Graph Diffusion-Embedding Network (GDEN)
We present a novel graph diffusion-embedding networks (GDEN) for graph structured data. GDEN is motivated by our closed-form formulation on regularized feature diffusion on graph. GDEN integrates both regularized feature diffusion and low-dimensional embedding simultaneously in a unified network model. Moreover, based on GDEN, we can naturally deal with structured data with multiple graph structures. Experiments on semi-supervised learning tasks on several benchmark datasets demonstrate the better performance of the proposed GDEN when comparing with the traditional GCN models. …

Context-Aware Dynamic Block (CDB)
Although deeper and larger neural networks have achieved better performance nowadays, the complex network structure and increasing computational cost cannot meet the demands of many resource-constrained applications. An effective way to address this problem is to make use of dynamic inference mechanism. Existing methods usually choose to execute or skip an entire specific layer through a switch structure, which can only alter the depth of the network. In this paper, we propose a dynamic inference method called Context-aware Dynamic Block (CDB), which provides more path selection choices in terms of network width and depth during inference. The execution of CDB is determined by a context-aware group controller, which can take into account both historical and object category information. The proposed method can be easily incorporated into most modern network architectures. Experimental results on ImageNet and CIFAR-100 demonstrate the superiority of our method on both efficiency and overall classification quality. To be specific, we integrate CDB block into ResNet-101 and find that our method significantly outperforms their counterparts and saves 45.1% FLOPs. …

Activation Atlases
Activation Atlases (in collaboration with Google researchers), is a new technique for visualizing what interactions between neurons can represent. …

Leader-Follower Elastic Averaging Stochastic Gradient Descent (LEASGD)
Distributed learning systems have enabled training large-scale models over large amount of data in significantly shorter time. In this paper, we focus on decentralized distributed deep learning systems and aim to achieve differential privacy with good convergence rate and low communication cost. To achieve this goal, we propose a new learning algorithm LEASGD (Leader-Follower Elastic Averaging Stochastic Gradient Descent), which is driven by a novel Leader-Follower topology and a differential privacy model.We provide a theoretical analysis of the convergence rate and the trade-off between the performance and privacy in the private setting.The experimental results show that LEASGD outperforms state-of-the-art decentralized learning algorithm DPSGD by achieving steadily lower loss within the same iterations and by reducing the communication cost by 30%. In addition, LEASGD spends less differential privacy budget and has higher final accuracy result than DPSGD under private setting. …