arXiv subjects

Yu Liu

Publications and source records attributed to Yu Liu.

At least 865 records · Page 48Linked to original sources

arXiv2021

Spatio-temporal Parking Behaviour Forecasting and Analysis Before and During COVID-19

Parking demand forecasting and behaviour analysis have received increasing attention in recent years because of their critical role in mitigating traffic congestion and understanding travel behaviours. However, previous studies usually only consider temporal dependence but ignore the spatial correlations among parking lots for parking prediction. This is mainly due to the lack of direct physical connections or observable interactions between them. Thus, how to quantify the spatial correlation remains a significant challenge. To bridge the gap, in this study, we propose a spatial-aware parking prediction framework, which includes two steps, i.e. spatial connection graph construction and spatio-temporal forecasting. A case study in Ningbo, China is conducted using parking data of over one million records before and during COVID-19. The results show that the approach is superior on parking occupancy forecasting than baseline methods, especially for the cases with high temporal irregularity such as during COVID-19. Our work has revealed the impact of the pandemic on parking behaviour and also accentuated the importance of modelling spatial dependence in parking behaviour forecasting, which can benefit future studies on epidemiology and human travel behaviours.

stat.AP↗

arXiv2021

Unified Regularity Measures for Sample-wise Learning and Generalization

Fundamental machine learning theory shows that different samples contribute unequally both in learning and testing processes. Contemporary studies on DNN imply that such sample difference is rooted on the distribution of intrinsic pattern information, namely sample regularity. Motivated by the recent discovery on network memorization and generalization, we proposed a pair of sample regularity measures for both processes with a formulation-consistent representation. Specifically, cumulative binary training/generalizing loss (CBTL/CBGL), the cumulative number of correct classiffcations of the training/testing sample within training stage, is proposed to quantize the stability in memorization-generalization process; while forgetting/mal-generalizing events, i.e., the mis-classification of previously learned or generalized sample, are utilized to represent the uncertainty of sample regularity with respect to optimization dynamics. Experiments validated the effectiveness and robustness of the proposed approaches for mini-batch SGD optimization. Further applications on training/testing sample selection show the proposed measures sharing the unified computing procedure could benefit for both tasks.

cs.LG↗

arXiv2021

Real-time Keypoints Detection for Autonomous Recovery of the Unmanned Ground Vehicle

The combination of a small unmanned ground vehicle (UGV) and a large unmanned carrier vehicle allows more flexibility in real applications such as rescue in dangerous scenarios. The autonomous recovery system, which is used to guide the small UGV back to the carrier vehicle, is an essential component to achieve a seamless combination of the two vehicles. This paper proposes a novel autonomous recovery framework with a low-cost monocular vision system to provide accurate positioning and attitude estimation of the UGV during navigation. First, we introduce a light-weight convolutional neural network called UGV-KPNet to detect the keypoints of the small UGV from the images captured by a monocular camera. UGV-KPNet is computationally efficient with a small number of parameters and provides pixel-level accurate keypoints detection results in real-time. Then, six degrees of freedom pose is estimated using the detected keypoints to obtain positioning and attitude information of the UGV. Besides, we are the first to create a large-scale real-world keypoints dataset of the UGV. The experimental results demonstrate that the proposed system achieves state-of-the-art performance in terms of both accuracy and speed on UGV keypoint detection, and can further boost the 6-DoF pose estimation for the UGV.

cs.CV↗

arXiv2021

Pulsed characteristic-function measurement of a thermalizing harmonic oscillator

We present a method for the direct measurement of the Wigner characteristic function of a thermalizing harmonic oscillator that is completely inaccessible for control or measurement. The strategy employs a recently proposed probe-measurement-based scheme [Phys. Rev. Lett. 122, 110406 (2019)] which relies on the pulsed control of a two-level probe. We generalize this scheme to the case of a nonunitary time evolution of the target harmonic oscillator, describing its thermalization through contact to a finite-temperature environment, given in the form of a Lindblad master equation. This generalization is achieved using a superoperator formalism and yields analytical expressions for the direct measurement of the characteristic function, accounting for the decoherence during the measurement process.

