arXiv subjects

Yu Wang

Publications and source records attributed to Yu Wang.

At least 1,171 records · Page 65Linked to original sources

arXiv2020

DRF: A Framework for High-Accuracy Autonomous Driving Vehicle Modeling

An accurate vehicle dynamic model is the key to bridge the gap between simulation and real road test in autonomous driving. In this paper, we present a Dynamic model-Residual correction model Framework (DRF) for vehicle dynamic modeling. On top of any existing open-loop dynamic model, this framework builds a Residual Correction Model (RCM) by integrating deep Neural Networks (NN) with Sparse Variational Gaussian Process (SVGP) model. RCM takes a sequence of vehicle control commands and dynamic status for a certain time duration as modeling inputs, extracts underlying context from this sequence with deep encoder networks, and predicts open-loop dynamic model prediction errors. Five vehicle dynamic models are derived from DRF via encoder variation. Our contribution is consolidated by experiments on evaluation of absolute trajectory error and similarity between DRF outputs and the ground truth. Compared to classic rule-based and learning-based vehicle dynamic models, DRF accomplishes as high as 74.12% to 85.02% of absolute trajectory error drop among all DRF variations.

cs.RO↗

arXiv2020

State sharding model on the blockchain

Blockchain is an incrementally updated ledger maintained by distributed nodes rather than centralized organizations. The current blockchain technology faces scalability issues, which include two aspects: low transaction throughput and high storage capacity costs. This paper studies the blockchain structure based on state sharding technology, and mainly solves the problem of non-scalability of block chain storage. This paper designs and implements the blockchain state sharding scheme, proposes a specific state sharding data structure and algorithm implementation, and realizes a complete blockchain structure so that the blockchain has the advantages of high throughput, processing a large number of transactions and saving storage costs. Experimental results show that a blockchain network with more than 100,000 nodes can be divided into 1024 shards. A blockchain network with this structure can process 500,000 transactions in about 5 seconds. If the consensus time of the blockchain is about 10 seconds, and the block generation time of the blockchain system of the sharding mechanism is 15 seconds, the transaction throughput can reach 33,000 tx/sec. Experimental results show that the throughput of the proposed protocol increases with the increase of the network node size. This confirms the scalability of the blockchain structure based on sharding technology.

cs.CR↗

arXiv2020

Stabilization of hexazine rings in potassium polynitride at high pressure

Polynitrogen molecules represent the ultimate high energy-density materials as they have a huge potential chemical energy originating from their high enthalpy. However, synthesis and storage of such compounds remain a big challenge because of difficulties to find energy efficient synthetic routes and stabilization mechanisms. Compounds of metals with nitrogen represent promising candidates for realization of energetic polynitrogen compounds, which are also environmentally benign. Here we report the synthesis of polynitrogen planar N6 hexazine rings, stabilized in K2N6 compound, which was formed from K azide upon laser heating in a diamond anvil cell at high pressures in excess of 45 GPa and remains metastable down to 20 GPa. Synchrotron X-ray diffraction and Raman spectroscopy are used to identify this material, also exhibiting metallic luster, being all consistent with theoretically predicted structural, vibrational and electronic properties. The documented here N6 hexazine rings represent new highly energetic polynitrogens, which have a potential for future recovery and utilization.

cond-mat.mtrl-sci↗

arXiv2020

Large array of Schrödinger cat states facilitated by an optical waveguide

Quantum engineering using photonic structures offer new capabilities for atom-photon interactions for quantum optics and atomic physics, which could eventually lead to integrated quantum devices. Despite the rapid progress in the variety of structures, coherent excitation of the motional states of atoms in a photonic waveguide using guided modes has yet to be demonstrated. Here, we use the waveguide mode of a hollow-core photonic crystal fibre to manipulate the mechanical Fock states of single atoms in a harmonic potential inside the fibre. We create a large array of Schrödinger cat states, a quintessential feature of quantum physics and a key element in quantum information processing and metrology, of approximately 15000 atoms along the fibre by entangling the electronic state with the coherent harmonic oscillator state of each individual atom. Our results provide a useful step for quantum information and simulation with a wide range of photonic waveguide systems.

