arXiv subjects

Yu Liu

Publications and source records attributed to Yu Liu.

At least 1,099 records · Page 61Linked to original sources

arXiv2017

Learning Deep Features via Congenerous Cosine Loss for Person Recognition

Person recognition aims at recognizing the same identity across time and space with complicated scenes and similar appearance. In this paper, we propose a novel method to address this task by training a network to obtain robust and representative features. The intuition is that we directly compare and optimize the cosine distance between two features - enlarging inter-class distinction as well as alleviating inner-class variance. We propose a congenerous cosine loss by minimizing the cosine distance between samples and their cluster centroid in a cooperative way. Such a design reduces the complexity and could be implemented via softmax with normalized inputs. Our method also differs from previous work in person recognition that we do not conduct a second training on the test subset. The identity of a person is determined by measuring the similarity from several body regions in the reference set. Experimental results show that the proposed approach achieves better classification accuracy against previous state-of-the-arts.

cs.CV↗

arXiv2017

Zoom Out-and-In Network with Recursive Training for Object Proposal

In this paper, we propose a zoom-out-and-in network for generating object proposals. We utilize different resolutions of feature maps in the network to detect object instances of various sizes. Specifically, we divide the anchor candidates into three clusters based on the scale size and place them on feature maps of distinct strides to detect small, medium and large objects, respectively. Deeper feature maps contain region-level semantics which can help shallow counterparts to identify small objects. Therefore we design a zoom-in sub-network to increase the resolution of high level features via a deconvolution operation. The high-level features with high resolution are then combined and merged with low-level features to detect objects. Furthermore, we devise a recursive training pipeline to consecutively regress region proposals at the training stage in order to match the iterative regression at the testing stage. We demonstrate the effectiveness of the proposed method on ILSVRC DET and MS COCO datasets, where our algorithm performs better than the state-of-the-arts in various evaluation metrics. It also increases average precision by around 2% in the detection system.

cs.CV↗

arXiv2017

Hearts of twin Cotorsion pairs on extriangulated categories

In this article, we study the heart of a cotorsion pairs on an exact category and a triangulated category in a unified meathod, by means of the notion of an extriangulated category. We prove that the heart is abelian, and construct a cohomological functor to the heart. If the extriangulated category has enough projectives, this functor gives an equivalence between the heart and the category of coherent functors over the coheart modulo projectives. We also show how an n-cluster tilting subcategory of an extriangulated category gives rise to a family of cotorsion pairs with equivalent hearts.

math.CT↗

arXiv2017

Ultracold Molecular Assembly

Chemical reactions can be surprisingly efficient at ultracold temperatures ( < 1mK) due to the wave nature of atoms and molecules. The study of reactions in the ultracold regime is a new research frontier enabled by cooling and trapping techniques developed in atomic and molecular physics. In addition, ultracold molecular gases that offer diverse molecular internal states and large electric dipolar interactions are sought after for studies of strongly interacting many-body quantum physics. Here we propose a new approach for producing ultracold molecules in the absolute internal and motional quantum ground state, where single molecules are assembled one by one from individual atoms. The scheme involves laser cooling, optical trapping, Raman sideband cooling, and coherent molecular state transfer. As a crucial initial step, we demonstrate quantum control of constituent atoms, including 3D ground-state cooling of a single Cs atom, in a simple apparatus. As laser technology advances to shorter wavelengths, additional atoms will be amenable to laser-cooling, allowing more diverse, and eventually more complex, molecules to be assembled with full quantum control.

physics.atom-ph↗

arXiv2016

Channel Measurements and Models for High-Speed Train Wireless Communication Systems in Tunnel Scenarios: A Survey

The rapid developments of high-speed trains (HSTs) introduce new challenges to HST wireless communication systems. Realistic HST channel models play a critical role in designing and evaluating HST communication systems. Due to the length limitation, bounding of tunnel itself, and waveguide effect, channel characteristics in tunnel scenarios are very different from those in other HST scenarios. Therefore, accurate tunnel channel models considering both large-scale and small-scale fading characteristics are essential for HST communication systems. Moreover, certain characteristics of tunnel channels have not been investigated sufficiently. This article provides a comprehensive review of the measurement campaigns in tunnels and presents some tunnel channel models using various modeling methods. Finally, future directions in HST tunnel channel measurements and modeling are discussed.

cs.NI↗

arXiv2016

Superconducting Order from Disorder in 2H-TaSe$_{2-x}$S$_{x}$ (0$\leq$x$\leq$2)

