arXiv subjects

Fei Ding

Publications and source records attributed to Fei Ding.

At least 91 records · Page 5Linked to original sources

arXiv2022

XMP-Font: Self-Supervised Cross-Modality Pre-training for Few-Shot Font Generation

Generating a new font library is a very labor-intensive and time-consuming job for glyph-rich scripts. Few-shot font generation is thus required, as it requires only a few glyph references without fine-tuning during test. Existing methods follow the style-content disentanglement paradigm and expect novel fonts to be produced by combining the style codes of the reference glyphs and the content representations of the source. However, these few-shot font generation methods either fail to capture content-independent style representations, or employ localized component-wise style representations, which is insufficient to model many Chinese font styles that involve hyper-component features such as inter-component spacing and "connected-stroke". To resolve these drawbacks and make the style representations more reliable, we propose a self-supervised cross-modality pre-training strategy and a cross-modality transformer-based encoder that is conditioned jointly on the glyph image and the corresponding stroke labels. The cross-modality encoder is pre-trained in a self-supervised manner to allow effective capture of cross- and intra-modality correlations, which facilitates the content-style disentanglement and modeling style representations of all scales (stroke-level, component-level and character-level). The pre-trained encoder is then applied to the downstream font generation task without fine-tuning. Experimental comparisons of our method with state-of-the-art methods demonstrate our method successfully transfers styles of all scales. In addition, it only requires one reference glyph and achieves the lowest rate of bad cases in the few-shot font generation task 28% lower than the second best

cs.CV↗

arXiv2022

Decoupled IoU Regression for Object Detection

Non-maximum suppression (NMS) is widely used in object detection pipelines for removing duplicated bounding boxes. The inconsistency between the confidence for NMS and the real localization confidence seriously affects detection performance. Prior works propose to predict Intersection-over-Union (IoU) between bounding boxes and corresponding ground-truths to improve NMS, while accurately predicting IoU is still a challenging problem. We argue that the complex definition of IoU and feature misalignment make it difficult to predict IoU accurately. In this paper, we propose a novel Decoupled IoU Regression (DIR) model to handle these problems. The proposed DIR decouples the traditional localization confidence metric IoU into two new metrics, Purity and Integrity. Purity reflects the proportion of the object area in the detected bounding box, and Integrity refers to the completeness of the detected object area. Separately predicting Purity and Integrity can divide the complex mapping between the bounding box and its IoU into two clearer mappings and model them independently. In addition, a simple but effective feature realignment approach is also introduced to make the IoU regressor work in a hindsight manner, which can make the target mapping more stable. The proposed DIR can be conveniently integrated with existing two-stage detectors and significantly improve their performance. Through a simple implementation of DIR with HTC, we obtain 51.3% AP on MS COCO benchmark, which outperforms previous methods and achieves state-of-the-art.

cs.CV↗

arXiv2021

Multi-level Knowledge Distillation via Knowledge Alignment and Correlation

Knowledge distillation (KD) has become an important technique for model compression and knowledge transfer. In this work, we first perform a comprehensive analysis of the knowledge transferred by different KD methods. We demonstrate that traditional KD methods, which minimize the KL divergence of softmax outputs between networks, are related to the knowledge alignment of an individual sample only. Meanwhile, recent contrastive learning-based KD methods mainly transfer relational knowledge between different samples, namely, knowledge correlation. While it is important to transfer the full knowledge from teacher to student, we introduce the Multi-level Knowledge Distillation (MLKD) by effectively considering both knowledge alignment and correlation. MLKD is task-agnostic and model-agnostic, and can easily transfer knowledge from supervised or self-supervised pretrained teachers. We show that MLKD can improve the reliability and transferability of learned representations. Experiments demonstrate that MLKD outperforms other state-of-the-art methods on a large number of experimental settings including different (a) pretraining strategies (b) network architectures (c) datasets (d) tasks.

