arXiv subjects

Yi Zhang

Publications and source records attributed to Yi Zhang.

At least 991 records · Page 55Linked to original sources

arXiv2021

Rip van Winkle's Razor: A Simple Estimate of Overfit to Test Data

Traditional statistics forbids use of test data (a.k.a. holdout data) during training. Dwork et al. 2015 pointed out that current practices in machine learning, whereby researchers build upon each other's models, copying hyperparameters and even computer code -- amounts to implicitly training on the test set. Thus error rate on test data may not reflect the true population error. This observation initiated {\em adaptive data analysis}, which provides evaluation mechanisms with guaranteed upper bounds on this difference. With statistical query (i.e. test accuracy) feedbacks, the best upper bound is fairly pessimistic: the deviation can hit a practically vacuous value if the number of models tested is quadratic in the size of the test set. In this work, we present a simple new estimate, {\em Rip van Winkle's Razor}. It relies upon a new notion of \textquotedblleft information content\textquotedblright\ of a model: the amount of information that would have to be provided to an expert referee who is intimately familiar with the field and relevant science/math, and who has been just been woken up after falling asleep at the moment of the creation of the test data (like \textquotedblleft Rip van Winkle\textquotedblright\ of the famous fairy tale). This notion of information content is used to provide an estimate of the above deviation which is shown to be non-vacuous in many modern settings.

cs.LG↗

arXiv2021

Near octave-spanning perfect soliton crystals in AlN microresonators

The perfect soliton crystal (PSC) was recently discovered as an extraordinary Kerr soliton state with regularly distributed soliton pulses and enhanced comb line power spaced by multiples of the cavity free spectral ranges (FSRs). The modulation of continuous-wave excitation in optical microresonators and the tunable repetition rate characteristic will significantly enhance and extend the application potential of soliton microcombs for self-referencing comb source, terahertz wave generation, and arbitrary waveform generation. However, the reported PSC spectrum is generally narrow. Here, we demonstrate the deterministic accessing of versatile perfect soliton crystals in the AlN microresonators (FSR ~374 GHz), featuring a broad spectral range up to 0.96 of an octave-span (1170-2300 nm) and terahertz repetition rates (up to ~1.87 THz). The measured 60-fs short pulses and low-noise characteristics confirms the high coherence of the PSCs

physics.optics↗

arXiv2021

ATCSpeechNet: A multilingual end-to-end speech recognition framework for air traffic control systems

In this paper, a multilingual end-to-end framework, called as ATCSpeechNet, is proposed to tackle the issue of translating communication speech into human-readable text in air traffic control (ATC) systems. In the proposed framework, we focus on integrating the multilingual automatic speech recognition (ASR) into one model, in which an end-to-end paradigm is developed to convert speech waveform into text directly, without any feature engineering or lexicon. In order to make up for the deficiency of the handcrafted feature engineering caused by ATC challenges, a speech representation learning (SRL) network is proposed to capture robust and discriminative speech representations from the raw wave. The self-supervised training strategy is adopted to optimize the SRL network from unlabeled data, and further to predict the speech features, i.e., wave-to-feature. An end-to-end architecture is improved to complete the ASR task, in which a grapheme-based modeling unit is applied to address the multilingual ASR issue. Facing the problem of small transcribed samples in the ATC domain, an unsupervised approach with mask prediction is applied to pre-train the backbone network of the ASR model on unlabeled data by a feature-to-feature process. Finally, by integrating the SRL with ASR, an end-to-end multilingual ASR framework is formulated in a supervised manner, which is able to translate the raw wave into text in one model, i.e., wave-to-text. Experimental results on the ATCSpeech corpus demonstrate that the proposed approach achieves a high performance with a very small labeled corpus and less resource consumption, only 4.20% label error rate on the 58-hour transcribed corpus. Compared to the baseline model, the proposed approach obtains over 100% relative performance improvement which can be further enhanced with the increasing of the size of the transcribed samples.

