Searcharxiv⌕ Search

arXiv subjects

Yan Wu

Publications and source records attributed to Yan Wu.

At least 109 records · Page 6Linked to original sources
arXiv2021

Trilevel Neural Architecture Search for Efficient Single Image Super-Resolution

Modern solutions to the single image super-resolution (SISR) problem using deep neural networks aim not only at better performance accuracy but also at a lighter and computationally efficient model. To that end, recently, neural architecture search (NAS) approaches have shown some tremendous potential. Following the same underlying, in this paper, we suggest a novel trilevel NAS method that provides a better balance between different efficiency metrics and performance to solve SISR. Unlike available NAS, our search is more complete, and therefore it leads to an efficient, optimized, and compressed architecture. We innovatively introduce a trilevel search space modeling, i.e., hierarchical modeling on network-, cell-, and kernel-level structures. To make the search on trilevel spaces differentiable and efficient, we exploit a new sparsestmax technique that is excellent at generating sparse distributions of individual neural architecture candidates so that they can be better disentangled for the final selection from the enlarged search space. We further introduce the sorting technique to the sparsestmax relaxation for better network-level compression. The proposed NAS optimization additionally facilitates simultaneous search and training in a single phase, reducing search time and train time. Comprehensive evaluations on the benchmark datasets show our method's clear superiority over the state-of-the-art NAS in terms of a good trade-off between model size, performance, and efficiency.

cs.CV↗
arXiv2021

Site Mixing for Engineering Magnetic Topological Insulators

The van der Waals compound, MnBi$_2$Te$_4$, is the first intrinsic magnetic topological insulator, providing a materials platform for exploring exotic quantum phenomena such as the axion insulator state and the quantum anomalous Hall effect. However, intrinsic structural imperfections lead to bulk conductivity, and the roles of magnetic defects are still unknown. With higher concentrations of same types of magnetic defects, the isostructural compound MnSb$_2$Te$_4$ is a better model system for a systematic investigation of the connections among magnetic, topology and lattice defects. In this work, the impact of antisite defects on the magnetism and electronic structure is studied in MnSb$_2$Te$_4$. Mn-Sb site mixing leads to complex magnetic structures and tunes the interlayer magnetic coupling between antiferromagnetic and ferromagnetic. The detailed nonstoichiometry and site-mixing of MnSb$_2$Te$_4$ crystals depend on the growth parameters, which can lead to $\approx$40\% of Mn sites occupied by Sb and $\approx$15\% of Sb sites by Mn in as-grown crystals. Single crystal neutron diffraction and electron microscopy studies show nearly random distribution of the antisite defects. Band structure calculations suggest that the Mn-Sb site-mixing favors a FM interlayer coupling, consistent with experimental observation, but is detrimental to the band inversion required for a nontrivial topology. Our results suggest a long range magnetic order of Mn ions sitting on Bi sites in MnBi$_2$Te$_4$. The effects of site mixing should be considered in all layered heterostructures that consist of alternating magnetic and topological layers, including the entire family of MnTe(Bi$_2$Te$_3$)$_n$, its Sb analogs and their solid solution.

cond-mat.mtrl-sci↗
arXiv2021

6D Pose Estimation with Correlation Fusion

6D object pose estimation is widely applied in robotic tasks such as grasping and manipulation. Prior methods using RGB-only images are vulnerable to heavy occlusion and poor illumination, so it is important to complement them with depth information. However, existing methods using RGB-D data cannot adequately exploit consistent and complementary information between RGB and depth modalities. In this paper, we present a novel method to effectively consider the correlation within and across both modalities with attention mechanism to learn discriminative and compact multi-modal features. Then, effective fusion strategies for intra- and inter-correlation modules are explored to ensure efficient information flow between RGB and depth. To our best knowledge, this is the first work to explore effective intra- and inter-modality fusion in 6D pose estimation. The experimental results show that our method can achieve the state-of-the-art performance on LineMOD and YCB-Video dataset. We also demonstrate that the proposed method can benefit a real-world robot grasping task by providing accurate object pose estimation.

cs.CV↗
arXiv2021

Neural Architecture Search as Sparse Supernet

This paper aims at enlarging the problem of Neural Architecture Search (NAS) from Single-Path and Multi-Path Search to automated Mixed-Path Search. In particular, we model the NAS problem as a sparse supernet using a new continuous architecture representation with a mixture of sparsity constraints. The sparse supernet enables us to automatically achieve sparsely-mixed paths upon a compact set of nodes. To optimize the proposed sparse supernet, we exploit a hierarchical accelerated proximal gradient algorithm within a bi-level optimization framework. Extensive experiments on Convolutional Neural Network and Recurrent Neural Network search demonstrate that the proposed method is capable of searching for compact, general and powerful neural architectures.

