arXiv subjects

Dong Liu

Publications and source records attributed to Dong Liu.

At least 325 records · Page 18Linked to original sources

arXiv2019

Customizable Architecture Search for Semantic Segmentation

In this paper, we propose a Customizable Architecture Search (CAS) approach to automatically generate a network architecture for semantic image segmentation. The generated network consists of a sequence of stacked computation cells. A computation cell is represented as a directed acyclic graph, in which each node is a hidden representation (i.e., feature map) and each edge is associated with an operation (e.g., convolution and pooling), which transforms data to a new layer. During the training, the CAS algorithm explores the search space for an optimized computation cell to build a network. The cells of the same type share one architecture but with different weights. In real applications, however, an optimization may need to be conducted under some constraints such as GPU time and model size. To this end, a cost corresponding to the constraint will be assigned to each operation. When an operation is selected during the search, its associated cost will be added to the objective. As a result, our CAS is able to search an optimized architecture with customized constraints. The approach has been thoroughly evaluated on Cityscapes and CamVid datasets, and demonstrates superior performance over several state-of-the-art techniques. More remarkably, our CAS achieves 72.3% mIoU on the Cityscapes dataset with speed of 108 FPS on an Nvidia TitanXp GPU.

cs.CV↗

arXiv2019

Residual Objectness for Imbalance Reduction

For a long time, object detectors have suffered from extreme imbalance between foregrounds and backgrounds. While several sampling/reweighting schemes have been explored to alleviate the imbalance, they are usually heuristic and demand laborious hyper-parameters tuning, which is hard to achieve the optimality. In this paper, we first reveal that such the imbalance could be addressed in a learning-based manner. Guided by this illuminating observation, we propose a novel Residual Objectness (ResObj) mechanism that addresses the imbalance by end-to-end optimization, while no further hand-crafted sampling/reweighting is required. Specifically, by applying multiple cascaded objectness-related modules with residual connections, we formulate an elegant consecutive refinement procedure for distinguishing the foregrounds from backgrounds, thereby progressively addressing the imbalance. Extensive experiments present the effectiveness of our method, as well as its compatibility and adaptivity for both region-based and one-stage detectors, namely, the RetinaNet-ResObj, YOLOv3-ResObj and FasterRCNN-ResObj achieve relative 3.6%, 3.9%, 3.2% Average Precision (AP) improvements compared with their vanilla models on COCO, respectively.

cs.CV↗

arXiv2019

$α$ Belief Propagation as Fully Factorized Approximation

Belief propagation (BP) can do exact inference in loop-free graphs, but its performance could be poor in graphs with loops, and the understanding of its solution is limited. This work gives an interpretable belief propagation rule that is actually minimization of a localized $α$-divergence. We term this algorithm as $α$ belief propagation ($α$-BP). The performance of $α$-BP is tested in MAP (maximum a posterior) inference problems, where $α$-BP can outperform (loopy) BP by a significant margin even in fully-connected graphs.

cs.LG↗

arXiv2019

Two-Stream Action Recognition-Oriented Video Super-Resolution

We study the video super-resolution (SR) problem for facilitating video analytics tasks, e.g. action recognition, instead of for visual quality. The popular action recognition methods based on convolutional networks, exemplified by two-stream networks, are not directly applicable on video of low spatial resolution. This can be remedied by performing video SR prior to recognition, which motivates us to improve the SR procedure for recognition accuracy. Tailored for two-stream action recognition networks, we propose two video SR methods for the spatial and temporal streams respectively. On the one hand, we observe that regions with action are more important to recognition, and we propose an optical-flow guided weighted mean-squared-error loss for our spatial-oriented SR (SoSR) network to emphasize the reconstruction of moving objects. On the other hand, we observe that existing video SR methods incur temporal discontinuity between frames, which also worsens the recognition accuracy, and we propose a siamese network for our temporal-oriented SR (ToSR) training that emphasizes the temporal continuity between consecutive frames. We perform experiments using two state-of-the-art action recognition networks and two well-known datasets--UCF101 and HMDB51. Results demonstrate the effectiveness of our proposed SoSR and ToSR in improving recognition accuracy.

