arXiv subjects

Zheng Gong

Publications and source records attributed to Zheng Gong.

At least 73 records · Page 4Linked to original sources

arXiv2022

Lidar with Velocity: Correcting Moving Objects Point Cloud Distortion from Oscillating Scanning Lidars by Fusion with Camera

Lidar point cloud distortion from moving object is an important problem in autonomous driving, and recently becomes even more demanding with the emerging of newer lidars, which feature back-and-forth scanning patterns. Accurately estimating moving object velocity would not only provide a tracking capability but also correct the point cloud distortion with more accurate description of the moving object. Since lidar measures the time-of-flight distance but with a sparse angular resolution, the measurement is precise in the radial measurement but lacks angularly. Camera on the other hand provides a dense angular resolution. In this paper, Gaussian-based lidar and camera fusion is proposed to estimate the full velocity and correct the lidar distortion. A probabilistic Kalman-filter framework is provided to track the moving objects, estimate their velocities and simultaneously correct the point clouds distortions. The framework is evaluated on real road data and the fusion method outperforms the traditional ICP-based and point-cloud only method. The complete working framework is open-sourced (https://github.com/ISEE-Technology/lidar-with-velocity) to accelerate the adoption of the emerging lidars.

cs.RO↗

arXiv2022

JiuZhang: A Chinese Pre-trained Language Model for Mathematical Problem Understanding

This paper aims to advance the mathematical intelligence of machines by presenting the first Chinese mathematical pre-trained language model~(PLM) for effectively understanding and representing mathematical problems. Unlike other standard NLP tasks, mathematical texts are difficult to understand, since they involve mathematical terminology, symbols and formulas in the problem statement. Typically, it requires complex mathematical logic and background knowledge for solving mathematical problems. Considering the complex nature of mathematical texts, we design a novel curriculum pre-training approach for improving the learning of mathematical PLMs, consisting of both basic and advanced courses. Specially, we first perform token-level pre-training based on a position-biased masking strategy, and then design logic-based pre-training tasks that aim to recover the shuffled sentences and formulas, respectively. Finally, we introduce a more difficult pre-training task that enforces the PLM to detect and correct the errors in its generated solutions. We conduct extensive experiments on offline evaluation (including nine math-related tasks) and online $A/B$ test. Experimental results demonstrate the effectiveness of our approach compared with a number of competitive baselines. Our code is available at: \textcolor{blue}{\url{https://github.com/RUCAIBox/JiuZhang}}.

cs.CL↗

arXiv2022

ElitePLM: An Empirical Study on General Language Ability Evaluation of Pretrained Language Models

Nowadays, pretrained language models (PLMs) have dominated the majority of NLP tasks. While, little research has been conducted on systematically evaluating the language abilities of PLMs. In this paper, we present a large-scale empirical study on general language ability evaluation of PLMs (ElitePLM). In our study, we design four evaluation dimensions, i.e. memory, comprehension, reasoning, and composition, to measure ten widely-used PLMs within five categories. Our empirical results demonstrate that: (1) PLMs with varying training objectives and strategies are good at different ability tests; (2) fine-tuning PLMs in downstream tasks is usually sensitive to the data size and distribution; (3) PLMs have excellent transferability between similar tasks. Moreover, the prediction results of PLMs in our experiments are released as an open resource for more deep and detailed analysis on the language abilities of PLMs. This paper can guide the future work to select, apply, and design PLMs for specific tasks. We have made all the details of experiments publicly available at https://github.com/RUCAIBox/ElitePLM.