cs.CV↗

arXiv2021

Multi-shot Temporal Event Localization: a Benchmark

Current developments in temporal event or action localization usually target actions captured by a single camera. However, extensive events or actions in the wild may be captured as a sequence of shots by multiple cameras at different positions. In this paper, we propose a new and challenging task called multi-shot temporal event localization, and accordingly, collect a large scale dataset called MUlti-Shot EventS (MUSES). MUSES has 31,477 event instances for a total of 716 video hours. The core nature of MUSES is the frequent shot cuts, for an average of 19 shots per instance and 176 shots per video, which induces large intrainstance variations. Our comprehensive evaluations show that the state-of-the-art method in temporal action localization only achieves an mAP of 13.1% at IoU=0.5. As a minor contribution, we present a simple baseline approach for handling the intra-instance variations, which reports an mAP of 18.9% on MUSES and 56.9% on THUMOS14 at IoU=0.5. To facilitate research in this direction, we release the dataset and the project code at https://songbai.site/muses/ .

cs.CV↗

arXiv2021

Electrical Tuning of Fresnel Lens in Reflection

Optical metasurfaces have been extensively investigated, demonstrating diverse and multiple functionalities with complete control over the transmitted and reflected fields. Most optical metasurfaces are however static, with only a few configurations offering (rather limited) electrical control, thereby jeopardizing their application prospects in emerging flat optics technologies. Here, we suggest an approach to realize electrically tunable optical metasurfaces, demonstrating dynamic Fresnel lens focusing. The active Fresnel lens (AFL) exploits the electro-optic Pockels effect in a 300-nm-thick lithium niobate layer sandwiched between a continuous thick and nanostructured gold film serving as electrodes. We fabricate and characterize the AFL, focusing 800-900 nm radiation at the distance of 40 $\mathrmμ$m with the focusing efficiency of 15 % and demonstrating the modulation depth of 1.5 % with the driving voltage of $\pm 10$ V within the bandwidth of $\sim\! 4$ MHz. We believe that the electro-optic metasurface concept introduced is useful for designing dynamic flat optics components.

physics.optics↗

arXiv2021

Double cycle-consistent generative adversarial network for unsupervised conditional generation

Conditional generative models have achieved considerable success in the past few years, but usually require a lot of labeled data. Recently, ClusterGAN combines GAN with an encoder to achieve remarkable clustering performance via unsupervised conditional generation. However, it ignores the real conditional distribution of data, which leads to generating less diverse samples for each class and makes the encoder only achieve sub-optimal clustering performance. Here, we propose a new unsupervised conditional generation framework, Double Cycle-Consistent Conditional GAN (DC3-GAN), which can generate diverse class-conditioned samples. We enforce the encoder and the generator of GAN to form an encoder-generator pair in addition to the generator-encoder pair, which enables us to avoid the low-diversity generation and the triviality of latent features. We train the encoder-generator pair using real data, which can indirectly estimate the real conditional distribution. Meanwhile, this framework enforces the outputs of the encoder to match the inputs of GAN and the prior noise distribution, which disentangles latent space into two parts: one-hot discrete and continuous latent variables. The former can be directly expressed as clusters and the latter represents remaining unspecified factors. This work demonstrates that enhancing the diversity of unsupervised conditional generated samples can improve the clustering performance. Experiments on different benchmark datasets show that the proposed method outperforms existing generative model-based clustering methods, and also achieves the optimal disentanglement performance.

cs.LG↗

arXiv2021

Room-temperature on-chip orbital angular momentum single-photon sources

On-chip photon sources carrying orbital angular momentum (OAM) are in demand for high-capacity optical information processing in both classical and quantum regimes. However, currently-exploited integrated OAM sources have been primarily limited to the classical regime. Herein, we demonstrate a room-temperature on-chip integrated OAM source that emits well-collimated single photons, with a single-photon purity of g(2)(0) = 0.22, carrying entangled spin and orbital angular momentum states and forming two spatially separated entangled radiation channels with different polarization properties. The OAM-encoded single photons are generated by efficiently outcoupling diverging surface plasmon polaritons excited with a deterministically positioned quantum emitter via Archimedean spiral gratings. Our OAM single-photon sources bridge the gap between conventional OAM manipulation and nonclassical light sources, enabling high-dimensional and large-scale photonic quantum systems for information processing.

