1) 'Densely Connected Convolutional Networks' - G Huang, GB Huang, S Song, K You
obtained state-of-the-art results in image classification on CIFAR-10, CIFAR-100, SVHN, ImageNet
This paper also won CVPR 17 best paper award along with Apple's paper
2) 'U-Net: Convolutional Networks for Biomedical Image Segmentation' - O Ronneberger, P Fischer, T Brox
This paper is quite successful in medical image segmentation and enjoyed success in Kaggle too
3) 'The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation' -
obtained state-of-the-art results in segmentation on CamVid and Gatech datasets. This one is by Bengio group
4) 'Wasserstein GAN' - M Arjovsky, S Chintala, L Bottou
This paper solved a lot of issues with GANs stability. The change in loss function and training paradigm
make loss functions interpretable. They obtained good results even with a normal mlp on LSUN dataset.
1) 'Densely Connected Convolutional Networks' - G Huang, GB Huang, S Song, K You obtained state-of-the-art results in image classification on CIFAR-10, CIFAR-100, SVHN, ImageNet This paper also won CVPR 17 best paper award along with Apple's paper
2) 'U-Net: Convolutional Networks for Biomedical Image Segmentation' - O Ronneberger, P Fischer, T Brox This paper is quite successful in medical image segmentation and enjoyed success in Kaggle too
3) 'The One Hundred Layers Tiramisu: Fully Convolutional DenseNets for Semantic Segmentation' - obtained state-of-the-art results in segmentation on CamVid and Gatech datasets. This one is by Bengio group
4) 'Wasserstein GAN' - M Arjovsky, S Chintala, L Bottou This paper solved a lot of issues with GANs stability. The change in loss function and training paradigm make loss functions interpretable. They obtained good results even with a normal mlp on LSUN dataset.