cs.CL↗

arXiv2022

WuDaoMM: A large-scale Multi-Modal Dataset for Pre-training models

Compared with the domain-specific model, the vision-language pre-training models (VLPMs) have shown superior performance on downstream tasks with fast fine-tuning process. For example, ERNIE-ViL, Oscar and UNIMO trained VLPMs with a uniform transformers stack architecture and large amounts of image-text paired data, achieving remarkable results on downstream tasks such as image-text reference(IR and TR), vision question answering (VQA) and image captioning (IC) etc. During the training phase, VLPMs are always fed with a combination of multiple public datasets to meet the demand of large-scare training data. However, due to the unevenness of data distribution including size, task type and quality, using the mixture of multiple datasets for model training can be problematic. In this work, we introduce a large-scale multi-modal corpora named WuDaoMM, totally containing more than 650M image-text pairs. Specifically, about 600 million pairs of data are collected from multiple webpages in which image and caption present weak correlation, and the other 50 million strong-related image-text pairs are collected from some high-quality graphic websites. We also release a base version of WuDaoMM with 5 million strong-correlated image-text pairs, which is sufficient to support the common cross-modal model pre-training. Besides, we trained both an understanding and a generation vision-language (VL) model to test the dataset effectiveness. The results show that WuDaoMM can be applied as an efficient dataset for VLPMs, especially for the model in text-to-image generation task. The data is released at https://data.wudaoai.cn

cs.CV↗

arXiv2022

Deciphering in situ electron dynamics of ultrarelativistic plasma via polarization pattern of emitted gamma photons

Understanding and interpretation of the dynamics of ultrarelativistic plasma is a challenge, which calls for the development of methods for \textit{in situ} probing the plasma dynamical characteristics. We put forward a new method, harnessing polarization properties of $γ$-photons emitted from a non-pre-polarized plasma irradiated by a circularly polarized pulse. We show that the angular pattern of $γ$-photon linear polarization is explicitly correlated with the dynamics of the radiating electrons, which provides information on the laser-plasma interaction regime. Furthermore, with the $γ$-photon circular polarization originating from the electron radiative spin-flips, the plasma susceptibility to quantum electrodynamical processes is gauged. Our study demonstrates that the polarization signal of emitted $γ$-photons can be a versatile information source, which would be beneficial for the research fields of laser-driven plasma, accelerator science, and laboratory astrophysics.

physics.plasm-ph↗

arXiv2022

Monolithic Kerr and electro-optic hybrid microcombs

Advances in microresonator-based soliton generation promise chip-scale integration of optical frequency comb for applications spanning from time keeping to frequency synthesis. Miniaturized cavities harness Kerr nonlinearity and enable terahertz soliton repetition rates. However, such high repetition rates are not amenable to direct electronic detection. Here, we demonstrate hybrid Kerr and electro-optic microcombs using the lithium niobate thin film that exhibits both Kerr and Pockels nonlinearities. By interleaving the high-repetition-rate Kerr soliton comb with the low-repetition-rate electro-optic comb on the same waveguide, the wide Kerr soliton mode spacing is divided within a single chip, allowing for subsequent electronic detection and feedback control of the soliton repetition rate. Our work establishes an integrated electronic interface to Kerr solitons of terahertz repetition rates, paving the path towards chipscale optical-to-microwave frequency division and comb locking.

physics.optics↗

arXiv2021

Retrieving transient magnetic fields of ultrarelativistic laser plasma via ejected electron polarization

Interaction of an ultrastrong short laser pulse with non-prepolarized near-critical density plasma is investigated in an ultrarelativistic regime, with an emphasis on the radiative spin polarization of ejected electrons. Our particle-in-cell simulations show explicit correlations between the angle resolved electron polarization and the structure and properties of the transient quasistatic plasma magnetic field. While the magnitude of the spin signal is the indicator of the magnetic field strength created by the longitudinal electron current, the asymmetry of electron polarization is found to gauge the island-like magnetic distribution which emerges due to the transverse current induced by the laser wave front. Our studies demonstrate that the spin degree of freedom of ejected electrons could potentially serve as an efficient tool to retrieve the features of strong plasma fields.

physics.plasm-ph↗

arXiv2021

A life-long SLAM approach using adaptable local maps based on rasterized LIDAR images