physics.optics↗

arXiv2021

Dynamic MEMS-based optical metasurfaces

Optical metasurfaces (OMSs) have shown unprecedented capabilities for versatile wavefront manipulations at the subwavelength scale, thus opening fascinating perspectives for next generation ultracompact optical devices and systems. However, to date, most well-established OMSs are static, featuring well-defined optical responses determined by OMS configurations set during their fabrication. Dynamic OMS configurations investigated so far by using controlled constituent materials or geometrical parameters often exhibit specific limitations and reduced reconfigurability. Here, by combining a thin-film piezoelectric micro-electro-mechanical system (MEMS) with a gap-surface plasmon based OMS, we develop an electrically driven dynamic MEMS-OMS platform that offers controllable phase and amplitude modulation of the reflected light by finely actuating the MEMS mirror. Using this platform, we demonstrate MEMS-OMS components for polarization-independent beam steering and two-dimensional focusing with high modulation efficiencies (~ 50%), broadband operation (~ 20% near the operating wavelength of 800 nm) and fast responses (< 0.4 ms). The developed MEMS-OMS platform offers flexible solutions for realizing complex dynamic 2D wavefront manipulations that could be used in reconfigurable and adaptive optical networks and systems.

physics.optics↗

arXiv2020

Deterministic preparation of spin qubits in droplet etched GaAs quantum dots using quasi-resonant excitation

We present a first comprehensive study on deterministic spin preparation employing excited state resonances of droplet etched GaAs quantum dots. This achievement facilitates future investigations of spin qubit based quantum memories using the GaAs quantum dot material platform. By observation of excitation spectra for a range of fundamental excitonic transitions the properties of different quantum dot energy levels, i.e. shells, are revealed. The innovative use of polarization resolved excitation and detection in quasi-resonant excitation spectroscopy facilitates determination of $85$ $\%$ maximum spin preparation fidelity - irrespective of the relative orientations of lab and quantum dot polarization eigenbases. Additionally, the characteristic non-radiative decay time is investigated as a function of ground state, excitation resonance and excitation power level, yielding decay times as low as $29$ ps for s-p-shell exited state transitions. Finally, by time resolved correlation spectroscopy it is demonstrated that the employed excitation scheme has a significant impact on the electronic environment of quantum dot transitions thereby influencing its charge and coherence.

cond-mat.mes-hall↗

arXiv2020

Maximally entangled and GHz-clocked on-demand photon pair source

We present a 1 GHz-clocked, maximally entangled and on-demand photon pair source based on droplet etched GaAs quantum dots using two-photon excitation. By employing these GaP microlensenhanced devices in conjunction with their substantial brightness, raw entanglement fidelities of up to $0.95 \pm 0.01$ and post-selected photon indistinguishabilities of up to $0.93 \pm 0.01$, the suitability for quantum repeater based long range quantum entanglement distribution schemes is shown. Comprehensive investigations of a complete set of polarization selective two-photon correlations as well as time resolved Hong-Ou-Mandel interferences facilitate innovative methods that determine quantities such as photon extraction and excitation efficiencies as well as pure dephasing directly - opposed to commonly employed indirect techniques.

quant-ph↗

arXiv2020

Unsupervised Hierarchical Graph Representation Learning by Mutual Information Maximization

Graph representation learning based on graph neural networks (GNNs) can greatly improve the performance of downstream tasks, such as node and graph classification. However, the general GNN models do not aggregate node information in a hierarchical manner, and can miss key higher-order structural features of many graphs. The hierarchical aggregation also enables the graph representations to be explainable. In addition, supervised graph representation learning requires labeled data, which is expensive and error-prone. To address these issues, we present an unsupervised graph representation learning method, Unsupervised Hierarchical Graph Representation (UHGR), which can generate hierarchical representations of graphs. Our method focuses on maximizing mutual information between "local" and high-level "global" representations, which enables us to learn the node embeddings and graph embeddings without any labeled data. To demonstrate the effectiveness of the proposed method, we perform the node and graph classification using the learned node and graph embeddings. The results show that the proposed method achieves comparable results to state-of-the-art supervised methods on several benchmarks. In addition, our visualization of hierarchical representations indicates that our method can capture meaningful and interpretable clusters.