quant-ph↗

arXiv2021

Biased tracer reconstruction with halo mass information

Plenty of crucial information about our Universe is encoded in the cosmic large-scale structure (LSS). However, the extractions of these information are usually hindered by the nonlinearities of the LSS, which can be largely alleviated by various techniques known as the reconstruction. In realistic applications, the efficiencies of these methods are always degraded by many limiting factors, a quite important one being the shot noise induced by the finite number density of biased matter tracers (i.e., luminous galaxies or dark matter halos) in observations. In this work, we explore the gains of biased tracer reconstruction achieved from halo mass information, which can suppress shot noise component and dramatically improves the cross-correlation between tracer field and dark matter. To this end, we first closely study the clustering biases and the stochasticity properties of halo fields with various number densities under different weighting schemes, i.e., the uniform, mass and optimal weightings. Then, we apply the biased tracer reconstruction method to these different weighted halo fields and investigate how linear bias and observational mass scatter affect the reconstruction performance. Our results demonstrate that halo masses are critical information for significantly improving the performance of biased tracer reconstruction, indicating a great application potential for substantially promoting the precision of cosmological measurements [especially for baryon acoustic oscillations (BAO)] in the ambitious on-going and future galaxy surveys.

astro-ph.CO↗

arXiv2021

Ergodic Numerical Approximation to Periodic Measures of Stochastic Differential Equations

In this paper, we consider numerical approximation to periodic measure of a time periodic stochastic differential equations (SDEs) under weakly dissipative condition. For this we first study the existence of the periodic measure $ρ_t$ and the large time behaviour of $\mathcal{U}(t+s,s,x) := \mathbb{E}ϕ(X_{t}^{s,x})-\intϕdρ_t,$ where $X_t^{s,x}$ is the solution of the SDEs and $ϕ$ is a test function being smooth and of polynomial growth at infinity. We prove $\mathcal{U}$ and all its spatial derivatives decay to 0 with exponential rate on time $t$ in the sense of average on initial time $s$. We also prove the existence and the geometric ergodicity of the periodic measure of the discretized semi-flow from the Euler-Maruyama scheme and moment estimate of any order when the time step is sufficiently small (uniform for all orders). We thereafter obtain that the weak error for the numerical scheme of infinite horizon is of the order $1$ in terms of the time step. We prove that the choice of step size can be uniform for all test functions $ϕ$. Subsequently we are able to estimate the average periodic measure with ergodic numerical schemes.

math.PR↗

arXiv2021

Curvature Graph Neural Network

Graph neural networks (GNNs) have achieved great success in many graph-based tasks. Much work is dedicated to empowering GNNs with the adaptive locality ability, which enables measuring the importance of neighboring nodes to the target node by a node-specific mechanism. However, the current node-specific mechanisms are deficient in distinguishing the importance of nodes in the topology structure. We believe that the structural importance of neighboring nodes is closely related to their importance in aggregation. In this paper, we introduce discrete graph curvature (the Ricci curvature) to quantify the strength of structural connection of pairwise nodes. And we propose Curvature Graph Neural Network (CGNN), which effectively improves the adaptive locality ability of GNNs by leveraging the structural property of graph curvature. To improve the adaptability of curvature to various datasets, we explicitly transform curvature into the weights of neighboring nodes by the necessary Negative Curvature Processing Module and Curvature Normalization Module. Then, we conduct numerous experiments on various synthetic datasets and real-world datasets. The experimental results on synthetic datasets show that CGNN effectively exploits the topology structure information, and the performance is improved significantly. CGNN outperforms the baselines on 5 dense node classification benchmark datasets. This study deepens the understanding of how to utilize advanced topology information and assign the importance of neighboring nodes from the perspective of graph curvature and encourages us to bridge the gap between graph theory and neural networks.

cs.LG↗

arXiv2021

Bimolecular chemistry in the ultracold regime

Advances in atomic, molecular, and optical (AMO) physics techniques allowed the cooling of simple molecules down to the ultracold regime ($\lesssim$ 1 mK), and opened the opportunities to study chemical reactions with unprecedented levels of control. This review covers recent developments in studying bimolecular chemistry at ultralow temperatures. We begin with a brief overview of methods for producing, manipulating, and detecting ultracold molecules. We then survey experimental works that exploit the controllability of ultracold molecules to probe and modify their long-range interactions. Further combining the use of physical chemistry techniques, such as mass spectrometry and ion imaging, significantly improved the detection of ultracold reactions and enabled explorations of their dynamics in the short-range. We discuss a series of studies on the reaction KRb + KRb $\rightarrow$ K$_2$ + Rb$_2$ initiated below 1 $μ$K, including the direct observation of a long-lived complex, the demonstration of product rotational state control via conserved nuclear spins, and a test of the statistical model using the complete quantum state distribution of the products.