Most real-time autonomous robot applications require a robot to traverse through a dynamic space for a long time. In some cases, a robot needs to work in the same environment. Such applications give rise to the problem of a life-long SLAM system. Life-long SLAM presents two main challenges i.e. the tracking should not fail in a dynamic environment and the need for a robust and efficient mapping strategy. The system should update maps with new information; while also keeping track of older observations. But, mapping for a long time can require higher computational requirements. In this paper, we propose a solution to the problem of life-long SLAM. We represent the global map as a set of rasterized images of local maps along with a map management system responsible for updating local maps and keeping track of older values. We also present an efficient approach of using the bag of visual words method for loop closure detection and relocalization. We evaluate the performance of our system on the KITTI dataset and an indoor dataset. Our loop closure system reported recall and precision of above 90 percent. The computational cost of our system is much lower as compared to state-of-the-art methods. Our method reports lower computational requirements even for long-term operation.

cs.RO↗

arXiv2021

WenLan: Bridging Vision and Language by Large-Scale Multi-Modal Pre-Training

Multi-modal pre-training models have been intensively explored to bridge vision and language in recent years. However, most of them explicitly model the cross-modal interaction between image-text pairs, by assuming that there exists strong semantic correlation between the text and image modalities. Since this strong assumption is often invalid in real-world scenarios, we choose to implicitly model the cross-modal correlation for large-scale multi-modal pre-training, which is the focus of the Chinese project `WenLan' led by our team. Specifically, with the weak correlation assumption over image-text pairs, we propose a two-tower pre-training model called BriVL within the cross-modal contrastive learning framework. Unlike OpenAI CLIP that adopts a simple contrastive learning method, we devise a more advanced algorithm by adapting the latest method MoCo into the cross-modal scenario. By building a large queue-based dictionary, our BriVL can incorporate more negative samples in limited GPU resources. We further construct a large Chinese multi-source image-text dataset called RUC-CAS-WenLan for pre-training our BriVL model. Extensive experiments demonstrate that the pre-trained BriVL model outperforms both UNITER and OpenAI CLIP on various downstream tasks.

cs.CV↗

arXiv2021

Radiative polarization dynamics of relativistic electrons in an intense electromagnetic field

We propose a self-consistent model which utilizes the polarization vector to theoretically describe the evolution of spin polarization of relativistic electrons in an intense electromagnetic field. The variation of radiative polarization due to instantaneous no photon emission is introduced into our model, which extends the applicability of the polarization vector model derived from the nonlinear Compton scattering under local constant crossed-field approximation to the complex electromagnetic environment in laser plasma interaction. According to this model, we develop a Monte Carlo method to simulate the electron spin under the influence of radiation and precession simultaneously. Our model is consistent with the quantum physical picture that spin can only be described by a probability distribution before measurement, and it contains the entire information on the spin. The correctness of our model is confirmed by the successful reproduction of the Sokolov-Ternov effect and the comparison of the simulation results with other models in the literature. The results show the superiority in accuracy, applicability, and computational efficiency of our model, and we believe that our model is a better choice to deal with the electron spin in particle-in-cell simulation for laser plasma interaction.

physics.plasm-ph↗

arXiv2021

Super-Heavy Ions Acceleration Driven by Ultrashort Laser Pulses at Ultrahigh Intensity

The acceleration of super-heavy ions (SHIs) from plasmas driven by ultrashort (tens of femtoseconds) laser pulses is a challenging topic waiting for breakthrough. The detecting and controlling of the ionization process, and the adoption of the optimal acceleration scheme are crucial for the generation of highly energetic SHIs. Here, we report the experimental results on the generation of deeply ionized super-heavy ions (Au) with unprecedented energy of 1.2 GeV utilizing ultrashort laser pulses (22 fs) at the intensity of 10^22 W/cm2. A novel self-calibrated diagnostic method was developed to acquire the absolute energy spectra and charge state distributions of Au ions abundant at the charge state of 51+ and reaching up to 61+. The measured charge state distributions supported by 2D particle-in-cell simulations serves as an additional tool to inspect the ionization dynamics associated with SHI acceleration, revealing that the laser intensity is the crucial parameter for the acceleration of Au ions over the pulse duration. The use of double-layer targets results in a prolongation of the acceleration time without sacrificing the strength of acceleration field, which is highly favorable for the generation of high-energy super heavy ions.