cs.CV↗

arXiv2019

Frank-Wolfe Network: An Interpretable Deep Structure for Non-Sparse Coding

The problem of $L_p$-norm constrained coding is to convert signal into code that lies inside an $L_p$-ball and most faithfully reconstructs the signal. Previous works under the name of sparse coding considered the cases of $L_0$ and $L_1$ norms. The cases with $p>1$ values, i.e. non-sparse coding studied in this paper, remain a difficulty. We propose an interpretable deep structure namely Frank-Wolfe Network (F-W Net), whose architecture is inspired by unrolling and truncating the Frank-Wolfe algorithm for solving an $L_p$-norm constrained problem with $p\geq 1$. We show that the Frank-Wolfe solver for the $L_p$-norm constraint leads to a novel closed-form nonlinear unit, which is parameterized by $p$ and termed $pool_p$. The $pool_p$ unit links the conventional pooling, activation, and normalization operations, making F-W Net distinct from existing deep networks either heuristically designed or converted from projected gradient descent algorithms. We further show that the hyper-parameter $p$ can be made learnable instead of pre-chosen in F-W Net, which gracefully solves the non-sparse coding problem even with unknown $p$. We evaluate the performance of F-W Net on an extensive range of simulations as well as the task of handwritten digit recognition, where F-W Net exhibits strong learning capability. We then propose a convolutional version of F-W Net, and apply the convolutional F-W Net into image denoising and super-resolution tasks, where F-W Net all demonstrates impressive effectiveness, flexibility, and robustness.

cs.CV↗

arXiv2019

Model-Free Unsupervised Learning for Optimization Problems with Constraints

In many optimization problems in wireless communications, the expressions of objective function or constraints are hard or even impossible to derive, which makes the solutions difficult to find. In this paper, we propose a model-free learning framework to solve constrained optimization problems without the supervision of the optimal solution. Neural networks are used respectively for parameterizing the function to be optimized, parameterizing the Lagrange multiplier associated with instantaneous constraints, and approximating the unknown objective function or constraints. We provide learning algorithms to train all the neural networks simultaneously, and reveal the connections of the proposed framework with reinforcement learning. Numerical and simulation results validate the proposed framework and demonstrate the efficiency of model-free learning by taking power control problem as an example.

cs.LG↗

arXiv2019

Composition-Aware Image Aesthetics Assessment

Automatic image aesthetics assessment is important for a wide variety of applications such as on-line photo suggestion, photo album management and image retrieval. Previous methods have focused on mapping the holistic image content to a high or low aesthetics rating. However, the composition information of an image characterizes the harmony of its visual elements according to the principles of art, and provides richer information for learning aesthetics. In this work, we propose to model the image composition information as the mutual dependency of its local regions, and design a novel architecture to leverage such information to boost the performance of aesthetics assessment. To achieve this, we densely partition an image into local regions and compute aesthetics-preserving features over the regions to characterize the aesthetics properties of image content. With the feature representation of local regions, we build a region composition graph in which each node denotes one region and any two nodes are connected by an edge weighted by the similarity of the region features. We perform reasoning on this graph via graph convolution, in which the activation of each node is determined by its highly correlated neighbors. Our method naturally uncovers the mutual dependency of local regions in the network training procedure, and achieves the state-of-the-art performance on the benchmark visual aesthetics datasets.

cs.CV↗

arXiv2019

Simple restricted modules for Neveu-Schwarz algebra

In this paper, we give a construction of simple modules generalizing and including both highest weight and Whittaker modules for the Neveu-Schwarz algebra, in the spirit of the work of Mazorchuk and Zhao on simple Virasoro modules. We establish a 1-1 correspondence between simple restricted Neveu-Schwarz modules and simple modules of a family of finite dimensional solvable Lie superalgebras associated to the Neveu-Schwarz algebra. Moreover, for two of these superalgebras all simple modules are classified.

math.RT↗

arXiv2019

High-Efficiency Second Harmonic Generation of Low-Temporal-Coherent Light Pulse

The nonlinear frequency conversion of low-temporal-coherent light holds a variety of applications and has attracted considerable interest. However, its physical mechanism remains relatively unexplored, and the conversion efficiency and bandwidth are extremely insufficient. Here, considering the instantaneous broadband characteristic, we establish a model of second harmonic generation (SHG) of low-temporal-coherent pulse, and reveal its differences from the coherent conditions. It is found that the second harmonic (SH) of low-temporal-coherent light is produced by not only the degenerate SH processes but also crossed sum-frequency processes. On the basis of this, we propose a method for realizing low-temporal-coherent SHG with high efficiency and broad bandwidth, and experimentally demonstrate a conversion efficiency up to 70% with a bandwidth of 3.1 THz (2.9 nm centered at 528 nm). To the best of our knowledge, this is the highest efficiency and broadest bandwidth of low-temporal-coherent SHG, and its efficiency is almost the same with that of the narrowband coherent condition. Furthermore, the spectral evolution characteristics of the broadband low-temporal-coherent pulse in SHG process are revealed in experiments, that the SH power spectral density (PSD) is proportional to the self-convolution of the fundamental wave PSD, which is greatly different from that of the coherent process. Our research opens a door for the study of the low-coherent nonlinear optical processes.

