arXiv subjects

Hao Chen

Publications and source records attributed to Hao Chen.

At least 1,333 records · Page 74Linked to original sources

arXiv2020

Weakly Inscribed Polyhedra

We study convex polyhedra in $\mathbb{R}\mathbb{P}^3$ with all their vertices on a sphere. We do not require, in particular, that the polyhedra lie in the interior of the sphere, hence the term "weakly inscribed". Such polyhedra can be interpreted as ideal polyhedra, if we regard $\mathbb{R}\mathbb{P}^3$ as a combination of the hyperbolic space and the de Sitter space, with the sphere as the common ideal boundary. We have three main results: (1) the $1$-skeleta of weakly inscribed polyhedra are characterized in a purely combinatorial way, (2) the exterior dihedral angles are characterized by linear programming, and (3) we also describe the hyperbolic-de Sitter structure induced on the boundary of weakly inscribed polyhedra.

math.MG↗

arXiv2020

New Constructions of Subspace Codes Using Subsets of MRD codes in Several Blocks

A basic problem for the constant dimension subspace coding is to determine the maximal possible size A_q (n, d, k) of a set of k-dimensional subspaces in Fnq such that the subspace distance satisfies d(U, V )> or =d for any two different subspaces U andV in this set. We present two new constructions of constant dimension subspace codes using subsets of maximal rank-distance (MRD) codes in several blocks. This method is firstly applied to the linkage construction and secondly to arbitrary number of blocks of lifting MRD codes. In these two constructions, subsets of MRD codes with bounded ranks play an essential role. The Delsarte theorem of the rank distribution of MRD codes is an important ingredient to count codewords in our constructed constant dimension subspace codes. We give many new lower bounds for A_q (n, d, k). More than 110 new constant dimension subspace codes better than previously best known codes are constructed.

cs.IT↗

arXiv2019

Secondary Voltage Control of Microgrids Using Nonlinear Multiple Models Adaptive Control

This paper proposes a novel model-free secondary voltage control (SVC) for microgrids using nonlinear multiple models adaptive control. The proposed method is comprised of two components. Firstly, a linear robust adaptive controller is designed to guarantee the voltage stability in the bounded-input-bounded-output (BIBO) manner, which is more consistent with the operation requirements of microgrids. Secondly, a nonlinear adaptive controller is developed to improve the voltage tracking performance with the help of artificial neural networks (ANNs). A switching mechanism is proposed to coordinate such two controllers for guaranteeing the closed-loop stability while achieving accurate voltage tracking. Given our method leverages a data-driven real-time identification, it only relies on the input and output data of microgrids without resorting to any prior information of primary control and grid models, thus exhibiting good robustness, ease of deployment and disturbance rejection.

math.OC↗

arXiv2019

Blind Inpainting of Large-scale Masks of Thin Structures with Adversarial and Reinforcement Learning

Several imaging applications (vessels, retina, plant roots, road networks from satellites) require the accurate segmentation of thin structures for subsequent analysis. Discontinuities (gaps) in the extracted foreground may hinder down-stream image-based analysis of biomarkers, organ structure and topology. In this paper, we propose a general post-processing technique to recover such gaps in large-scale segmentation masks. We cast this problem as a blind inpainting task, where the regions of missing lines in the segmentation masks are not known to the algorithm, which we solve with an adversarially trained neural network. One challenge of using large images is the memory capacity of current GPUs. The typical approach of dividing a large image into smaller patches to train the network does not guarantee global coherence of the reconstructed image that preserves structure and topology. We use adversarial training and reinforcement learning (Policy Gradient) to endow the model with both global context and local details. We evaluate our method in several datasets in medical imaging, plant science, and remote sensing. Our experiments demonstrate that our model produces the most realistic and complete inpainted results, outperforming other approaches. In a dedicated study on plant roots we find that our approach is also comparable to human performance. Implementation available at \url{https://github.com/Hhhhhhhhhhao/Thin-Structure-Inpainting}.

cs.CV↗

arXiv2019

Auxiliary Learning for Deep Multi-task Learning

Multi-task learning (MTL) is an efficient solution to solve multiple tasks simultaneously in order to get better speed and performance than handling each single-task in turn. The most current methods can be categorized as either: (i) hard parameter sharing where a subset of the parameters is shared among tasks while other parameters are task-specific; or (ii) soft parameter sharing where all parameters are task-specific but they are jointly regularized. Both methods suffer from limitations: the shared hidden layers of the former are difficult to optimize due to the competing objectives while the complexity of the latter grows linearly with the increasing number of tasks. To mitigate those drawbacks, this paper proposes an alternative, where we explicitly construct an auxiliary module to mimic the soft parameter sharing for assisting the optimization of the hard parameter sharing layers in the training phase. In particular, the auxiliary module takes the outputs of the shared hidden layers as inputs and is supervised by the auxiliary task loss. During training, the auxiliary module is jointly optimized with the MTL network, serving as a regularization by introducing an inductive bias to the shared layers. In the testing phase, only the original MTL network is kept. Thus our method avoids the limitation of both categories. We evaluate the proposed auxiliary module on pixel-wise prediction tasks, including semantic segmentation, depth estimation, and surface normal prediction with different network structures. The extensive experiments over various settings verify the effectiveness of our methods.

