arXiv subjects

Jie Hu

Publications and source records attributed to Jie Hu.

At least 217 records · Page 12Linked to original sources

arXiv2021

Properly colored short cycles in edge-colored graphs

Properly colored cycles in edge-colored graphs are closely related to directed cycles in oriented graphs. As an analogy of the well-known Caccetta-Häggkvist Conjecture, we study the existence of properly colored cycles of bounded length in an edge-colored graph. We first prove that for all integers $s$ and $t$ with $t\geq s\geq2$, every edge-colored graph $G$ with no properly colored $K_{s,t}$ contains a spanning subgraph $H$ which admits an orientation $D$ such that every directed cycle in $D$ is a properly colored cycle in $G$. Using this result, we show that for $r\geq4$, if the Caccetta-Häggkvist Conjecture holds , then every edge-colored graph of order $n$ with minimum color degree at least $n/r+2\sqrt{n}+1$ contains a properly colored cycle of length at most $r$. In addition, we also obtain an asymptotically tight total color degree condition which ensures a properly colored (or rainbow) $K_{s,t}$.

math.CO↗

arXiv2021

Unifying Nonlocal Blocks for Neural Networks

The nonlocal-based blocks are designed for capturing long-range spatial-temporal dependencies in computer vision tasks. Although having shown excellent performance, they still lack the mechanism to encode the rich, structured information among elements in an image or video. In this paper, to theoretically analyze the property of these nonlocal-based blocks, we provide a new perspective to interpret them, where we view them as a set of graph filters generated on a fully-connected graph. Specifically, when choosing the Chebyshev graph filter, a unified formulation can be derived for explaining and analyzing the existing nonlocal-based blocks (e.g., nonlocal block, nonlocal stage, double attention block). Furthermore, by concerning the property of spectral, we propose an efficient and robust spectral nonlocal block, which can be more robust and flexible to catch long-range dependencies when inserted into deep neural networks than the existing nonlocal blocks. Experimental results demonstrate the clear-cut improvements and practical applicabilities of our method on image classification, action recognition, semantic segmentation, and person re-identification tasks.

cs.CV↗

arXiv2021

Predicting Quantum Potentials by Deep Neural Network and Metropolis Sampling

The hybridizations of machine learning and quantum physics have caused essential impacts to the methodology in both fields. Inspired by quantum potential neural network, we here propose to solve the potential in the Schrodinger equation provided the eigenstate, by combining Metropolis sampling with deep neural network, which we dub as Metropolis potential neural network (MPNN). A loss function is proposed to explicitly involve the energy in the optimization for its accurate evaluation. Benchmarking on the harmonic oscillator and hydrogen atom, MPNN shows excellent accuracy and stability on predicting not just the potential to satisfy the Schrodinger equation, but also the eigen-energy. Our proposal could be potentially applied to the ab-initio simulations, and to inversely solving other partial differential equations in physics and beyond.

quant-ph↗

arXiv2021

Scalable and Adaptive Graph Neural Networks with Self-Label-Enhanced training

It is hard to directly implement Graph Neural Networks (GNNs) on large scaled graphs. Besides of existed neighbor sampling techniques, scalable methods decoupling graph convolutions and other learnable transformations into preprocessing and post classifier allow normal minibatch training. By replacing redundant concatenation operation with attention mechanism in SIGN, we propose Scalable and Adaptive Graph Neural Networks (SAGN). SAGN can adaptively gather neighborhood information among different hops. To further improve scalable models on semi-supervised learning tasks, we propose Self-Label-Enhance (SLE) framework combining self-training approach and label propagation in depth. We add base model with a scalable node label module. Then we iteratively train models and enhance train set in several stages. To generate input of node label module, we directly apply label propagation based on one-hot encoded label vectors without inner random masking. We find out that empirically the label leakage has been effectively alleviated after graph convolutions. The hard pseudo labels in enhanced train set participate in label propagation with true labels. Experiments on both inductive and transductive datasets demonstrate that, compared with other sampling-based and sampling-free methods, SAGN achieves better or comparable results and SLE can further improve performance.

cs.LG↗

arXiv2021

Graph Inference Representation: Learning Graph Positional Embeddings with Anchor Path Encoding

Learning node representations that incorporate information from graph structure benefits wide range of tasks on graph. The majority of existing graph neural networks (GNNs) have limited power in capturing position information for a given node. The idea of positioning nodes with selected anchors has been exploited, yet mainly relying on explicit labeling of distance information. Here we propose Graph Inference Representation (GIR), an anchor based GNN model encoding path information related to pre-selected anchors for each node. Abilities to get position-aware embeddings are theoretically and experimentally investigated on GIR and its core variants. Further, the complementarity between GIRs and typical GNNs is demonstrated. We show that GIRs get outperformed results in position-aware scenarios, and performances on typical GNNs could be improved by fusing GIR embeddings.