cs.CV↗
arXiv2021

Radio frequency spectrum analyzer with a 5 THz bandwidth based on nonlinear optics in a CMOS compatible high-index doped silica waveguide

We report an all-optical radio-frequency (RF) spectrum analyzer with a bandwidth greater than 5 terahertz (THz), based on a 50-cm long spiral waveguide in a CMOS-compatible high-index doped silica platform. By carefully mapping out the dispersion profile of the waveguides for different thicknesses, we identify the optimal design to achieve near zero dispersion in the C-band. To demonstrate the capability of the RF spectrum analyzer, we measure the optical output of a femtosecond fiber laser with an ultrafast optical RF spectrum in the terahertz regime.

physics.optics↗
arXiv2021

Reentrance of spin-driven ferroelectricity through rotational tunneling of ammonium

Quantum effects fundamentally engender exotic physical phenomena in macroscopic systems, which advance next-generation technological applications. Rotational tunneling that represents the quantum phenomenon of the librational motion of molecules is ubiquitous in hydrogen-contained materials. However, its direct manifestation in realizing macroscopic physical properties is elusive. Here we report an observation of reentrant ferroelectricity under low pressure that is mediated by the rotational tunneling of ammonium ions in molecule-based (NH$_4$)$_2$FeCl$_5 \cdot$H$_2$O. Applying a small pressure leads to a transition from spin-driven ferroelectricity to paraelectricity coinciding with the stabilization of a collinear magnetic phase. Such a transition is attributed to the hydrogen bond fluctuations via the rotational tunneling of ammonium groups as supported by theoretical calculations. Higher pressure lifts the quantum fluctuations and leads to a reentrant ferroelectric phase concomitant with another incommensurate magnetic phase. These results demonstrate that the rotational tunneling emerges as a new route to control magnetic-related properties in soft magnets, opening avenues for designing multi-functional materials and realizing potential quantum control.

cond-mat.mtrl-sci↗
arXiv2020

Training Generative Adversarial Networks by Solving Ordinary Differential Equations

The instability of Generative Adversarial Network (GAN) training has frequently been attributed to gradient descent. Consequently, recent methods have aimed to tailor the models and training procedures to stabilise the discrete updates. In contrast, we study the continuous-time dynamics induced by GAN training. Both theory and toy experiments suggest that these dynamics are in fact surprisingly stable. From this perspective, we hypothesise that instabilities in training GANs arise from the integration error in discretising the continuous dynamics. We experimentally verify that well-known ODE solvers (such as Runge-Kutta) can stabilise training - when combined with a regulariser that controls the integration error. Our approach represents a radical departure from previous methods which typically use adaptive optimisation and stabilisation techniques that constrain the functional space (e.g. Spectral Normalisation). Evaluation on CIFAR-10 and ImageNet shows that our method outperforms several strong baselines, demonstrating its efficacy.

stat.ML↗
arXiv2020

Dense Dual-Path Network for Real-time Semantic Segmentation

Semantic segmentation has achieved remarkable results with high computational cost and a large number of parameters. However, real-world applications require efficient inference speed on embedded devices. Most previous works address the challenge by reducing depth, width and layer capacity of network, which leads to poor performance. In this paper, we introduce a novel Dense Dual-Path Network (DDPNet) for real-time semantic segmentation under resource constraints. We design a light-weight and powerful backbone with dense connectivity to facilitate feature reuse throughout the whole network and the proposed Dual-Path module (DPM) to sufficiently aggregate multi-scale contexts. Meanwhile, a simple and effective framework is built with a skip architecture utilizing the high-resolution feature maps to refine the segmentation output and an upsampling module leveraging context information from the feature maps to refine the heatmaps. The proposed DDPNet shows an obvious advantage in balancing accuracy and speed. Specifically, on Cityscapes test dataset, DDPNet achieves 75.3% mIoU with 52.6 FPS for an input of 1024 X 2048 resolution on a single GTX 1080Ti card. Compared with other state-of-the-art methods, DDPNet achieves a significant better accuracy with a comparable speed and fewer parameters.