physics.atom-ph↗

arXiv2021

Detection of Long-Lived Complexes in Ultracold Atom-Molecule Collisions

We investigate collisional loss in an ultracold mixture of $^{40}$K$^{87}$Rb molecules and $^{87}$Rb atoms, where chemical reactions between the two species are energetically forbidden. Through direct detection of the KRb$_{2}^{*}$ intermediate complexes formed from atom-molecule collisions, we show that a $1064$ nm laser source used for optical trapping of the sample can efficiently deplete the complex population via photo-excitation, an effect which can explain the universal two-body loss observed in the mixture. By monitoring the time-evolution of the KRb$_{2}^{*}$ population after a sudden reduction in the $1064$ nm laser intensity, we measure the lifetime of the complex ($0.39(6)$ ms), as well as the photo-excitation rate for $1064$ nm light ($0.50(3)$ $μ$s$^{-1}($kW/cm$^{2})^{-1}$). The observed lifetime is ${\sim}10^{5}$ times longer than recent estimates based on the Rice-Ramsperger-Kassel-Marcus statistical theory, which calls for new insight to explain such a dramatic discrepancy.

physics.atom-ph↗

arXiv2021

FNAS: Uncertainty-Aware Fast Neural Architecture Search

Reinforcement learning (RL)-based neural architecture search (NAS) generally guarantees better convergence yet suffers from the requirement of huge computational resources compared with gradient-based approaches, due to the rollout bottleneck -- exhaustive training for each sampled generation on proxy tasks. In this paper, we propose a general pipeline to accelerate the convergence of the rollout process as well as the RL process in NAS. It is motivated by the interesting observation that both the architecture and the parameter knowledge can be transferred between different experiments and even different tasks. We first introduce an uncertainty-aware critic (value function) in Proximal Policy Optimization (PPO) to utilize the architecture knowledge in previous experiments, which stabilizes the training process and reduces the searching time by 4 times. Further, an architecture knowledge pool together with a block similarity function is proposed to utilize parameter knowledge and reduces the searching time by 2 times. It is the first to introduce block-level weight sharing in RLbased NAS. The block similarity function guarantees a 100% hitting ratio with strict fairness. Besides, we show that a simply designed off-policy correction factor used in "replay buffer" in RL optimization can further reduce half of the searching time. Experiments on the Mobile Neural Architecture Search (MNAS) search space show the proposed Fast Neural Architecture Search (FNAS) accelerates standard RL-based NAS process by ~10x (e.g. ~256 2x2 TPUv2 x days / 20,000 GPU x hour -> 2,000 GPU x hour for MNAS), and guarantees better performance on various vision tasks.

cs.LG↗

arXiv2021

The impact and recovery of asteroid 2018 LA

The June 2, 2018, impact of asteroid 2018 LA over Botswana is only the second asteroid detected in space prior to impacting over land. Here, we report on the successful recovery of meteorites. Additional astrometric data refine the approach orbit and define the spin period and shape of the asteroid. Video observations of the fireball constrain the asteroid's position in its orbit and were used to triangulate the location of the fireball's main flare over the Central Kalahari Game Reserve. 23 meteorites were recovered. A consortium study of eight of these classifies Motopi Pan as a HED polymict breccia derived from howardite, cumulate and basaltic eucrite, and diogenite lithologies. Before impact, 2018 LA was a solid rock of about 156 cm diameter with high bulk density about 2.85 g/cm3, a relatively low albedo pV about 0.25, no significant opposition effect on the asteroid brightness, and an impact kinetic energy of about 0.2 kt. The orbit of 2018 LA is consistent with an origin at Vesta (or its Vestoids) and delivery into an Earth-impacting orbit via the nu_6 resonance. The impact that ejected 2018 LA in an orbit towards Earth occurred 22.8 +/- 3.8 Ma ago. Zircons record a concordant U-Pb age of 4563 +/- 11 Ma and a consistent 207Pb/206Pb age of 4563 +/- 6 Ma. A much younger Pb-Pb phosphate resetting age of 4234 +/- 41 Ma was found. From this impact chronology, we discuss what is the possible source crater of Motopi Pan and the age of Vesta's Veneneia impact basin.