physics.optics↗

arXiv2019

Deep Learning-Based Video Coding: A Review and A Case Study

The past decade has witnessed great success of deep learning technology in many disciplines, especially in computer vision and image processing. However, deep learning-based video coding remains in its infancy. This paper reviews the representative works about using deep learning for image/video coding, which has been an actively developing research area since the year of 2015. We divide the related works into two categories: new coding schemes that are built primarily upon deep networks (deep schemes), and deep network-based coding tools (deep tools) that shall be used within traditional coding schemes or together with traditional coding tools. For deep schemes, pixel probability modeling and auto-encoder are the two approaches, that can be viewed as predictive coding scheme and transform coding scheme, respectively. For deep tools, there have been several proposed techniques using deep learning to perform intra-picture prediction, inter-picture prediction, cross-channel prediction, probability distribution prediction, transform, post- or in-loop filtering, down- and up-sampling, as well as encoding optimizations. In the hope of advocating the research of deep learning-based video coding, we present a case study of our developed prototype video codec, namely Deep Learning Video Coding (DLVC). DLVC features two deep tools that are both based on convolutional neural network (CNN), namely CNN-based in-loop filter (CNN-ILF) and CNN-based block adaptive resolution coding (CNN-BARC). Both tools help improve the compression efficiency by a significant margin. With the two deep tools as well as other non-deep coding tools, DLVC is able to achieve on average 39.6\% and 33.0\% bits saving than HEVC, under random-access and low-delay configurations, respectively. The source code of DLVC has been released for future researches.

cs.MM↗

arXiv2019

On The Classification-Distortion-Perception Tradeoff

Signal degradation is ubiquitous and computational restoration of degraded signal has been investigated for many years. Recently, it is reported that the capability of signal restoration is fundamentally limited by the perception-distortion tradeoff, i.e. the distortion and the perceptual difference between the restored signal and the ideal `original' signal cannot be made both minimal simultaneously. Distortion corresponds to signal fidelity and perceptual difference corresponds to perceptual naturalness, both of which are important metrics in practice. Besides, there is another dimension worthy of consideration, namely the semantic quality or the utility for recognition purpose, of the restored signal. In this paper, we extend the previous perception-distortion tradeoff to the case of classification-distortion-perception (CDP) tradeoff, where we introduced the classification error rate of the restored signal in addition to distortion and perceptual difference. Two versions of the CDP tradeoff are considered, one using a predefined classifier and the other dealing with the optimal classifier for the restored signal. For both versions, we can rigorously prove the existence of the CDP tradeoff, i.e. the distortion, perceptual difference, and classification error rate cannot be made all minimal simultaneously. Our findings can be useful especially for computer vision researches where some low-level vision tasks (signal restoration) serve for high-level vision tasks (visual understanding).

cs.IT↗

arXiv2019

High-Resolution Representations for Labeling Pixels and Regions

High-resolution representation learning plays an essential role in many vision problems, e.g., pose estimation and semantic segmentation. The high-resolution network (HRNet)~\cite{SunXLW19}, recently developed for human pose estimation, maintains high-resolution representations through the whole process by connecting high-to-low resolution convolutions in \emph{parallel} and produces strong high-resolution representations by repeatedly conducting fusions across parallel convolutions. In this paper, we conduct a further study on high-resolution representations by introducing a simple yet effective modification and apply it to a wide range of vision tasks. We augment the high-resolution representation by aggregating the (upsampled) representations from all the parallel convolutions rather than only the representation from the high-resolution convolution as done in~\cite{SunXLW19}. This simple modification leads to stronger representations, evidenced by superior results. We show top results in semantic segmentation on Cityscapes, LIP, and PASCAL Context, and facial landmark detection on AFLW, COFW, $300$W, and WFLW. In addition, we build a multi-level representation from the high-resolution representation and apply it to the Faster R-CNN object detection framework and the extended frameworks. The proposed approach achieves superior results to existing single-model networks on COCO object detection. The code and models have been publicly available at \url{https://github.com/HRNet}.