cs.CV↗

arXiv2019

DirectPose: Direct End-to-End Multi-Person Pose Estimation

We propose the first direct end-to-end multi-person pose estimation framework, termed DirectPose. Inspired by recent anchor-free object detectors, which directly regress the two corners of target bounding-boxes, the proposed framework directly predicts instance-aware keypoints for all the instances from a raw input image, eliminating the need for heuristic grouping in bottom-up methods or bounding-box detection and RoI operations in top-down ones. We also propose a novel Keypoint Alignment (KPAlign) mechanism, which overcomes the main difficulty: lack of the alignment between the convolutional features and predictions in this end-to-end framework. KPAlign improves the framework's performance by a large margin while still keeping the framework end-to-end trainable. With the only postprocessing non-maximum suppression (NMS), our proposed framework can detect multi-person keypoints with or without bounding-boxes in a single shot. Experiments demonstrate that the end-to-end paradigm can achieve competitive or better performance than previous strong baselines, in both bottom-up and top-down methods. We hope that our end-to-end approach can provide a new perspective for the human pose estimation task.

cs.CV↗

arXiv2019

Decentralized Consensus Optimization Based on Parallel Random Walk

The alternating direction method of multipliers (ADMM) has recently been recognized as a promising approach for large-scale machine learning models. However, very few results study ADMM from the aspect of communication costs, especially jointly with running time. In this letter, we investigate the communication efficiency and running time of ADMM in solving the consensus optimization problem over decentralized networks. We first review the effort of random walk ADMM (W-ADMM), which reduces communication costs at the expense of running time. To accelerate the convergence speed of W-ADMM, we propose the parallel random walk ADMM (PW-ADMM) algorithm, where multiple random walks are active at the same time. Moreover, to further reduce the running time of PW-ADMM, the intelligent parallel random walk ADMM (IPW-ADMM) algorithm is proposed through integrating the \textit{Random Walk with Choice} with PW-ADMM. By numerical results from simulations, we demonstrate that the proposed algorithms can be both communication efficient and fast in running speed compared with state-of-the-art methods.

cs.DC↗

arXiv2019

Automatic Freeway Bottleneck Identification and Visualization using Image Processing Techniques

This paper develops an automatic freeway bottleneck identification and visualization algorithm using a combination of image processing techniques and traffic flow theory. Unlike previous studies that are based solely on loop detector data, the proposed method can use traffic measurements from various sensing technologies. Four steps are included in the proposed algorithm. First, the raw spatiotemporal speed data are transformed into binary matrices using image binarization techniques. Second, two post-processer filters are developed to clean the binary matrices by filtering scattered noise cells and localized congested regions. Subsequently, the roadway geometry information is used to remove the impact of acceleration zones downstream of bottlenecks and thus locate bottlenecks more precisely. Finally, the major characteristics of bottlenecks including activation and deactivation points, shockwave speeds and traffic delay caused by bottleneck are automatically extracted and visualized. The proposed algorithm is tested using loop detector data from I-5 demonstrating that the proposed method outperforms the state-of-the-art methods for congestion identification. The second test using INRIX data from I-66 demonstrates ability of the proposed algorithm to accurately extract and visualize bottleneck characteristics.

eess.IV↗

arXiv2019

Towards More Realistic Human-Robot Conversation: A Seq2Seq-based Body Gesture Interaction System

This paper presents a novel system that enables intelligent robots to exhibit realistic body gestures while communicating with humans. The proposed system consists of a listening model and a speaking model used in corresponding conversational phases. Both models are adapted from the sequence-to-sequence (seq2seq) architecture to synthesize body gestures represented by the movements of twelve upper-body keypoints. All the extracted 2D keypoints are firstly 3D-transformed, then rotated and normalized to discard irrelevant information. Substantial videos of human conversations from Youtube are collected and preprocessed to train the listening and speaking models separately, after which the two models are evaluated using metrics of mean squared error (MSE) and cosine similarity on the test dataset. The tuned system is implemented to drive a virtual avatar as well as Pepper, a physical humanoid robot, to demonstrate the improvement on conversational interaction abilities of our method in practice.