astro-ph.EP↗

arXiv2021

The first-principles research on the role of surface in the heavy fermion compound CeRh$_2$Si$_2$

In the heavy fermion materials, the characteristic energy scales of many exotic strongly correlated phenomena (Kondo effect, magnetic order, superconductivity, etc.) are at milli-electron-volt order, implying that the heavy fermion materials are surface sensitive. Here, we investigate the electronic structures for Si- and Ce-terminated surfaces of CeRh$_2$Si$_2$ by first-principles methods. Our research reveals three notable impacts of surface effects on electronic structures, which are consistent with recent angle-resolved photoemission spectroscopy (ARPES) experiments. Firstly, the relaxation of surface crystal structures changes the relative position of Fermi level, adjusts the dispersion of bands and enhances the Kondo resonance. Secondly, the decrease of the hybridization between the Ce-4$f$ and conduction electrons in the surface layer leads to a weaker Kondo resonance peak and the shift of spin-orbit bands. Thirdly, the variation of crystal electric field around surface Ce atoms affects the splitting of Kondo resonance peaks, and also pushes down the lower-Hubbard bands of surface 4$f$ electrons. Moreover, we find the characteristic of bulk's lower-Hubbard bands, which was overlooked in previous works. Our investigation suggests that these surface effects are potentially important and highlighted in the future researches on properties of strongly correlated materials.

cond-mat.str-el↗

arXiv2021

Actor-Context-Actor Relation Network for Spatio-Temporal Action Localization

Localizing persons and recognizing their actions from videos is a challenging task towards high-level video understanding. Recent advances have been achieved by modeling direct pairwise relations between entities. In this paper, we take one step further, not only model direct relations between pairs but also take into account indirect higher-order relations established upon multiple elements. We propose to explicitly model the Actor-Context-Actor Relation, which is the relation between two actors based on their interactions with the context. To this end, we design an Actor-Context-Actor Relation Network (ACAR-Net) which builds upon a novel High-order Relation Reasoning Operator and an Actor-Context Feature Bank to enable indirect relation reasoning for spatio-temporal action localization. Experiments on AVA and UCF101-24 datasets show the advantages of modeling actor-context-actor relations, and visualization of attention maps further verifies that our model is capable of finding relevant higher-order relations to support action detection. Notably, our method ranks first in the AVA-Kineticsaction localization task of ActivityNet Challenge 2020, out-performing other entries by a significant margin (+6.71mAP). Training code and models will be available at https://github.com/Siyu-C/ACAR-Net.

cs.CV↗

arXiv2021

Role-Aware Modeling for N-ary Relational Knowledge Bases

N-ary relational knowledge bases (KBs) represent knowledge with binary and beyond-binary relational facts. Especially, in an n-ary relational fact, the involved entities play different roles, e.g., the ternary relation PlayCharacterIn consists of three roles, ACTOR, CHARACTER and MOVIE. However, existing approaches are often directly extended from binary relational KBs, i.e., knowledge graphs, while missing the important semantic property of role. Therefore, we start from the role level, and propose a Role-Aware Modeling, RAM for short, for facts in n-ary relational KBs. RAM explores a latent space that contains basis vectors, and represents roles by linear combinations of these vectors. This way encourages semantically related roles to have close representations. RAM further introduces a pattern matrix that captures the compatibility between the role and all involved entities. To this end, it presents a multilinear scoring function to measure the plausibility of a fact composed by certain roles and entities. We show that RAM achieves both theoretical full expressiveness and computation efficiency, which also provides an elegant generalization for approaches in binary relational KBs. Experiments demonstrate that RAM outperforms representative baselines on both n-ary and binary relational datasets.

cs.AI↗

arXiv2021

Integrating Information Theory and Adversarial Learning for Cross-modal Retrieval

