Deep learning has achieved remarkable performance in various tasks thanks to massive labeled datasets. However, there are often cases where labeling large amount of data is challenging or infeasible due to high labeling cost such as labeling by experts or long labeling time per large-scale data sample (e.g., video, very large image). Active learning is one of the ways to query the most informative samples to be annotated among massive unlabeled pool. Two promising directions for active learning that have been recently explored are data distribution-based approach to select data points that are far from current labeled pool and model uncertainty-based approach that relies on the perspective of task model. Unfortunately, the former does not exploit structures from tasks and the latter does not seem to well-utilize overall data distribution. Here, we propose the methods that simultaneously take advantage of both data distribution and model uncertainty approaches. Our proposed methods exploit variational adversarial active learning (VAAL), that considered data distribution of both label and unlabeled pools, by incorporating learning loss prediction module and RankCGAN concept into VAAL by modeling loss prediction as a ranker. We demonstrate that our proposed methods outperform recent state-of-theart active learning methods on various balanced and imbalanced benchmark datasets.

Recently, deep neural network (DNN) based methods for low-dose CT have been investigated to achieve excellent performance in both image quality and computational speed. However, almost all methods using DNNs for low-dose CT require clean ground truth data with full radiation dose to train the DNNs. In this work, we attempt to train DNNs for low-dose CT reconstructions with reduced tube current by investigating unsupervised training of DNNs for denoising sensor measurements or sinograms without full-dose ground truth images. In other words, our proposed methods allow training of DNNs with only noisy low-dose CT measurements. First, the Poisson Unbiased Risk Estimator (PURE) is investigated to train a DNN for denoising CT measurements, and a method is proposed for reconstructing the CT image using filtered back-projection (FBP) and the DNN trained with PURE. Then, the CT forward model-based Weighted Stein's Unbiased Risk Estimator (WSURE) is proposed to train a DNN for denoising CT sinograms and to subsequently reconstruct the CT image using FBP. Our proposed methods achieve excellent performance in both fast computation and reconstructed image quality, which is more comparable to the results of the DNNs trained with full-dose ground truth data than other state-of-the-art denoising methods such as the BM3D, Deep Image Prior, and Deep Decoder. Contents Contents ii List of Figures iv List of Tables viI.LIST OF FIGURES 4.3 Visual results of the denoised sinograms obtained using various methods for I 0 = 4 . The red and yellow boxes represent the enlarged views and their corresponding residuals, respectively. The numbers in orange in the images indicate the RMSE values. The intensity ranges were (-1000, 3000) and (-100,150

Recently, there has been extensive research interest in training deep networks to denoise images without clean reference. However, the representative approaches such as Noise2Noise, Noise2Void, Stein's unbiased risk estimator (SURE), etc. seem to differ from one another and it is difficult to find the coherent mathematical structure. To address this, here we present a novel approach, called Noise2Score, which reveals a missing link in order to unite these seemingly different approaches. Specifically, we show that image denoising problems without clean images can be addressed by finding the mode of the posterior distribution and that the Tweedie's formula offers an explicit solution through the score function (i.e. the gradient of loglikelihood). Our method then uses the recent finding that the score function can be stably estimated from the noisy images using the amortized residual denoising autoencoder, the method of which is closely related to Noise2Noise or Nose2Void. Our Noise2Score approach is so universal that the same network training can be used to remove noises from images that are corrupted by any exponential family distributions and noise parameters. Using extensive experiments with Gaussian, Poisson, and Gamma noises, we show that Noise2Score significantly outperforms the state-of-the-art self-supervised denoising methods in the benchmark data set such as (C)BSD68, Set12, and Kodak, etc.

This paper reviews the first NTIRE challenge on video super-resolution (restoration of rich details in lowresolution video frames) with focus on proposed solutions and results. A new REalistic and Diverse Scenes dataset (REDS) was employed. The challenge was divided into 2 tracks. Track 1 employed standard bicubic downscaling setup while Track 2 had realistic dynamic motion blurs. Each competition had 124 and 104 registered participants. There were total 14 teams in the final testing phase. They gauge the state-of-the-art in video super-resolution.