physics.plasm-ph↗

arXiv2021

The Efficient Hedging Frontier with Deep Neural Networks

The trade off between risks and returns gives rise to multi-criteria optimisation problems that are well understood in finance, efficient frontiers being the tool to navigate their set of optimal solutions. Motivated by the recent advances in the use of deep neural networks in the context of hedging vanilla options when markets have frictions, we introduce the Efficient Hedging Frontier (EHF) by enriching the pipeline with a filtering step that allows to trade off costs and risks. This way, a trader's risk preference is matched with an expected hedging cost on the frontier, and the corresponding hedging strategy can be computed with a deep neural network. We further develop our framework to improve the EHF and find better hedging strategies. By adding a random forest classifier to the pipeline to forecast market movements, we show how the frontier shifts towards lower costs and reduced risks, which indicates that the overall hedging performances have improved. In addition, by designing a new recurrent neural network, we also find strategies on the frontier where hedging costs are even lower.

q-fin.CP↗

arXiv2021

6-DOF Feature based LIDAR SLAM using ORB Features from Rasterized Images of 3D LIDAR Point Cloud

An accurate and computationally efficient SLAM algorithm is vital for modern autonomous vehicles. To make a lightweight the algorithm, most SLAM systems rely on feature detection from images for vision SLAM or point cloud for laser-based methods. Feature detection through a 3D point cloud becomes a computationally challenging task. In this paper, we propose a feature detection method by projecting a 3D point cloud to form an image and apply the vision-based feature detection technique. The proposed method gives repeatable and stable features in a variety of environments. Based on such features, we build a 6-DOF SLAM system consisting of tracking, mapping, and loop closure threads. For loop detection, we employ a 2-step approach i.e. nearest key-frames detection and loop candidate verification by matching features extracted from rasterized LIDAR images. Furthermore, we utilize a key-frame structure to achieve a lightweight SLAM system. The proposed system is evaluated with implementation on the KITTI dataset and the University of Michigan Ford Campus dataset. Through experimental results, we show that the algorithm presented in this paper can substantially reduce the computational cost of feature detection from the point cloud and the whole SLAM system while giving accurate results.

cs.RO↗

arXiv2021

Emission of electromagnetic waves as a stopping mechanism for nonlinear collisionless ionization waves in a high-$β$ regime

A high energy density plasma embedded in a neutral gas is able to launch an outward-propagating nonlinear electrostatic ionization wave that traps energetic electrons. The trapping maintains a strong sheath electric field, enabling rapid and long-lasting wave propagation aided by field ionization. Using 1D3V kinetic simulations, we examine the propagation of the ionization wave in the presence of a transverse MG-level magnetic field with the objective to identify qualitative changes in a regime where the initial thermal pressure of the plasma exceeds the pressure of the magnetic field ($β>1$). Our key finding is that the magnetic field stops the propagation by causing the energetic electrons sustaining the wave to lose their energy by emitting an electromagnetic wave. The emission is accompanied by the magnetic field expulsion from the plasma and an increased electron loss from the trapping wave structure. The described effect provides a mechanism mitigating rapid plasma expansion for those applications that involve an embedded plasma, such as high-flux neutron production from laser-irradiated deuterium gas jets.