quant-ph↗

arXiv2020

Joint Contrastive Learning with Infinite Possibilities

This paper explores useful modifications of the recent development in contrastive learning via novel probabilistic modeling. We derive a particular form of contrastive loss named Joint Contrastive Learning (JCL). JCL implicitly involves the simultaneous learning of an infinite number of query-key pairs, which poses tighter constraints when searching for invariant features. We derive an upper bound on this formulation that allows analytical solutions in an end-to-end training manner. While JCL is practically effective in numerous computer vision applications, we also theoretically unveil the certain mechanisms that govern the behavior of JCL. We demonstrate that the proposed formulation harbors an innate agency that strongly favors similarity within each instance-specific class, and therefore remains advantageous when searching for discriminative features among distinct instances. We evaluate these proposals on multiple benchmarks, demonstrating considerable improvements over existing algorithms. Code is publicly available at: https://github.com/caiqi/Joint-Contrastive-Learning.

cs.CV↗

arXiv2020

How Effective is Task-Agnostic Data Augmentation for Pretrained Transformers?

Task-agnostic forms of data augmentation have proven widely effective in computer vision, even on pretrained models. In NLP similar results are reported most commonly for low data regimes, non-pretrained models, or situationally for pretrained models. In this paper we ask how effective these techniques really are when applied to pretrained transformers. Using two popular varieties of task-agnostic data augmentation (not tailored to any particular task), Easy Data Augmentation (Wei and Zou, 2019) and Back-Translation (Sennrichet al., 2015), we conduct a systematic examination of their effects across 5 classification tasks, 6 datasets, and 3 variants of modern pretrained transformers, including BERT, XLNet, and RoBERTa. We observe a negative result, finding that techniques which previously reported strong improvements for non-pretrained models fail to consistently improve performance for pretrained transformers, even when training data is limited. We hope this empirical analysis helps inform practitioners where data augmentation techniques may confer improvements.

cs.LG↗

arXiv2020

Model-Free Reinforcement Learning for Stochastic Games with Linear Temporal Logic Objectives

We study the problem of synthesizing control strategies for Linear Temporal Logic (LTL) objectives in unknown environments. We model this problem as a turn-based zero-sum stochastic game between the controller and the environment, where the transition probabilities and the model topology are fully unknown. The winning condition for the controller in this game is the satisfaction of the given LTL specification, which can be captured by the acceptance condition of a deterministic Rabin automaton (DRA) directly derived from the LTL specification. We introduce a model-free reinforcement learning (RL) methodology to find a strategy that maximizes the probability of satisfying a given LTL specification when the Rabin condition of the derived DRA has a single accepting pair. We then generalize this approach to LTL formulas for which the Rabin condition has a larger number of accepting pairs, providing a lower bound on the satisfaction probability. Finally, we illustrate applicability of our RL method on two motion planning case studies.

cs.RO↗

arXiv2020

IrO2 Surface Complexions Identified Through Machine Learning and Surface Investigations

A Gaussian Approximation Potential (GAP) was trained using density-functional theory data to enable a global geometry optimization of low-index rutile IrO2 facets through simulated annealing. Ab initio thermodynamics identifies (101) and (111) (1x1)-terminations competitive with (110) in reducing environments. Experiments on single crystals find that (101) facets dominate, and exhibit the theoretically predicted (1x1) periodicity and X-ray photoelectron spectroscopy (XPS) core level shifts. The obtained structures are analogous to the complexions discussed in the context of ceramic battery materials.

cond-mat.mtrl-sci↗

arXiv2020

DL-IAPS and PJSO: A Path/Speed Decoupled Trajectory Optimization and its Application in Autonomous Driving

This paper presents a free space trajectory optimization algorithm of autonomous driving vehicle, which decouples the collision-free trajectory planning problem into a Dual-Loop Iterative Anchoring Path Smoothing (DL-IAPS) and a Piece-wise Jerk Speed Optimization (PJSO). The work leads to remarkable driving performance improvements including more precise collision avoidance, higher control feasibility and better driving comfort, as those are often hard to realize in other existing path/speed decoupled trajectory optimization methods. Our algorithm's efficiency, robustness and adaptiveness to complex driving scenarios have been validated by both simulations and real on-road tests.

cs.RO↗

arXiv2020

Verifying Stochastic Hybrid Systems with Temporal Logic Specifications via Model Reduction