cs.LG↗

arXiv2021

Data-Driven Design-by-Analogy: State of the Art and Future Directions

Design-by-Analogy (DbA) is a design methodology wherein new solutions, opportunities or designs are generated in a target domain based on inspiration drawn from a source domain; it can benefit designers in mitigating design fixation and improving design ideation outcomes. Recently, the increasingly available design databases and rapidly advancing data science and artificial intelligence technologies have presented new opportunities for developing data-driven methods and tools for DbA support. In this study, we survey existing data-driven DbA studies and categorize individual studies according to the data, methods, and applications in four categories, namely, analogy encoding, retrieval, mapping, and evaluation. Based on both nuanced organic review and structured analysis, this paper elucidates the state of the art of data-driven DbA research to date and benchmarks it with the frontier of data science and AI research to identify promising research opportunities and directions for the field. Finally, we propose a future conceptual data-driven DbA system that integrates all propositions.

cs.AI↗

arXiv2021

ISTR: End-to-End Instance Segmentation with Transformers

End-to-end paradigms significantly improve the accuracy of various deep-learning-based computer vision models. To this end, tasks like object detection have been upgraded by replacing non-end-to-end components, such as removing non-maximum suppression by training with a set loss based on bipartite matching. However, such an upgrade is not applicable to instance segmentation, due to its significantly higher output dimensions compared to object detection. In this paper, we propose an instance segmentation Transformer, termed ISTR, which is the first end-to-end framework of its kind. ISTR predicts low-dimensional mask embeddings, and matches them with ground truth mask embeddings for the set loss. Besides, ISTR concurrently conducts detection and segmentation with a recurrent refinement strategy, which provides a new way to achieve instance segmentation compared to the existing top-down and bottom-up frameworks. Benefiting from the proposed end-to-end mechanism, ISTR demonstrates state-of-the-art performance even with approximation-based suboptimal embeddings. Specifically, ISTR obtains a 46.8/38.6 box/mask AP using ResNet50-FPN, and a 48.1/39.9 box/mask AP using ResNet101-FPN, on the MS COCO dataset. Quantitative and qualitative results reveal the promising potential of ISTR as a solid baseline for instance-level recognition. Code has been made available at: https://github.com/hujiecpp/ISTR.

cs.CV↗

arXiv2021

Learning the Superpixel in a Non-iterative and Lifelong Manner

Superpixel is generated by automatically clustering pixels in an image into hundreds of compact partitions, which is widely used to perceive the object contours for its excellent contour adherence. Although some works use the Convolution Neural Network (CNN) to generate high-quality superpixel, we challenge the design principles of these networks, specifically for their dependence on manual labels and excess computation resources, which limits their flexibility compared with the traditional unsupervised segmentation methods. We target at redefining the CNN-based superpixel segmentation as a lifelong clustering task and propose an unsupervised CNN-based method called LNS-Net. The LNS-Net can learn superpixel in a non-iterative and lifelong manner without any manual labels. Specifically, a lightweight feature embedder is proposed for LNS-Net to efficiently generate the cluster-friendly features. With those features, seed nodes can be automatically assigned to cluster pixels in a non-iterative way. Additionally, our LNS-Net can adapt the sequentially lifelong learning by rescaling the gradient of weight based on both channel and spatial context to avoid overfitting. Experiments show that the proposed LNS-Net achieves significantly better performance on three benchmarks with nearly ten times lower complexity compared with other state-of-the-art methods.

cs.CV↗

arXiv2021

Involution: Inverting the Inherence of Convolution for Visual Recognition

Convolution has been the core ingredient of modern neural networks, triggering the surge of deep learning in vision. In this work, we rethink the inherent principles of standard convolution for vision tasks, specifically spatial-agnostic and channel-specific. Instead, we present a novel atomic operation for deep neural networks by inverting the aforementioned design principles of convolution, coined as involution. We additionally demystify the recent popular self-attention operator and subsume it into our involution family as an over-complicated instantiation. The proposed involution operator could be leveraged as fundamental bricks to build the new generation of neural networks for visual recognition, powering different deep learning models on several prevalent benchmarks, including ImageNet classification, COCO detection and segmentation, together with Cityscapes segmentation. Our involution-based models improve the performance of convolutional baselines using ResNet-50 by up to 1.6% top-1 accuracy, 2.5% and 2.4% bounding box AP, and 4.7% mean IoU absolutely while compressing the computational cost to 66%, 65%, 72%, and 57% on the above benchmarks, respectively. Code and pre-trained models for all the tasks are available at https://github.com/d-li14/involution.