physics.plasm-ph↗

arXiv2021

Ultralow-threshold thin-film lithium niobate optical parametric oscillator

Materials with strong $χ^{(2)}$ optical nonlinearity, especially lithium niobate, play a critical role in building optical parametric oscillators (OPOs). However, chip-scale integration of low-loss $χ^{(2)}$ materials remains challenging and limits the threshold power of on-chip $χ^{(2)}$ OPO. Here we report the first on-chip lithium niobate optical parametric oscillator at the telecom wavelengths using a quasi-phase matched, high-quality microring resonator, whose threshold power ($\sim$30 $μ$W) is 400 times lower than that in previous $χ^{(2)}$ integrated photonics platforms. An on-chip power conversion efficiency of 11% is obtained at a pump power of 93 $μ$W. The OPO wavelength tuning is achieved by varying the pump frequency and chip temperature. With the lowest power threshold among all on-chip OPOs demonstrated so far, as well as advantages including high conversion efficiency, flexibility in quasi-phase matching and device scalability, the thin-film lithium niobate OPO opens new opportunities for chip-based tunable classical and quantum light sources and provides an potential platform for realizing photonic neural networks.

physics.optics↗

arXiv2020

III-Nitride nanophotonics for beyond-octave soliton generation and self-referencing

Frequency microcombs, successors to mode-locked laser and fiber combs, enable miniature rulers of light for applications including precision metrology, molecular fingerprinting, and exoplanet discoveries. To enable the frequency ruling function, microcombs must be stabilized by locking their carrier-envelop offset frequency. So far, the microcomb stabilization remains compounded by the elaborate optics external to the chip, thus evading its scaling benefit. To address this challenge, here we demonstrate a nanophotonic chip solution based on aluminum nitride thin films, which simultaneously offer optical Kerr nonlinearity for generating octave soliton combs and Pockels nonlinearity for enabling heterodyne detection of the offset frequency. The agile dispersion control of crystalline III-Nitride photonics permits high-fidelity generation of solitons with features including 1.5-octave comb span, dual dispersive waves, and sub-terahertz repetition rates down to 220 gigahertz. These attractive characteristics, aided by on-chip phase-matched aluminum nitride waveguides, allow the full determination of the offset frequency. Our proof-of-principle demonstration represents an important milestone towards fully-integrated self-locked microcombs for portable optical atomic clocks and frequency synthesizers.

physics.app-ph↗

arXiv2020

Energetic spin-polarized proton beams from two-stage coherent acceleration in laser-driven plasma

We propose a scheme to overcome the great challenge of polarization loss in spin-polarized ion acceleration. When a petawatt laser pulse penetrates through a compound plasma target consisting of a double layer slab and prepolarized hydrogen halide gas, a strong forward moving quasistatic longitudinal electric field is constructed by the self-generated laser-driven plasma. This field with a varying drift velocity efficiently boosts the prepolarized protons via a two-stage coherent acceleration process. Its merit is not only achieving a highly energetic beam but also eliminating the undesired polarization loss of the accelerated protons. We study the proton dynamics via Hamiltonian analyses, specifically deriving the threshold of triggering the two-stage coherent acceleration. To confirm the theoretical predictions, we perform three-dimensional PIC simulations, where unprecedented proton beams with energy approximating half GeV and polarization ratio $\sim$ 94\% are obtained.

physics.plasm-ph↗

arXiv2020

Cascaded Generation of a Sub-10-Attosecond Half-Cycle Pulse

Sub-10-attosecond pulses with half-cycle electric fields provide exceptional options to detect and manipulate electrons in the atomic timescale. However, the availability of such pulses is still challenging. Here, we propose a method to generate isolated sub-10-attosecond half-cycle pulses based on a cascade process naturally happening in plasma. A 100s-attosecond pulse is first generated by shooting a moderate overdense plasma with a one-cycle femtosecond pulse. After that, the generated attosecond pulse cascadedly produce a sub-10-attosecond half-cycle pulse in the transmission direction by unipolarly perturbing a nanometer-thin relativistic electron sheet naturally form in the plasma. Two-dimensional particle-in-cell simulations indicate that an isolated half-cycle pulse with the duration of 8.3 attoseconds can be produced. Apart from one-cycle driving pulse, such a scheme also can be realized with a commercial 100-TW 25-fs driving laser by shaping the pulse with a relativistic plasma lens in advance.

physics.plasm-ph↗