We present a scalable methodology to verify stochastic hybrid systems. Using the Mori-Zwanzig reduction method, we construct a finite state Markov chain reduction of a given stochastic hybrid system and prove that this reduced Markov chain is approximately equivalent to the original system in a distributional sense. Approximate equivalence of the stochastic hybrid system and its Markov chain reduction means that analyzing the Markov chain with respect to a suitably strengthened property, allows us to conclude whether the original stochastic hybrid system meets its temporal logic specifications. We present the first statistical model checking algorithms to verify stochastic hybrid systems against correctness properties, expressed in the linear inequality linear temporal logic (iLTL) or the metric interval temporal logic (MITL).

math.OC↗

arXiv2020

DualLip: A System for Joint Lip Reading and Generation

Lip reading aims to recognize text from talking lip, while lip generation aims to synthesize talking lip according to text, which is a key component in talking face generation and is a dual task of lip reading. In this paper, we develop DualLip, a system that jointly improves lip reading and generation by leveraging the task duality and using unlabeled text and lip video data. The key ideas of the DualLip include: 1) Generate lip video from unlabeled text with a lip generation model, and use the pseudo pairs to improve lip reading; 2) Generate text from unlabeled lip video with a lip reading model, and use the pseudo pairs to improve lip generation. We further extend DualLip to talking face generation with two additionally introduced components: lip to face generation and text to speech generation. Experiments on GRID and TCD-TIMIT demonstrate the effectiveness of DualLip on improving lip reading, lip generation, and talking face generation by utilizing unlabeled data. Specifically, the lip generation model in our DualLip system trained with only10% paired data surpasses the performance of that trained with the whole paired data. And on the GRID benchmark of lip reading, we achieve 1.16% character error rate and 2.71% word error rate, outperforming the state-of-the-art models using the same amount of paired data.

cs.MM↗

arXiv2020

In the Eyes of the Beholder: Analyzing Social Media Use of Neutral and Controversial Terms for COVID-19

During the COVID-19 pandemic, "Chinese Virus" emerged as a controversial term for coronavirus. To some, it may seem like a neutral term referring to the physical origin of the virus. To many others, however, the term is in fact attaching ethnicity to the virus. While both arguments appear reasonable, quantitative analysis of the term's real-world usage is lacking to shed light on the issues behind the controversy. In this paper, we attempt to fill this gap. To model the substantive difference of tweets with controversial terms and those with non-controversial terms, we apply topic modeling and LIWC-based sentiment analysis. To test whether "Chinese Virus" and "COVID-19" are interchangeable, we formulate it as a classification task, mask out these terms, and classify them using the state-of-the-art transformer models. Our experiments consistently show that the term "Chinese Virus" is associated with different substantive topics and sentiment compared with "COVID-19" and that the two terms are easily distinguishable by looking at their context.

cs.SI↗

arXiv2020

SONYC-UST-V2: An Urban Sound Tagging Dataset with Spatiotemporal Context

We present SONYC-UST-V2, a dataset for urban sound tagging with spatiotemporal information. This dataset is aimed for the development and evaluation of machine listening systems for real-world urban noise monitoring. While datasets of urban recordings are available, this dataset provides the opportunity to investigate how spatiotemporal metadata can aid in the prediction of urban sound tags. SONYC-UST-V2 consists of 18510 audio recordings from the "Sounds of New York City" (SONYC) acoustic sensor network, including the timestamp of audio acquisition and location of the sensor. The dataset contains annotations by volunteers from the Zooniverse citizen science platform, as well as a two-stage verification with our team. In this article, we describe our data collection procedure and propose evaluation metrics for multilabel classification of urban sound tags. We report the results of a simple baseline model that exploits spatiotemporal information.

cs.SD↗

arXiv2020

A Generic Graph-based Neural Architecture Encoding Scheme for Predictor-based NAS

This work proposes a novel Graph-based neural ArchiTecture Encoding Scheme, a.k.a. GATES, to improve the predictor-based neural architecture search. Specifically, different from existing graph-based schemes, GATES models the operations as the transformation of the propagating information, which mimics the actual data processing of neural architecture. GATES is a more reasonable modeling of the neural architectures, and can encode architectures from both the "operation on node" and "operation on edge" cell search spaces consistently. Experimental results on various search spaces confirm GATES's effectiveness in improving the performance predictor. Furthermore, equipped with the improved performance predictor, the sample efficiency of the predictor-based neural architecture search (NAS) flow is boosted. Codes are available at https://github.com/walkerning/aw_nas.