cs.CV↗
arXiv2020

Asymptotic property C of the wreath product ZwrZ

Using the relationship between transfinite asymptotic dimension and asymptotic property C, we obtain that the wreath product Z wr Z has asymptotic property C. Specifically, we prove that the transfinite asymptotic dimension of the wreath product Z wr Z is no more than omega+ 1.

math.GR↗
arXiv2020

Multi-UAV Coverage Path Planning for the Inspection of Large and Complex Structures

We present a multi-UAV Coverage Path Planning (CPP) framework for the inspection of large-scale, complex 3D structures. In the proposed sampling-based coverage path planning method, we formulate the multi-UAV inspection applications as a multi-agent coverage path planning problem. By combining two NP-hard problems: Set Covering Problem (SCP) and Vehicle Routing Problem (VRP), a Set-Covering Vehicle Routing Problem (SC-VRP) is formulated and subsequently solved by a modified Biased Random Key Genetic Algorithm (BRKGA) with novel, efficient encoding strategies and local improvement heuristics. We test our proposed method for several complex 3D structures with the 3D model extracted from OpenStreetMap. The proposed method outperforms previous methods, by reducing the length of the planned inspection path by up to 48%

cs.RO↗
arXiv2020

LOGAN: Latent Optimisation for Generative Adversarial Networks

Training generative adversarial networks requires balancing of delicate adversarial dynamics. Even with careful tuning, training may diverge or end up in a bad equilibrium with dropped modes. In this work, we improve CS-GAN with natural gradient-based latent optimisation and show that it improves adversarial dynamics by enhancing interactions between the discriminator and the generator. Our experiments demonstrate that latent optimisation can significantly improve GAN training, obtaining state-of-the-art performance for the ImageNet ($128 \times 128$) dataset. Our model achieves an Inception Score (IS) of $148$ and an Fréchet Inception Distance (FID) of $3.4$, an improvement of $17\%$ and $32\%$ in IS and FID respectively, compared with the baseline BigGAN-deep model with the same architecture and number of parameters.

cs.LG↗
arXiv2020

Canted antiferromagnetic order in the monoaxial chiral magnets V$_{1/3}$TaS$_2$ and V$_{1/3}$NbS$_{2}$

The Dzyaloshinskii-Moriya (DM) interaction is present in the transition metal dichalcogenides (TMDC) magnets of form $M_{1/3}T$S$_{2}$ ($M$ $=$ 3d transition metal, $T$ $\in$ {Nb, Ta}), given that the intercalants $M$ form $\sqrt{3}\times\sqrt{3}$ superlattices within the structure of the parent materials $T$S$_2$ and break the centrosymmetry. Competition between the DM and ferromagnetic exchange interactions in these systems has been shown to stabilize a topological defect known as a chiral soliton in select intercalated TMDCs, initiating interest both in terms of fundamental physics and the potential for technological applications. In the current article, we report on our study of the materials V$_{1/3}$TaS$_2$ and V$_{1/3}$NbS$_2$, using a combination of x-ray powder diffraction, magnetization and single crystal neutron diffraction. Historically identified as ferromagnets, our diffraction results instead reveal that vanadium spins in these compounds are arranged into an A-type antiferromagnetic configuration at low temperatures. Refined moments are 1.37(6)$μ_{B}$ and 1.50(9)$μ_{B}$ for V$_{1/3}$TaS$_2$ and V$_{1/3}$NbS$_2$, respectively. Transition temperatures $T_{c}$~$=$~32K for V$_{1/3}$TaS$_{2}$ and 50K for V$_{1/3}$NbS$_{2}$ are obtained from the magnetization and neutron diffraction results. We attribute the small net magnetization observed in the low-temperature phases to a subtle ($\sim$2$^{\circ}$) canting of XY-spins in the out-of-plane direction. These new results are indicative of dominant antiferromagnetic exchange interactions between the vanadium moments in adjacent ab-planes, likely eliminating the possibility of identifying stable chiral solitons in the current materials.

cond-mat.mtrl-sci↗
arXiv2020

Examples of metric spaces with asymptotic property $C$

We construct a class of metric spaces whose transfinite asymptotic dimension and complementary-finite asymptotic dimension are both $ω+k$ for any $k\in\mathbb{N}$, where $ω$ is the smallest infinite ordinal number and a metric space whose transfinite asymptotic dimension and complementary-finite asymptotic dimension are both $2ω$. Moreover, we study the relationship between asymptotic dimension growth, transfinite asymptotic dimension and finite decomposition complexity.