We report on the emergence of robust superconducting order in single crystal alloys of 2H-TaSe$_{2-x}$S$_{x}$ (0$\leq$x$\leq$2) . The critical temperature of the alloy is surprisingly higher than that of the two end compounds TaSe$_{2}$ and TaS$_{2}$. The evolution of superconducting critical temperature T$_{c} (x)$ correlates with the full width at half maximum of the Bragg peaks and with the linear term of the high temperature resistivity. The conductivity of the crystals near the middle of the alloy series is higher or similar than that of either one of the end members 2H-TaSe$_{2}$ and/or 2H-TaS$_{2}$. It is known that in these materials superconductivity (SC) is in close competition with charge density wave (CDW) order. We interpret our experimental findings in a picture where disorder tilts this balance in favor of superconductivity by destroying the CDW order.

cond-mat.supr-con↗

arXiv2016

Adiabatic quantum parameter amplification for generic robust quantum sensing

Quantum enhanced sensing provides a powerful tool for the precise measurement of physical parameters that is applicable in many areas of science and technology. The achievable gain in sensitivity is largely limited by the influence of noise and decoherence. Here, we propose a paradigm of adiabatic quantum parameter amplification to overcome this limitation. We demonstrate that it allows to achieve generic robust quantum sensing, namely it is robust against noise that may even acting on the same degree of freedom as the field. Furthermore, the proposal achieves entanglement-free Heisenberg limit sensitivity that surpasses the limit of classical statistics.

quant-ph↗

arXiv2016

On the Exploration of Convolutional Fusion Networks for Visual Recognition

Despite recent advances in multi-scale deep representations, their limitations are attributed to expensive parameters and weak fusion modules. Hence, we propose an efficient approach to fuse multi-scale deep representations, called convolutional fusion networks (CFN). Owing to using 1$\times$1 convolution and global average pooling, CFN can efficiently generate the side branches while adding few parameters. In addition, we present a locally-connected fusion module, which can learn adaptive weights for the side branches and form a discriminatively fused feature. CFN models trained on the CIFAR and ImageNet datasets demonstrate remarkable improvements over the plain CNNs. Furthermore, we generalize CFN to three new tasks, including scene recognition, fine-grained recognition and image retrieval. Our experiments show that it can obtain consistent improvements towards the transferring tasks.

cs.CV↗

arXiv2016

Distinct surface and bulk charge density waves in ultrathin 1T-TaS2

We employ low-frequency Raman spectroscopy to study the nearly commensurate (NC) to commensurate (C) charge density wave (CDW) transition in 1T-TaS2 ultrathin flakes protected from oxidation. We identify new modes originating from C phase CDW phonons that are distinct from those seen in bulk 1T-TaS2. We attribute these to CDW modes from the surface layers. By monitoring individual modes with temperature, we find that surfaces undergo a separate, low-hysteresis NC-C phase transition that is decoupled from the transition in the bulk layers. This indicates the activation of a secondary phase nucleation process in the limit of weak interlayer interaction, which can be understood from energy considerations.

cond-mat.mtrl-sci↗

arXiv2016

POI: Multiple Object Tracking with High Performance Detection and Appearance Feature

Detection and learning based appearance feature play the central role in data association based multiple object tracking (MOT), but most recent MOT works usually ignore them and only focus on the hand-crafted feature and association algorithms. In this paper, we explore the high-performance detection and deep learning based appearance feature, and show that they lead to significantly better MOT results in both online and offline setting. We make our detection and appearance feature publicly available. In the following part, we first summarize the detection and appearance feature, and then introduce our tracker named Person of Interest (POI), which has both online and offline version.

cs.CV↗

arXiv2016

Simulating heavy fermion physics in optical lattice: Periodic Anderson model with harmonic trapping potential

Periodic Anderson model (PAM), where local electron orbitals interplay with itinerant electronic carriers, plays an essential role in our understanding on heavy fermion materials. Motivated by recent proposal of simulating Kondo lattice model (KLM) in terms of alkaline-earth metal atoms, we make a further step toward simulation of PAM, which includes crucial charge/valence fluctuation of local f-electron beyond purely low-energy spin fluctuation in KLM. To realize PAM, transition induced by suitable laser between electronic excited and ground state of alkaline-earth metal atoms ($^{1}S_{0}$$\rightleftharpoons$$^{3}P_{0}$) is introduced, and it leads to effective hybridization between local electron and conduction electron in PAM. Generally, the $SU(N)$ version of PAM can be realized by our proposal, which gives a unique opportunity to detect large-$N$ physics without complexity in realistic materials. In the present work, high temperature physical feature of standard ($SU(2)$) PAM with harmonic trapping potential is detailed analyzed by quantum Monte Carlo and dynamic mean-field theory. Indications for near-future experiments are provided. We expect our theoretical proposal and (hopefully) forthcoming experiment will deepen our understanding on heavy fermion systems and at the same time triggers further studies on related Mott physics, quantum criticality and non-trivial topology in both inhomogeneous and nonequilibrium realm.