cs.CV↗

arXiv2019

Classification of Neurodevelopmental Age in Normal Infants Using 3D-CNN based on Brain MRI

Human brain development is rapid during infancy and early childhood. Many disease processes impair this development. Therefore, brain developmental age estimation (BDAE) is essential for all diseases affecting cognitive development. Brain magnetic resonance imaging (MRI) of infants shows brain growth and morphologic patterns during childhood. Therefore, we can estimate the developmental age from brain images. However, MRI analysis is time-consuming because each scan contains millions of data points (voxels). We investigated the three-dimensional convolutional neural network (3D CNN), a deep learning algorithm, to rapidly classify neurodevelopmental age with high accuracy based on MRIs. MRIs from normal newborns were obtained from the National Institute of Mental Health (NIMH) Data Archive. Age categories of pediatric MRIs were 3 wks + 1 wk, 1 yr + 2 wks, and 3 yrs + 4 wks. We trained a BDAE method using T1, T2, and proton density (PD) images from MRI scans of 112 individuals using 3D CNN. Compared with the known age, our method has a sensitivity of 99% and specificity of 98.3%. Moreover, our 3D CNN model has better performance in neurodevelopmental age estimation than does 2D CNN.

eess.IV↗

arXiv2019

Long Optimal or Small-Defect LRC Codes with Unbounded Minimum Distances

A code over a finite field is called locally recoverable code (LRC) if every coordinate symbol can be determined by a small number (at most r, this parameter is called locality) of other coordinate symbols. For a linear code with length n, dimension k and locality r, its minimum distance d satisfies a Singleton-like bound. A code attaining this bound is called optimal. Many families of optimal locally recoverable codes have been constructed by using different techniques in finite fields or algebraic curves. However no optimal LRC code over a general finite field of q elements with the length n around the square of q, the locality r larger than or equal to 24 and the minimum distance d larger than or equal to 9 has been constructed. In this paper for any given finite field of q elements, any given r between 1 and q-1 and given d in certain range, we give an optimal LRC code with length n around the square of q, locality r and minimum distance d. This is the only known family of optimal LRC codes with lengths n around the square of q and unbounded localities and minimum distances d larger than or equal to 9. We also gives an asymptotic bound for q-ary r locality LRC codes better than the previously konwn bound. Many long r-locality LRC codes with small defects are also constructed.

cs.IT↗

arXiv2019

Sending-or-Not-Sending with Independent Lasers: Secure Twin-Field Quantum Key Distribution Over 509 km

Twin field quantum key distribution promises high key rates at long distance to beat the rate distance limit. Here, applying the sending or not sending TF QKD protocol, we experimentally demonstrate a secure key distribution breaking the absolute key rate limit of repeaterless QKD over 509 km, 408 km ultra-low loss optical fibre and 350 km standard optical fibre. Two independent lasers are used as the source with remote frequency locking technique over 500 km fiber distance; Practical optical fibers are used as the optical path with appropriate noise filtering; And finite key effects are considered in the key rate analysis. The secure key rates obtained at different distances are more than 5 times higher than the conditional limit of repeaterless QKD, a bound value assuming the same detection loss in the comparison. The achieved secure key rate is also higher than that a traditional QKD protocol running with a perfect repeaterless QKD device and even if an infinite number of sent pulses. Our result shows that the protocol and technologies applied in this experiment enable TF QKD to achieve high secure key rate at long distribution distance, and hence practically useful for field implementation of intercity QKD.

quant-ph↗

arXiv2019

Matryoshka: Fuzzing Deeply Nested Branches

Greybox fuzzing has made impressive progress in recent years, evolving from heuristics-based random mutation to approaches for solving individual path constraints. However, they have difficulty solving path constraints that involve deeply nested conditional statements, which are common in image and video decoders, network packet analyzers, and checksum tools. We propose an approach for addressing this problem. First, we identify all the control flow-dependent conditional statements of the target conditional statement. Next, we select the data flow-dependent conditional statements. Finally, we use three strategies to find an input that satisfies all conditional statements simultaneously. We implemented this approach in a tool called Matryoshka and compared its effectiveness on 13 open source programs against other state-of-the-art fuzzers. Matryoshka found significantly more unique crashes than AFL, QSYM, and Angora. We manually classified those crashes into 41 unique new bugs, and obtained 12 CVEs. Our evaluation also uncovered the key technique contributing to Matryoshka's impressive performance: it collects only the nesting constraints that may cause the target conditional statements unreachable, which greatly simplifies the constraints that it has to solve.