Accurately matching visual and textual data in cross-modal retrieval has been widely studied in the multimedia community. To address these challenges posited by the heterogeneity gap and the semantic gap, we propose integrating Shannon information theory and adversarial learning. In terms of the heterogeneity gap, we integrate modality classification and information entropy maximization adversarially. For this purpose, a modality classifier (as a discriminator) is built to distinguish the text and image modalities according to their different statistical properties. This discriminator uses its output probabilities to compute Shannon information entropy, which measures the uncertainty of the modality classification it performs. Moreover, feature encoders (as a generator) project uni-modal features into a commonly shared space and attempt to fool the discriminator by maximizing its output information entropy. Thus, maximizing information entropy gradually reduces the distribution discrepancy of cross-modal features, thereby achieving a domain confusion state where the discriminator cannot classify two modalities confidently. To reduce the semantic gap, Kullback-Leibler (KL) divergence and bi-directional triplet loss are used to associate the intra- and inter-modality similarity between features in the shared space. Furthermore, a regularization term based on KL-divergence with temperature scaling is used to calibrate the biased label classifier caused by the data imbalance issue. Extensive experiments with four deep models on four benchmarks are conducted to demonstrate the effectiveness of the proposed approach.

cs.CV↗

arXiv2021

Self-supervised Video Representation Learning by Context and Motion Decoupling

A key challenge in self-supervised video representation learning is how to effectively capture motion information besides context bias. While most existing works implicitly achieve this with video-specific pretext tasks (e.g., predicting clip orders, time arrows, and paces), we develop a method that explicitly decouples motion supervision from context bias through a carefully designed pretext task. Specifically, we take the keyframes and motion vectors in compressed videos (e.g., in H.264 format) as the supervision sources for context and motion, respectively, which can be efficiently extracted at over 500 fps on the CPU. Then we design two pretext tasks that are jointly optimized: a context matching task where a pairwise contrastive loss is cast between video clip and keyframe features; and a motion prediction task where clip features, passed through an encoder-decoder network, are used to estimate motion features in a near future. These two tasks use a shared video backbone and separate MLP heads. Experiments show that our approach improves the quality of the learned video representation over previous works, where we obtain absolute gains of 16.0% and 11.1% in video retrieval recall on UCF101 and HMDB51, respectively. Moreover, we find the motion prediction to be a strong regularization for video networks, where using it as an auxiliary task improves the accuracy of action recognition with a margin of 7.4%~13.8%.

cs.CV↗

arXiv2021

MulDE: Multi-teacher Knowledge Distillation for Low-dimensional Knowledge Graph Embeddings

Link prediction based on knowledge graph embeddings (KGE) aims to predict new triples to automatically construct knowledge graphs (KGs). However, recent KGE models achieve performance improvements by excessively increasing the embedding dimensions, which may cause enormous training costs and require more storage space. In this paper, instead of training high-dimensional models, we propose MulDE, a novel knowledge distillation framework, which includes multiple low-dimensional hyperbolic KGE models as teachers and two student components, namely Junior and Senior. Under a novel iterative distillation strategy, the Junior component, a low-dimensional KGE model, asks teachers actively based on its preliminary prediction results, and the Senior component integrates teachers' knowledge adaptively to train the Junior component based on two mechanisms: relation-specific scaling and contrast attention. The experimental results show that MulDE can effectively improve the performance and training speed of low-dimensional KGE models. The distilled 32-dimensional model is competitive compared to the state-of-the-art high-dimensional methods on several widely-used datasets.

cs.AI↗

arXiv2021

Lifelong Person Re-Identification via Adaptive Knowledge Accumulation

Person ReID methods always learn through a stationary domain that is fixed by the choice of a given dataset. In many contexts (e.g., lifelong learning), those methods are ineffective because the domain is continually changing in which case incremental learning over multiple domains is required potentially. In this work we explore a new and challenging ReID task, namely lifelong person re-identification (LReID), which enables to learn continuously across multiple domains and even generalise on new and unseen domains. Following the cognitive processes in the human brain, we design an Adaptive Knowledge Accumulation (AKA) framework that is endowed with two crucial abilities: knowledge representation and knowledge operation. Our method alleviates catastrophic forgetting on seen domains and demonstrates the ability to generalize to unseen domains. Correspondingly, we also provide a new and large-scale benchmark for LReID. Extensive experiments demonstrate our method outperforms other competitors by a margin of 5.8% mAP in generalising evaluation.

cs.CV↗