cs.CV↗

arXiv2021

Architecture Disentanglement for Deep Neural Networks

Understanding the inner workings of deep neural networks (DNNs) is essential to provide trustworthy artificial intelligence techniques for practical applications. Existing studies typically involve linking semantic concepts to units or layers of DNNs, but fail to explain the inference process. In this paper, we introduce neural architecture disentanglement (NAD) to fill the gap. Specifically, NAD learns to disentangle a pre-trained DNN into sub-architectures according to independent tasks, forming information flows that describe the inference processes. We investigate whether, where, and how the disentanglement occurs through experiments conducted with handcrafted and automatically-searched network architectures, on both object-based and scene-based datasets. Based on the experimental results, we present three new findings that provide fresh insights into the inner logic of DNNs. First, DNNs can be divided into sub-architectures for independent tasks. Second, deeper layers do not always correspond to higher semantics. Third, the connection type in a DNN affects how the information flows across layers, leading to different disentanglement behaviors. With NAD, we further explain why DNNs sometimes give wrong predictions. Experimental results show that misclassified images have a high probability of being assigned to task sub-architectures similar to the correct ones. Code will be available at: https://github.com/hujiecpp/NAD.

cs.CV↗

arXiv2021

Image-to-image Translation via Hierarchical Style Disentanglement

Recently, image-to-image translation has made significant progress in achieving both multi-label (\ie, translation conditioned on different labels) and multi-style (\ie, generation with diverse styles) tasks. However, due to the unexplored independence and exclusiveness in the labels, existing endeavors are defeated by involving uncontrolled manipulations to the translation results. In this paper, we propose Hierarchical Style Disentanglement (HiSD) to address this issue. Specifically, we organize the labels into a hierarchical tree structure, in which independent tags, exclusive attributes, and disentangled styles are allocated from top to bottom. Correspondingly, a new translation process is designed to adapt the above structure, in which the styles are identified for controllable translations. Both qualitative and quantitative results on the CelebA-HQ dataset verify the ability of the proposed HiSD. We hope our method will serve as a solid baseline and provide fresh insights with the hierarchically organized annotations for future research in image-to-image translation. The code has been released at https://github.com/imlixinyang/HiSD.

cs.CV↗

arXiv2021

Performance Analysis and Improvement on DSRC Application for V2V Communication

In this paper, we focus on the performance of vehicle-to-vehicle (V2V) communication adopting the Dedicated Short Range Communication (DSRC) application in periodic broadcast mode. An analytical model is studied and a fixed point method is used to analyze the packet delivery ratio (PDR) and mean delay based on the IEEE 802.11p standard in a fully connected network under the assumption of perfect PHY performance. With the characteristics of V2V communication, we develop the Semi-persistent Contention Density Control (SpCDC) scheme to improve the DSRC performance. We use Monte Carlo simulation to verify the results obtained by the analytical model. The simulation results show that the packet delivery ratio in SpCDC scheme increases more than 10% compared with IEEE 802.11p in heavy vehicle load scenarios. Meanwhile, the mean reception delay decreases more than 50%, which provides more reliable road safety.

cs.NI↗

arXiv2020

Manipulating Giant Rashba Valley Splitting and Quantum Hall States in Few-Layer Black Arsenic by Electrostatic Gating

Exciting phenomena may emerge in non-centrosymmetric two-dimensional (2D) electronic systems when spin-orbit coupling (SOC) interplays dynamically with Coulomb interactions, band topology, and external modulating forces, etc. Here, we report illuminating synergetic effects between SOC and Stark in centrosymmetric few-layer black arsenic (BAs), manifested as giant Rashba valley splitting and exotic quantum Hall states (QHS) reversibly controlled by electrostatic gating. The unusual finding is rooted in the puckering square lattice of BAs, in which heavy $4p$ orbitals form highly asymmetric $Γ$ valley with the $p_{z}$ symmetry and $D$ valleys of the $p_{x}$ origin, located at the Brillouin zone (BZ) center and near the time reversal invariant momenta of $X$, respectively. When the structure inversion symmetry is broken by perpendicular electric field, giant Rashba SOC is activated for the $p_{x}$ bands to produce strong spin-polarized $D^{+}$ and $D^{-}$ valleys related by time-reversal symmetry, coexisting with weak $Γ$ Rashba bands constrained by the $p_{z}$ symmetry. Intriguingly, strong Stark effect shows the same $p_{x}$-orbital selectiveness for $D$, collectively shifting the valence band maximum of $D^{\pm}$ valleys to exceed the $Γ$ pockets. Such an orchestrating effect between SOC and Stark allows us to realize gate-tunable spin valley manipulations for 2D hole gas, as revealed by unconventional magnetic field triggered even-to-odd transitions in QHS. For electron doping, the quantization of the $Γ$ Rashba bands is characterized by peculiar density-dependent transitions in band topology from two parabolic valleys to a unique inner-outer helical structure when charge carrier concentrations increase.

