NLP

Read ICLR Proceedings The Proceedings of ICLR are probably the best concentrated place of good papers - Yoshua Bengio (during his interview with Andrew Ng) ("Of course NIPS and ICML papers are good too.."

Gradient-Based Optimizers Here are some high-performing grandient-based optimizers that are widely used in deep learning. $dW$ denotes the gradient matrix $\frac{\partial J}{\partial W}$ where $J$ is the cost function, $\alpha$ is the learning

Generating Labelled Data from Unlabelled Data An idea for generating unlabelled data from unlabelled data and two high-performing existing classifier. Collect a big amount of unlabelled data Use two(or maybe more?) high performance existing parsers to parse an