cs.LG↗

arXiv2020

Model-Free Voltage Regulation of Unbalanced Distribution Network Based on Surrogate Model and Deep Reinforcement Learning

Accurate knowledge of the distribution system topology and parameters is required to achieve good voltage controls, but this is difficult to obtain in practice. This paper develops a model-free approach based on the surrogate model and deep reinforcement learning (DRL). We have also extended it to deal with unbalanced three-phase scenarios. The key idea is to learn a surrogate model to capture the relationship between the power injections and voltage fluctuation of each node from historical data instead of using the original inaccurate model affected by errors and uncertainties. This allows us to integrate the DRL with the learned surrogate model. In particular, DRL is applied to learn the optimal control strategy from the experiences obtained by continuous interactions with the surrogate model. The integrated framework contains training three networks, i.e., surrogate model, actor, and critic networks, which fully leverage the strong nonlinear fitting ability of deep learning and DRL for online decision making. Several single-phase approaches have also been extended to deal with three-phase unbalance scenarios and the simulation results on the IEEE 123-bus system show that our proposed method can achieve similar performance as those that use accurate physical models.

eess.SY↗

arXiv2020

Distributed Voltage Regulation of Active Distribution System Based on Enhanced Multi-agent Deep Reinforcement Learning

This paper proposes a data-driven distributed voltage control approach based on the spectrum clustering and the enhanced multi-agent deep reinforcement learning (MADRL) algorithm. Via the unsupervised clustering, the whole distribution system can be decomposed into several sub-networks according to the voltage and reactive power sensitivity. Then, the distributed control problem of each sub-network is modeled as Markov games and solved by the enhanced MADRL algorithm, where each sub-network is modeled as an adaptive agent. Deep neural networks are used in each agent to approximate the policy function and the action value function. All agents are centrally trained to learn the optimal coordinated voltage regulation strategy while executed in a distributed manner to make decisions based on only local information. The proposed method can significantly reduce the requirements of communications and knowledge of system parameters. It also effectively deals with uncertainties and can provide online coordinated control based on the latest local information. Comparison results with other existing model-based and data-driven methods on IEEE 33-bus and 123-bus systems demonstrate the effectiveness and benefits of the proposed approach.

eess.SY↗

arXiv2020

On-Chip Spin-Orbit Controlled Excitation of Quantum Emitters Coupled to Hybrid Plasmonic Nanocircuits

On-chip realization of complex photonic functionalities is essential for further progress in planar integrated nanophotonics, especially when involving nonclassical light sources such as quantum emitters (QEs). Hybrid plasmonic nanocircuits integrated with QEs have been attracting considerable attention due to the prospects of significantly enhancing QE emission rates and miniaturizing quantum nanophotonic components. Spin-orbit interactions on subwavelength scales have been increasingly explored in both conventional and quantum nanophotonics for realization and utilization of the spin-dependent flow of light. Here, we propose and realize a dielectric-loaded plasmonic nanocircuit consisting of an achiral spin-orbit coupler for unidirectional routing of pump radiation into branched QE-integrated waveguides. We demonstrate experimentally the circular-polarization controlled coupling of 532-nm pump laser light into polymer-loaded branched waveguides followed by the excitation of spatially separated (by a distance of ~ 10 μm) QEs, nanodiamonds, with multiple nitrogen vacancy centres, that are embedded in and efficiently coupled to the corresponding waveguides. The realization of on-chip spin-orbit controlled excitation of different QEs coupled to branched waveguides opens new avenues for designing complex quantum plasmonic nanocircuits exploiting the spin degree of freedom within chiral quantum nanophotonics.