cond-mat.mes-hall↗

arXiv2020

A Survey on Machine Reading Comprehension: Tasks, Evaluation Metrics and Benchmark Datasets

Machine Reading Comprehension (MRC) is a challenging Natural Language Processing(NLP) research field with wide real-world applications. The great progress of this field in recent years is mainly due to the emergence of large-scale datasets and deep learning. At present, a lot of MRC models have already surpassed human performance on various benchmark datasets despite the obvious giant gap between existing MRC models and genuine human-level reading comprehension. This shows the need for improving existing datasets, evaluation metrics, and models to move current MRC models toward "real" understanding. To address the current lack of comprehensive survey of existing MRC tasks, evaluation metrics, and datasets, herein, (1) we analyze 57 MRC tasks and datasets and propose a more precise classification method of MRC tasks with 4 different attributes; (2) we summarized 9 evaluation metrics of MRC tasks, 7 attributes and 10 characteristics of MRC datasets; (3) We also discuss key open issues in MRC research and highlighted future research directions. In addition, we have collected, organized, and published our data on the companion website(https://mrc-datasets.github.io/) where MRC researchers could directly access each MRC dataset, papers, baseline projects, and the leaderboard.

cs.CL↗

arXiv2020

Gap Probabilities in the Laguerre Unitary Ensemble and Discrete Painlevé Equations

In this paper we study a certain recurrence relation, that can be used to generate ladder operators for the Laguerre Unitary ensemble, from the point of view of Sakai's geometric theory of Painlevé equations. On one hand, this gives us one more detailed example of the appearance of discrete Painlevé equations in the theory of orthogonal polynomials. On the other hand, it serves as a good illustration of the effectiveness of a recently proposed procedure on how to reduce such recurrences to some canonical discrete Painlevé equations.

nlin.SI↗

arXiv2020

The Steiner $k$-eccentricity on trees

We study the Steiner $k$-eccentricity on trees, which generalizes the previous one in the paper [X.~Li, G.~Yu, S.~Klavžar, On the average Steiner 3-eccentricity of trees, arXiv:2005.10319, 2020]. To support the algorithm, we achieve much stronger properties for the Steiner $k$-ecc tree than that in the previous paper. Based on this, a linear time algorithm is devised to calculate the Steiner $k$-eccentricity of a vertex in a tree. On the other hand, the lower and upper bounds of the average Steiner $k$-eccentricity index of a tree on order $n$ are established based on a novel technique which is quite different from that in the previous paper but much easier to follow.

math.CO↗

arXiv2020

Energy Self-Sustainability in Full-Spectrum 6G

Full-spectrum ranging from sub 6 GHz to THz and visible light will be exploited in 6G in order to reach unprecedented key-performance-indicators (KPIs). However, extraordinary amount of energy will be consumed by network infrastructure, while functions of massively deployed Internet of Everything (IoE) devices are limited by embedded batteries. Therefore, energy self-sustainable 6G is proposed in this article. First of all, it may achieve network-wide energy efficiency by exploiting cell-free and airborne access networks as well as by implementing intelligent holographic environments. Secondly, by exploiting radio-frequency/visible light signals for providing on-demand wireless power transfer (WPT) and for enabling passive backscatter communication, ``zero-energy'' devices may become a reality. Furthermore, IoE devices actively adapt their transceivers for better performance to a dynamic environment. This article aims to provide a first glance at primary designing principles of energy self-sustainable 6G.

cs.NI↗

arXiv2020

A Convolutional Neural Network-based Patent Image Retrieval Method for Design Ideation

The patent database is often used in searches of inspirational stimuli for innovative design opportunities because of its large size, extensive variety and rich design information in patent documents. However, most patent mining research only focuses on textual information and ignores visual information. Herein, we propose a convolutional neural network (CNN)-based patent image retrieval method. The core of this approach is a novel neural network architecture named Dual-VGG that is aimed to accomplish two tasks: visual material type prediction and international patent classification (IPC) class label prediction. In turn, the trained neural network provides the deep features in the image embedding vectors that can be utilized for patent image retrieval and visual mapping. The accuracy of both training tasks and patent image embedding space are evaluated to show the performance of our model. This approach is also illustrated in a case study of robot arm design retrieval. Compared to traditional keyword-based searching and Google image searching, the proposed method discovers more useful visual information for engineering design.

cs.CV↗