cs.CL↗

arXiv2021

DAN-Net: Dual-Domain Adaptive-Scaling Non-local Network for CT Metal Artifact Reduction

Metal implants can heavily attenuate X-rays in computed tomography (CT) scans, leading to severe artifacts in reconstructed images, which significantly jeopardize image quality and negatively impact subsequent diagnoses and treatment planning. With the rapid development of deep learning in the field of medical imaging, several network models have been proposed for metal artifact reduction (MAR) in CT. Despite the encouraging results achieved by these methods, there is still much room to further improve performance. In this paper, a novel Dual-domain Adaptive-scaling Non-local network (DAN-Net) for MAR. We correct the corrupted sinogram using adaptive scaling first to preserve more tissue and bone details as a more informative input. Then, an end-to-end dual-domain network is adopted to successively process the sinogram and its corresponding reconstructed image generated by the analytical reconstruction layer. In addition, to better suppress the existing artifacts and restrain the potential secondary artifacts caused by inaccurate results of the sinogram-domain network, a novel residual sinogram learning strategy and nonlocal module are leveraged in the proposed network model. In the experiments, the proposed DAN-Net demonstrates performance competitive with several state-of-the-art MAR methods in both qualitative and quantitative aspects.

physics.med-ph↗

arXiv2021

Gapless vortex bound states in superconducting topological semimetals

We find that the vortex bound states in superconducting topological semimetals are gapless owing to topological massless excitations in their normal states. We demonstrate this universal result in a variety of semimetals including Dirac and Weyl semimetals, threefold degenerate spin-1 fermions, spin-3/2 Rarita-Schwinger-Weyl fermion semimetals and other exotic fermion semimetals. The formation of these gapless bound states are closely related to their Andreev specular reflection and propagating Andreev modes in $π$ phase superconductor-normal metal-superconductor (SNS) junctions. We further demonstrate that these gapless states are topologically protected and can be derived from a topological pumping process.

cond-mat.supr-con↗

arXiv2021

Tactical Decision Making for Emergency Vehicles Based on A Combinational Learning Method

Increasing the response time of emergency vehicles(EVs) could lead to an immeasurable loss of property and life. On this account, tactical decision making for EVs' microscopic control remains an indispensable issue to be improved. In this paper, a rule-based avoiding strategy(AS) is devised, that CVs in the prioritized zone ahead of EV should accelerate or change their lane to avoid it. Besides, a novel DQN method with speed-adaptive compact state space (SC-DQN) is put forward to fit in EVs' high-speed feature and generalize in various road topologies. Afterward, the execution of AS feedback to the input of SC-DQN so that they joint organically as a combinational method. The following approach reveals that DRL could complement rule-based avoiding strategy in generalization, and on the contrary, the rule-based avoiding strategy could complement DRL in stability, and their combination could lead to less response time, lower collision rate and smoother trajectory.

cs.AI↗

arXiv2021

RGB-D Salient Object Detection via 3D Convolutional Neural Networks

RGB-D salient object detection (SOD) recently has attracted increasing research interest and many deep learning methods based on encoder-decoder architectures have emerged. However, most existing RGB-D SOD models conduct feature fusion either in the single encoder or the decoder stage, which hardly guarantees sufficient cross-modal fusion ability. In this paper, we make the first attempt in addressing RGB-D SOD through 3D convolutional neural networks. The proposed model, named RD3D, aims at pre-fusion in the encoder stage and in-depth fusion in the decoder stage to effectively promote the full integration of RGB and depth streams. Specifically, RD3D first conducts pre-fusion across RGB and depth modalities through an inflated 3D encoder, and later provides in-depth feature fusion by designing a 3D decoder equipped with rich back-projection paths (RBPP) for leveraging the extensive aggregation ability of 3D convolutions. With such a progressive fusion strategy involving both the encoder and decoder, effective and thorough interaction between the two modalities can be exploited and boost the detection accuracy. Extensive experiments on six widely used benchmark datasets demonstrate that RD3D performs favorably against 14 state-of-the-art RGB-D SOD approaches in terms of four key evaluation metrics. Our code will be made publicly available: https://github.com/PPOLYpubki/RD3D.