math.FA↗
arXiv2020

Successive dielectric anomalies and magnetoelectric coupling in honeycomb Fe$_4$Nb$_2$O$_9$

By combining single crystal x-ray and neutron diffraction, and the magnetodielectric measurements on single crystal Fe4Nb2O9, we present the magnetic structure and the symmetry-allowed magnetoelectric coupling in Fe4Nb2O9. It undergoes an antiferromagnetic transition at TN=93 K, followed by a displacive transition at TS=70 K. The temperature-dependent dielectric constant of Fe4Nb2O9 is strongly anisotropic with the first anomaly at 93 K due to the exchange striction as a result of the long range spin order, and the second one at 70 K emanating from the structural phase transition primarily driven by the O atomic displacements. Magneticfield induced magnetoelectric coupling was observed in single crystal Fe4Nb2O9 and is compatible with the solved magnetic structure that is characteristic of antiferromagnetically arranged ferromagnetic chains in the honeycomb plane. We propose that such magnetic symmetry should be immune to external magnetic fields to some extent favored by the freedom of rotation of moments in the honeycomb plane, laying out a promising system to control the magnetoelectric properties by magnetic fields.

cond-mat.str-el↗
arXiv2020

Product Kanerva Machines: Factorized Bayesian Memory

An ideal cognitively-inspired memory system would compress and organize incoming items. The Kanerva Machine (Wu et al, 2018) is a Bayesian model that naturally implements online memory compression. However, the organization of the Kanerva Machine is limited by its use of a single Gaussian random matrix for storage. Here we introduce the Product Kanerva Machine, which dynamically combines many smaller Kanerva Machines. Its hierarchical structure provides a principled way to abstract invariant features and gives scaling and capacity advantages over single Kanerva Machines. We show that it can exhibit unsupervised clustering, find sparse and combinatorial allocation patterns, and discover spatial tunings that approximately factorize simple images by object.

cs.LG↗
arXiv2019

Efficient Robotic Task Generalization Using Deep Model Fusion Reinforcement Learning

Learning-based methods have been used to pro-gram robotic tasks in recent years. However, extensive training is usually required not only for the initial task learning but also for generalizing the learned model to the same task but in different environments. In this paper, we propose a novel Deep Reinforcement Learning algorithm for efficient task generalization and environment adaptation in the robotic task learning problem. The proposed method is able to efficiently generalize the previously learned task by model fusion to solve the environment adaptation problem. The proposed Deep Model Fusion (DMF) method reuses and combines the previously trained model to improve the learning efficiency and results.Besides, we also introduce a Multi-objective Guided Reward(MGR) shaping technique to further improve training efficiency.The proposed method was benchmarked with previous methods in various environments to validate its effectiveness.

cs.RO↗
arXiv2019

Exploring the "$L$--$σ$" relation of HII galaxies and giant extragalactic HII regions acting as standard candles

Cosmological applications of HII galaxies (HIIGx) and giant extragalactic HII regions (GEHR) to construct the Hubble diagram at higher redshifts require knowledge of the "$L$--$σ$" relation of the standard candles used. In this paper, we study the properties of a large sample of 156 sources (25 high-$z$ HII galaxies, 107 local HII galaxies, and 24 giant extragalactic HII regions) compiled by Terlevich et al.(2015). Using the the cosmological distances reconstructed through two new cosmology-independent methods, we investigate the correlation between the H$β$ emission-line luminosity $L$ and ionized-gas velocity dispersion $σ$. The method is based on non-parametric reconstruction using the measurements of Hubble parameters from cosmic clocks, as well as the simulated data of gravitational waves from the third-generation gravitational wave detector (the Einstein Telescope, ET), which can be considered as standard sirens. Assuming the emission-line luminosity versus ionized gas velocity dispersion relation, $\log L ($H$β) = α\log σ($H$β)+κ$, we find the full sample provides a tight constraint on the correlation parameters. However, similar analysis done on three different sub-samples seems to support the scheme of treating HII galaxies and giant extragalactic HII regions with distinct strategies. Using the corrected "$L$--$σ$" relation for the HII observational sample beyond the current reach of Type Ia supernovae, we obtain a value of the matter density parameter, $Ω_{m}=0.314\pm0.054$ (calibrated with standard clocks) and $Ω_{m}=0.311\pm0.049$ (calibrated with standard sirens), in the spatially flat $Λ$CDM cosmology.

astro-ph.CO↗