Yuhang Song

Main works: We show that currently popular GANs struggle to learn stochastic transitions in model-based RL with closely matched probability distribution. In response, we propose a novel GAN, namely SGAN, accomplished by a modification to the loss of the discriminator in traditional GAN’s paradigm. We propose the optimal SGAN we are expecting and give 3 pages’ theoretical proof to show how the proposed algorithm can achieve this optimal SGAN. In experiments, SGAN advances multiple domains (including a real-world domain) significantly.

Main works: We identify a class of function decompositions with properties that make them well suited to the critic role in a leading approach to GANs known as Wasserstein GANs. We show that Taylor and Fourier series decompositions belong to our class, provide examples of these critics outperforming standard GAN approaches. We show that this reparameterized critic performs better than standard gradient-penalty wGAN ap- proaches on a set of one-dimensional simulated domains.

Main works: We establish a database collecting subjectsąŕ head movement (HM) positions on panoramic video sequences, and we find from our database that the HM data are highly consistent across subjects. We further find that deep reinforcement learning (DRL) can be applied in predicting HM positions, seen as actions of an agent. Based on our findings, we propose a DRL based HM Prediction (DHP) approach in offline and online versions, called offline-DHP and online-DHP, respectively. Experimental results validate that offline-DHP and online-DHP are effective in predicting HM positions of panoramic video in offline and online manners, respectively. Experimental results also show that the learned offline-DHP model is capable of improving the performance of online-DHP.

Main works: We conduct a case study on human with similar games from Atari, and found the existence of hierarchical shared knowledges across similar tasks. Inspired by how human develop hierarchical knowledges, we propose a novel deep network, namely Generalization Tower Network (GTN), enabling task-label-free multi-task RL within a single model. The main novelty of GTN is to introduce vertical streams, the effectiveness of which is validated by Fisher Sensitivity (FS) analysis. Experimental results verify that our GTN architecture is able to advance the state-of-the-art multi-task RL, via being tested on 51 Atari games.

Main works: We proposed a dual MDPs structure to estimate the "learning potential" of an exploration noise. Accordingly, an algorithm is proposed to achieve deep exploration via modeling the "learning potential". Experimental results advance the state-of-the-art on part (65%) of the continuous control tasks from MuJoCo.

Main works: We focus on a challenging problem of model-free video quality control, and propose a novel PID-based model-free quality control (PQC) method for video coding. We proposed to apply the Laplace domain analysis to model the relationship between quantization parameter (QP) and control error in this formulation. Experimental results show that our PQC method is effective in both control accuracy and quality fluctuation.

The top entrepreneurship competition in China; Directly obtain graduate admission privilege to our university for this honor.

The top academic competition in our university; Directly obtain graduate admission privilege to our university for this honor;

The top innovation competition in our university; Directly obtain graduate admission privilege to our university for this honor;

1st Author&1st Prize

1st Author&1st Prize

1st Author&1st

1st Author&1st Prize

1st Prize

1st Author&1st Prize

1st Prize

1st Author&1st Prize

1st Prize

We propose an automatic counting software system based on smartphones in the Michelson interferometer experiment. The software comes with a smartphone camera to capture the interference fringe image. Trigonometric cosine Fourier transform is used to analyze the characteristics of the image changes of the interference fringes in the frequency domain. Based on this system, the number of interference fringes can be obtained automatically and directly. The system is simple to operate and has a counting accuracy of up to 0.5 fringe. With the popularity of smartphone applications, the system has broad application prospects.