cs.CV↗

arXiv2021

Exploring Lexical Irregularities in Hypothesis-Only Models of Natural Language Inference

Natural Language Inference (NLI) or Recognizing Textual Entailment (RTE) is the task of predicting the entailment relation between a pair of sentences (premise and hypothesis). This task has been described as a valuable testing ground for the development of semantic representations, and is a key component in natural language understanding evaluation benchmarks. Models that understand entailment should encode both, the premise and the hypothesis. However, experiments by Poliak et al. revealed a strong preference of these models towards patterns observed only in the hypothesis, based on a 10 dataset comparison. Their results indicated the existence of statistical irregularities present in the hypothesis that bias the model into performing competitively with the state of the art. While recast datasets provide large scale generation of NLI instances due to minimal human intervention, the papers that generate them do not provide fine-grained analysis of the potential statistical patterns that can bias NLI models. In this work, we analyze hypothesis-only models trained on one of the recast datasets provided in Poliak et al. for word-level patterns. Our results indicate the existence of potential lexical biases that could contribute to inflating the model performance.

cs.CL↗

arXiv2021

Atomic line defects and topological superconductivity in unconventional superconductors

Topological superconductors (TSCs) are correlated quantum states with simultaneous off-diagonal long-range order and nontrivial topological invariants. They produce gapless or zero energy boundary excitations, including Majorana zero modes and chiral Majorana edge states with topologically protected phase coherence essential for fault-tolerant quantum computing. Candidate TSCs are very rare in nature. Here, we propose a novel route toward emergent quasi-one-dimensional (1D) TSCs in naturally embedded quantum structures such as atomic line defects in unconventional spin-singlet $s$-wave and $d$-wave superconductors. We show that inversion symmetry breaking and charge transfer due to the missing atoms lead to the occupation of incipient impurity bands and mixed parity spin singlet and triplet Cooper pairing of neighboring electrons traversing the line defect. Nontrivial topological invariants arise and occupy a large part of the parameter space, including the time reversal symmetry breaking Zeeman coupling due to applied magnetic field or defect-induced magnetism, creating TSCs in different topological classes with robust Majorana zero modes at both ends of the line defect. Beyond providing a novel mechanism for the recent discovery of zero-energy bound states at both ends of an atomic line defect in monolayer Fe(Te,Se) superconductors, the findings pave the way for new material realizations of the simplest and most robust 1D TSCs using embedded quantum structures in unconventional superconductors with large pairing energy gaps and high transition temperatures.

cond-mat.supr-con↗

arXiv2021

Vibrational Spectroscopic Detection of Single Virus by a Confocal Interferometric Mid-Infrared Photothermal Microscope

We report a confocal interferometric mid-infrared photothermal (MIP) microscope and its application to label-free detection of biological nanoparticles down to single virus level. We apply the interferometric scattering principle to detect the weak photothermal effect induced by infrared absorption in a viral particle. We validated this method with two kinds of viruses, namely the vesicular stomatitis virus (VSV) and poxvirus. The single virus spectra generated by MIP microscopy show high consistency in the same group, with dominant peaks contributed by the amide I and amide II vibrations. The ratio of these two peaks are significantly different between VSV and poxvirus, highlighting the potential of using MIP microscopy for label-free differentiation of viral particles. This all-optical chemical imaging method opens a new way for spectroscopic detection of biological nanoparticles in a label-free manner and may facilitate in predicting and controlling outbreaks of emerging virus strains.

physics.app-ph↗

arXiv2020

Commutative matching Rota-Baxter operators, shuffle products with decorations and matching Zinbiel algebras