physics.optics↗

arXiv2019

Hierarchical Attention Networks for Medical Image Segmentation

The medical image is characterized by the inter-class indistinction, high variability, and noise, where the recognition of pixels is challenging. Unlike previous self-attention based methods that capture context information from one level, we reformulate the self-attention mechanism from the view of the high-order graph and propose a novel method, namely Hierarchical Attention Network (HANet), to address the problem of medical image segmentation. Concretely, an HA module embedded in the HANet captures context information from neighbors of multiple levels, where these neighbors are extracted from the high-order graph. In the high-order graph, there will be an edge between two nodes only if the correlation between them is high enough, which naturally reduces the noisy attention information caused by the inter-class indistinction. The proposed HA module is robust to the variance of input and can be flexibly inserted into the existing convolution neural networks. We conduct experiments on three medical image segmentation tasks including optic disc/cup segmentation, blood vessel segmentation, and lung segmentation. Extensive results show our method is more effective and robust than the existing state-of-the-art methods.

cs.CV↗

arXiv2019

Spinning single photons

Single photons carrying spin angular momentum (SAM), i.e., circularly polarized single photons generated typically by subjecting a quantum emitter (QE) to a strong magnetic field at low temperatures are at the core of chiral quantum optics enabling non-reciprocal single-photon configurations and deterministic spin-photon interfaces. Here we propose a conceptually new approach to the room-temperature generation of SAM-coded single photons (SSPs) entailing QE non-radiative coupling to surface plasmons that are transformed, by interacting with an optical metasurface, into a collimated stream of SSPs with the designed handedness. We report on the design, fabrication and characterization of SSP sources consisting of dielectric circular nanoridges with azimuthally varying widths deterministically fabricated on a dielectric-protected silver film around a nanodiamond containing a nitrogen-vacancy centre. With properly engineered phases of QE-originated fields scattered by nanoridges, the out-coupled photons feature a well-defined SAM (with the chirality > 0.8) and high directionality (collection efficiency up to 92%).

cond-mat.mes-hall↗

arXiv2019

Multifunctional Metamirrors Involving Simultaneous Linear-polarization Conversion and Focusing

Reducing a set of diverse bulk-optic-based optical components to a single ultrathin and compact element that enables the same complex functionality has become an emerging research area, propelling further integration and miniaturization in photonics. In this work, we establish a versatile metasurface platform based on gap-surface plasmon meta-atoms enabling efficient linear-polarization conversion along with the complete phase control over reflected fields. Capitalizing on the meta-atom design, multifunctional metamirrors involving linear-polarization conversion and focusing are experimentally demonstrated to generate various kinds of focused beams with distinct phase distributions and wavefronts, reproducing thereby the combined functionalities of conventional half-wave plates, lenses, and even spatial light modulators. The proof-of-concept fabricated metamirrors exhibit excellent capability of linear-polarization conversion and focusing within the wavelength range from 800 to 950 nm under linearly-polarized excitation. The multifunctional metamirrors design developed in this study opens new avenues in the advanced research and applications targeting photonics integration of diversified functionalities.

physics.optics↗

arXiv2019

Hierarchical Distributed Voltage Regulation in Networked Autonomous Grids

We propose a novel algorithm to solve optimal power flow (OPF) that aims at dispatching controllable distributed energy resources (DERs) for voltage regulation at minimum cost. The proposed algorithm features unprecedented scalability to large distribution networks by utilizing an information structure based on networked autonomous grids (AGs). Specifically, each AG is a subtree of a large distribution network that has a tree topology. The topology and line parameters of each AG are known only to a regional coordinator (RC) that is responsible for communicating with and dispatching the DERs within this AG. The reduced network, where each AG is treated as a node, is managed by a central coordinator (CC), which knows the topology and line parameters of the reduced network only and communicates with all the RCs. We jointly explore this information structure and the structure of the linearized distribution power flow (LinDistFlow) model to derive a hierarchical, distributed implementation of the primal-dual gradient algorithm that solves the OPF. The proposed implementation significantly reduces the computation burden compared to the centrally coordinated implementation of the primal-dual algorithm. Numerical results on a 4,521-node test feeder show that the proposed hierarchical distributed algorithm can achieve an improvement of more than tenfold in the speed of convergence compared to the centrally coordinated primal-dual algorithm.

math.OC↗