cs.CV↗

arXiv2019

Deep High-Resolution Representation Learning for Human Pose Estimation

This is an official pytorch implementation of Deep High-Resolution Representation Learning for Human Pose Estimation. In this work, we are interested in the human pose estimation problem with a focus on learning reliable high-resolution representations. Most existing methods recover high-resolution representations from low-resolution representations produced by a high-to-low resolution network. Instead, our proposed network maintains high-resolution representations through the whole process. We start from a high-resolution subnetwork as the first stage, gradually add high-to-low resolution subnetworks one by one to form more stages, and connect the mutli-resolution subnetworks in parallel. We conduct repeated multi-scale fusions such that each of the high-to-low resolution representations receives information from other parallel representations over and over, leading to rich high-resolution representations. As a result, the predicted keypoint heatmap is potentially more accurate and spatially more precise. We empirically demonstrate the effectiveness of our network through the superior pose estimation results over two benchmark datasets: the COCO keypoint detection dataset and the MPII Human Pose dataset. The code and models have been publicly available at \url{https://github.com/leoxiaobin/deep-high-resolution-net.pytorch}.

cs.CV↗

arXiv2018

Lattice-mismatched semiconductor heterostructures

Semiconductor heterostructure is a critical building block for modern semiconductor devices. However, forming semiconductor heterostructures of lattice-mismatch has been a great challenge for several decades. Epitaxial growth is infeasible to form abrupt heterostructures with large lattice-mismatch while mechanical-thermal bonding results in a high density of interface defects and therefore severely limits device applications. Here we show an ultra-thin oxide-interfaced approach for the successful formation of lattice-mismatched semiconductor heterostructures. Following the depiction of a theory on the role of interface oxide in forming the heterostructures, we describe experimental demonstrations of Ge/Si (diamond lattices), Si/GaAs (zinc blende lattice), GaAs/GaN (hexagon lattice), and Si/GaN heterostructures. Extraordinary electrical performances in terms of ideality factor, current on/off ratio, and reverse breakdown voltage are measured from p-n diodes fabricated from the four types of heterostructures, significantly outperforming diodes derived from other methods. Our demonstrations indicate the versatility of the ultra-thin-oxide-interface approach in forming lattice-mismatched heterostructures, open up a much larger possibility for material combinations for heterostructures, and pave the way toward broader applications in electronic and optoelectronic realms.

physics.app-ph↗

arXiv2018

Study of boiling heat transfer on the wall with an improved hybrid lattice Boltzmann model

In this paper, the energy non-conservation problem of the hybrid thermal LB phase change model is emphasized. To solve this problem, an infinite volume discrete scheme is introduced to deal with the diffusion term in the energy equation. After verified this model with energy conservation law in the simulation of the one-dimensional heat conduction across the liquid-vapor interface and the liquid film evaporation on the heating wall, the improved hybrid thermal LB model is then applied in the simulation of bubble nucleation and boiling heat transfer. With an innovative nucleation site treatment on the wall, the complete boiling stages as well as the boiling curve were successfully reproduced without adding any heating fluctuations. The bubble nucleation, growth, departure, and coalescence are well captured. Some basic features of boiling heat transfer including bubble nucleation site activation, bubble departure frequency, character of bubble shape at different regime, and horizontal movement of bubble during the boiling process were clearly observed in the numerical results.

physics.flu-dyn↗

arXiv2018

Entropy-regularized Optimal Transport Generative Models

We investigate the use of entropy-regularized optimal transport (EOT) cost in developing generative models to learn implicit distributions. Two generative models are proposed. One uses EOT cost directly in an one-shot optimization problem and the other uses EOT cost iteratively in an adversarial game. The proposed generative models show improved performance over contemporary models for image generation on MNSIT.

cs.LG↗

arXiv2018

Caching at the Wireless Edge: Design Aspects, Challenges and Future Directions

Caching at the wireless edge is a promising way of boosting spectral efficiency and reducing energy consumption of wireless systems. These improvements are rooted in the fact that popular contents are reused, asynchronously, by many users. In this article, we first introduce methods to predict the popularity distributions and user preferences, and the impact of erroneous information. We then discuss the two aspects of caching systems, namely content placement and delivery. We expound the key differences between wired and wireless caching, and outline the differences in the system arising from where the caching takes place, e.g., at base stations, or on the wireless devices themselves. Special attention is paid to the essential limitations in wireless caching, and possible tradeoffs between spectral efficiency, energy efficiency and cache size.

cs.NI↗

arXiv2018

Whittaker modules for the super-Virasoro algebras

In this paper, we define and study Whittaker modules for the super-Viraoro algebras, including the Neveu-Schwarz algebra and the Ramond algebra. We classify the simple Whittaker modules and obtain necessary and sufficient conditions for irreducibility of these modules.

math.RT↗