The Rota-Baxter algebra and the shuffle product are both algebraic structures arising from integral operators and integral equations. Free commutative Rota-Baxter algebras provide an algebraic framework for integral equations with the simple Riemann integral operator. The Zinbiel algebras form a category in which the shuffle product algebra is the free object. Motivated by algebraic structures underlying integral equations involving multiple integral operators and kernels, we study commutative matching Rota-Baxter algebras and construct the free objects making use of the shuffle product with multiple decorations. We also construct free commutative matching Rota-Baxter algebras in a relative context, to emulate the action of the integral operators on the coefficient functions in an integral equation. We finally show that free commutative \match Rota-Baxter algebras give the free matching Zinbiel algebra, generalizing the characterization of the shuffle product algebra as the free Zinbiel algebra obtained by Loday.

math.RA↗

arXiv2020

On Loss Functions and Recurrency Training for GAN-based Speech Enhancement Systems

Recent work has shown that it is feasible to use generative adversarial networks (GANs) for speech enhancement, however, these approaches have not been compared to state-of-the-art (SOTA) non GAN-based approaches. Additionally, many loss functions have been proposed for GAN-based approaches, but they have not been adequately compared. In this study, we propose novel convolutional recurrent GAN (CRGAN) architectures for speech enhancement. Multiple loss functions are adopted to enable direct comparisons to other GAN-based systems. The benefits of including recurrent layers are also explored. Our results show that the proposed CRGAN model outperforms the SOTA GAN-based models using the same loss functions and it outperforms other non-GAN based systems, indicating the benefits of using a GAN for speech enhancement. Overall, the CRGAN model that combines an objective metric loss function with the mean squared error (MSE) provides the best performance over comparison approaches across many evaluation metrics.

eess.AS↗

arXiv2020

Directly accessing octave-spanning dissipative Kerr soliton frequency combs in an AlN microring resonator

Self-referenced dissipative Kerr solitons (DKSs) based on optical microresonators offer prominent characteristics including miniaturization, low power consumption, broad spectral range and inherent coherence for various applications such as precision measurement, communications, microwave photonics, and astronomical spectrometer calibration. To date, octave-spanning DKSs with a free spectral range (FSR) of ~1 THz have been achieved only in ultrahigh-Q silicon nitride microresonators, with elaborate wavelength control required. Here we demonstrate an octave-spanning DKS in an aluminium nitride (AlN) microresonator with moderate loaded Q (500,000) and FSR of 374 GHz. In the design, a TE00 mode and a TE10 mode are nearly degenerate and act as pump and auxiliary modes. The presence of the auxiliary resonance balances the thermal dragging effect in dissipative soliton comb formation, crucially simplifying the DKS generation with a single pump and leading to a wide single soliton access window. We experimentally demonstrate stable DKS operation with a record single soliton step (~80 pm) and octave-spanning bandwidth (1100-2300 nm) through adiabatic pump tuning and on-chip power of 340 mW. Our scheme also allows for direct creation of the DKS state with high probability and without elaborate wavelength or power schemes being required to stabilize the soliton behavior.

physics.optics↗

arXiv2020

Inflection Point: A New Perspective on Photonic Nanojets

When light propagates through the edge or middle part of microparticle's incoming interface, there is a basic rule that light converges and diverges rapidly or slowly at the output port. These two parts are referred to as region of rapid change (RRC) and region of slow change (RSC), respectively. Finding the boundary point between RRC and RSC is the key to reveal and expound this rule scientifically. Based on the correlation between light convergence-divergence and the slope of emergent light, combined with the relationship between natural logarithm and growth in physical reality and the second derivative of a function in practical significance, we determine the boundary point between RRC and RSC, namely the inflection point. From such perspective, photonic nanojet (PNJ) and near-field focusing by light irradiation on RSC and RRC, as well as the position of the inflection point under different refractive index contrast and the field distribution of light-focusing, are studied with finite-element-method-based numerical simulation and ray-optics-based theoretical analysis. By illuminating light of different field intensity ratios to the regions divided by the inflection point, we demonstrate the generation of photonic hook (PH) and the modulation of PNJ/PH in a new manner.