cs.LG↗

arXiv2020

Hunting for the host galaxy groups of binary black holes and the application in constraining Hubble constant

The discovery of gravitational-wave (GW) signals, produced by the coalescence of stellar-mass binary black holes (SBBHs), opens a new window to study the astrophysical origins and dynamical evolutions of compact binaries. In addition, these GW events can be treated as the standard sirens to constrain various cosmological parameters. Both issues require the host identification for these GW events, with help of the spatial resolution of GW detector networks. In this paper, we investigate the capabilities of various detector networks for identifying the SBBHs' host galaxy groups, rather than their host galaxies, which can overcome the influence of galaxies' proper motions in dark matter halos for measuring the cosmological parameters. In our analysis, the group catalog of SDSS DR7 with redshift $z\in(0.01,0.1)$ is considered as an example of the application. We find that for the second-generation (2G) detector network, the host galaxy groups of around $(0.7-6.9)$ SBBHs can be identified per year assuming all sources are $30-30\ M_{\odot}$ binaries, and that all five detectors in the network are in lock 100\% of the time. For the 3G detector network, this number becomes $(3.9-40.0)$ yr$^{-1}$. We also investigate the potential constraint on the Hubble constant $H_0$ by these GW events, if their redshift information is extracted from the candidates of host galaxy groups. We find that, by five-year's full time observations, 2G detector network is expected to give a constraint of $ΔH_0/H_0\sim (1\%,4\%)$, which can be more than two order smaller if considering the 3G detector network.

astro-ph.CO↗

arXiv2020

Joint Visual and Wireless Signal Feature based Approach for High-Precision Indoor Localization

The existing localization systems for indoor applications basically rely on wireless signal. With the massive deployment of low-cost cameras, the visual image based localization become attractive as well. However, in the existing literature, the hybrid visual and wireless approaches simply combine the above schemes in a straight forward manner, and fail to explore the interactions between them. In this paper, we propose a joint visual and wireless signal feature based approach for high-precision indoor localization system. In this joint scheme, WiFi signals are utilized to compute the coarse area with likelihood probability and visual images are used to fine-tune the localization result. Based on the numerical results, we show that the proposed scheme can achieve 0.62m localization accuracy with near real-time running time.

eess.SP↗

arXiv2020

Shift Equivariance in Object Detection

Robustness to small image translations is a highly desirable property for object detectors. However, recent works have shown that CNN-based classifiers are not shift invariant. It is unclear to what extent this could impact object detection, mainly because of the architectural differences between the two and the dimensionality of the prediction space of modern detectors. To assess shift equivariance of object detection models end-to-end, in this paper we propose an evaluation metric, built upon a greedy search of the lower and upper bounds of the mean average precision on a shifted image set. Our new metric shows that modern object detection architectures, no matter if one-stage or two-stage, anchor-based or anchor-free, are sensitive to even one pixel shift to the input images. Furthermore, we investigate several possible solutions to this problem, both taken from the literature and newly proposed, quantifying the effectiveness of each one with the suggested metric. Our results indicate that none of these methods can provide full shift equivariance. Measuring and analyzing the extent of shift variance of different models and the contributions of possible factors, is a first step towards being able to devise methods that mitigate or even leverage such variabilities.

cs.CV↗

arXiv2020

Reduction of laser intensity noise over 1 MHz band for single atom trapping

We reduce the intensity noise of laser light by using an electro-optic modulator and a cousto-optic modulator in series. The electro-optic modulator reduces noise at high frequency(10 kHz to 1 MHz), while the acousto-optic modulator sets the average power of the light and reduces noise at low frequency (up to 10 kHz). The light is then used to trap single sodium atoms in an optical tweezer, where the lifetime of the atoms is limited by parametric heating due to laser noise at twice the trapping frequency. With our noise eater, the noise is reduced by up to 15 dB at these frequencies, and the lifetime is increased by an order of magnitude to around 6 seconds. Our technique is general and acts directly on the laser beam, expanding laser options for sensitive optical trapping applications.

physics.atom-ph↗