cs.CR↗

arXiv2019

CNN-based RGB-D Salient Object Detection: Learn, Select and Fuse

The goal of this work is to present a systematic solution for RGB-D salient object detection, which addresses the following three aspects with a unified framework: modal-specific representation learning, complementary cue selection and cross-modal complement fusion. To learn discriminative modal-specific features, we propose a hierarchical cross-modal distillation scheme, in which the well-learned source modality provides supervisory signals to facilitate the learning process for the new modality. To better extract the complementary cues, we formulate a residual function to incorporate complements from the paired modality adaptively. Furthermore, a top-down fusion structure is constructed for sufficient cross-modal interactions and cross-level transmissions. The experimental results demonstrate the effectiveness of the proposed cross-modal distillation scheme in zero-shot saliency detection and pre-training on a new modality, as well as the advantages in selecting and fusing cross-modal/cross-level complements.

cs.CV↗

arXiv2019

On a conjecture about a class of permutation quadrinomials

Very recently, Tu et al. presented a sufficient condition about $(a_1,a_2,a_3)$, see Theorem 1.1, such that $f(x) = x^{3\cdot 2^m} + a_1 x^{2^{m+1}+1}+ a_2 x^{2^m+2} + a_3 x^3$ is a class of permutation polynomials over $\gf_{2^{n}}$ with $n=2m$ and $m$ odd. In this present paper, we prove that the sufficient condition is also necessary.

cs.IT↗

arXiv2019

XONN: XNOR-based Oblivious Deep Neural Network Inference

Advancements in deep learning enable cloud servers to provide inference-as-a-service for clients. In this scenario, clients send their raw data to the server to run the deep learning model and send back the results. One standing challenge in this setting is to ensure the privacy of the clients' sensitive data. Oblivious inference is the task of running the neural network on the client's input without disclosing the input or the result to the server. This paper introduces XONN, a novel end-to-end framework based on Yao's Garbled Circuits (GC) protocol, that provides a paradigm shift in the conceptual and practical realization of oblivious inference. In XONN, the costly matrix-multiplication operations of the deep learning model are replaced with XNOR operations that are essentially free in GC. We further provide a novel algorithm that customizes the neural network such that the runtime of the GC protocol is minimized without sacrificing the inference accuracy. We design a user-friendly high-level API for XONN, allowing expression of the deep learning model architecture in an unprecedented level of abstraction. Extensive proof-of-concept evaluation on various neural network architectures demonstrates that XONN outperforms prior art such as Gazelle (USENIX Security'18) by up to 7x, MiniONN (ACM CCS'17) by 93x, and SecureML (IEEE S&P'17) by 37x. State-of-the-art frameworks require one round of interaction between the client and the server for each layer of the neural network, whereas, XONN requires a constant round of interactions for any number of layers in the model. XONN is first to perform oblivious inference on Fitnet architectures with up to 21 layers, suggesting a new level of scalability compared with state-of-the-art. Moreover, we evaluate XONN on four datasets to perform privacy-preserving medical diagnosis.

cs.CR↗

arXiv2019

Existence of the tetragonal and rhombohedral deformation families of the gyroid

We provide an existence proof for two 1-parameter families of embedded triply periodic minimal surfaces of genus three, namely the tG family with tetragonal symmetry that contains the gyroid, and the rGL family with rhombohedral symmetry that contains the gyroid and the Lidinoid, both discovered numerically in the 1990s. The existence was previously proved within a neighborhood of the gyroid and the Lidinoid, using Weierstrass data defined on branched rectangular tori. Our main contribution is to extend the technique to branched tori that are not necessarily rectangular.

math.DG↗

arXiv2019

An Active Learning Approach for Reducing Annotation Cost in Skin Lesion Analysis

Automated skin lesion analysis is very crucial in clinical practice, as skin cancer is among the most common human malignancy. Existing approaches with deep learning have achieved remarkable performance on this challenging task, however, heavily relying on large-scale labelled datasets. In this paper, we present a novel active learning framework for cost-effective skin lesion analysis. The goal is to effectively select and utilize much fewer labelled samples, while the network can still achieve state-of-the-art performance. Our sample selection criteria complementarily consider both informativeness and representativeness, derived from decoupled aspects of measuring model certainty and covering sample diversity. To make wise use of the selected samples, we further design a simple yet effective strategy to aggregate intra-class images in pixel space, as a new form of data augmentation. We validate our proposed method on data of ISIC 2017 Skin Lesion Classification Challenge for two tasks. Using only up to 50% of samples, our approach can achieve state-of-the-art performances on both tasks, which are comparable or exceeding the accuracies with full-data training, and outperform other well-known active learning methods by a large margin.

cs.CV↗