physics.optics↗

arXiv2020

LEARN++: Recurrent Dual-Domain Reconstruction Network for Compressed Sensing CT

Compressed sensing (CS) computed tomography has been proven to be important for several clinical applications, such as sparse-view computed tomography (CT), digital tomosynthesis and interior tomography. Traditional compressed sensing focuses on the design of handcrafted prior regularizers, which are usually image-dependent and time-consuming. Inspired by recently proposed deep learning-based CT reconstruction models, we extend the state-of-the-art LEARN model to a dual-domain version, dubbed LEARN++. Different from existing iteration unrolling methods, which only involve projection data in the data consistency layer, the proposed LEARN++ model integrates two parallel and interactive subnetworks to perform image restoration and sinogram inpainting operations on both the image and projection domains simultaneously, which can fully explore the latent relations between projection data and reconstructed images. The experimental results demonstrate that the proposed LEARN++ model achieves competitive qualitative and quantitative results compared to several state-of-the-art methods in terms of both artifact reduction and detail preservation.

physics.med-ph↗

arXiv2020

Interpreting machine learning of topological quantum phase transitions

There has been growing excitement over the possibility of employing artificial neural networks (ANNs) to gain new theoretical insight into the physics of quantum many-body problems. ``Interpretability'' remains a concern: can we understand the basis for the ANN's decision-making criteria in order to inform our theoretical understanding? ``Interpretable'' machine learning in quantum matter has to date been restricted to linear models, such as support vector machines, due to the greater difficulty of interpreting non-linear ANNs. Here we consider topological quantum phase transitions in models of Chern insulator, $\mathbb{Z}_2$ topological insulator, and $\mathbb{Z}_2$ quantum spin liquid, each using a shallow fully connected feed-forward ANN. The use of quantum loop topography, a ``domain knowledge''-guided approach to feature selection, facilitates the construction of faithful phase diagrams and semi-quantitative interpretation of the criteria in certain cases. To identify the topological phases, the ANNs learn physically meaningful features, such as topological invariants and deconfinement of loops. The interpretability in these cases suggests hope for theoretical progress based on future uses of ANN-based machine learning on quantum many-body problems.

cond-mat.dis-nn↗

arXiv2020

X-ray reprocessing in accreting pulsar GX 301-2 observed with Insight-HXMT

We investigate the absorption and emission features in observations of GX 301-2 detected with Insight-HXMT/LE in 2017-2019. At different orbital phases, we found prominent Fe Kalpha, Kbeta and Ni Kalpha lines, as well as Compton shoulders and Fe K-shell absorption edges. These features are due to the X-ray reprocessing caused by the interaction between the radiation from the source and surrounding accretion material. According to the ratio of iron lines Kalpha and Kbeta, we infer the accretion material is in a low ionisation state. We find an orbital-dependent local absorption column density, which has a large value and strong variability around the periastron. We explain its variability as a result of inhomogeneities of the accretion environment and/or instabilities of accretion processes. In addition, the variable local column density is correlated with the equivalent width of the iron Kalpha lines throughout the orbit, which suggests that the accretion material near the neutron star is spherically distributed.

astro-ph.HE↗

arXiv2020

Meta learning to classify intent and slot labels with noisy few shot examples

Recently deep learning has dominated many machine learning areas, including spoken language understanding (SLU). However, deep learning models are notorious for being data-hungry, and the heavily optimized models are usually sensitive to the quality of the training examples provided and the consistency between training and inference conditions. To improve the performance of SLU models on tasks with noisy and low training resources, we propose a new SLU benchmarking task: few-shot robust SLU, where SLU comprises two core problems, intent classification (IC) and slot labeling (SL). We establish the task by defining few-shot splits on three public IC/SL datasets, ATIS, SNIPS, and TOP, and adding two types of natural noises (adaptation example missing/replacing and modality mismatch) to the splits. We further propose a novel noise-robust few-shot SLU model based on prototypical networks. We show the model consistently outperforms the conventional fine-tuning baseline and another popular meta-learning method, Model-Agnostic Meta-Learning (MAML), in terms of achieving better IC accuracy and SL F1, and yielding smaller performance variation when noises are present.

cs.CL↗