cond-mat.quant-gas↗

arXiv2016

Crafting GBD-Net for Object Detection

The visual cues from multiple support regions of different sizes and resolutions are complementary in classifying a candidate box in object detection. Effective integration of local and contextual visual cues from these regions has become a fundamental problem in object detection. In this paper, we propose a gated bi-directional CNN (GBD-Net) to pass messages among features from different support regions during both feature learning and feature extraction. Such message passing can be implemented through convolution between neighboring support regions in two directions and can be conducted in various layers. Therefore, local and contextual visual patterns can validate the existence of each other by learning their nonlinear relationships and their close interactions are modeled in a more complex way. It is also shown that message passing is not always helpful but dependent on individual samples. Gated functions are therefore needed to control message transmission, whose on-or-offs are controlled by extra visual evidence from the input sample. The effectiveness of GBD-Net is shown through experiments on three object detection datasets, ImageNet, Pascal VOC2007 and Microsoft COCO. This paper also shows the details of our approach in wining the ImageNet object detection challenge of 2016, with source code provided on \url{https://github.com/craftGBD/craftGBD}.

cs.CV↗

arXiv2016

Atomic Lattice Disorder in Charge Density Wave Phases of Exfoliated Dichalcogenides (1T-TaS2)

Charge density waves (CDW) and their concomitant periodic lattice distortions (PLD) govern the electronic properties in many layered transition-metal dichalcogenides. In particular, 1T-TaS2 undergoes a metal-to-insulator phase transition as the PLD becomes commensurate with the crystal lattice. Here we directly image PLDs of the nearly-commensurate (NC) and commensurate (C) phases in thin exfoliated 1T-TaS2 using atomic resolution scanning transmission electron microscopy at room and cryogenic temperature. At low temperatures, we observe commensurate PLD superstructures, suggesting ordering of the CDWs both in- and out-of-plane. In addition, we discover stacking transitions in the atomic lattice that occur via one bond length shifts. Interestingly, the NC PLDs exist inside both the stacking domains and their boundaries. Transitions in stacking order are expected to create fractional shifts in the CDW between layers and may be another route to manipulate electronic phases in layered dichalcogenides.

cond-mat.mtrl-sci↗

arXiv2016

Three-dimensional bulk electronic structure of the Kondo lattice CeIn3 revealed by photoemission

We show the three-dimensional electronic structure of the Kondo lattice CeIn3 using soft x-ray angle resolved photoemission spectroscopy in the paramagnetic state. For the first time, we have directly observed the three-dimensional topology of the Fermi surface of CeIn3 by photoemission. The Fermi surface has a complicated hole pocket centred at the Γ-Z line and an elliptical electron pocket centred at the R point of the Brillouin zone. Polarization and photon-energy dependent photoemission results both indicate the nearly localized nature of the 4f electrons in CeIn3, consistent with the theoretical prediction by means of the combination of density functional theory and single-site dynamical meanfield theory. Those results illustrate that the f electrons of CeIn3, which is the parent material of CeMIn5 compounds, are closer to the localized description than the layered CeMIn5 compounds.

cond-mat.str-el↗

arXiv2016

Spatial interferences in the electron transport of heavy fermion materials

The scanning tunneling microscopy/spectroscopy and the point contact spectroscopy represent one of the major progresses in recent heavy fermion research. Both have revealed important information on the composite nature of the emergent heavy electron quasiparticles. However, a detailed and thorough microscopic understanding of the similarities and differences in the underlying physical processes of these techniques is still lacking. Here we study the electron transport in the normal state of the periodic Anderson lattice by using the Keldysh nonequilibrium Green's function technique. In addition to the well-known Fano interference between the conduction and f-electron channels, our results further reveal the effect of spatial interference between different spatial paths at the interface on the differential conductance and their interesting interplay with the band features such as the hybridization gap and the Van Hove singularity. We find that the spatial interference leads to a weighted average in the momentum space for the electron transport and could cause suppression of the electronic band features under certain circumstances. In particular, it reduces the capability of probing the f-electron spectral weight near the edges of the hybridization gap for large interface depending on the Fermi surface of the lead. Our results indicates an intrinsic inefficiency of the point contact spectroscopy in probing the f-electrons.

cond-mat.str-el↗

arXiv2016

Phase diagram of Kondo-Heisenberg model on honeycomb lattice with geometrical frustration

We calculated the phase diagram of the Kondo-Heisenberg model on two-dimensional honeycomb lattice with both nearest-neighbor and next-nearest-neighbor antiferromagnetic spin exchanges, to investigate the interplay between RKKY and Kondo interactions at presence of magnetic frustration. Within a mean-field decoupling technology in slave-fermion representation, we derived the zero-temperature phase diagram as a function of Kondo coupling $J_k$ and frustration strength $Q$. The geometrical frustration can destroy the magnetic order, driving the original antiferromagnetic (AF) phase to non-magnetic valence bond state (VBS). In addition, we found two distinct VBS. As $J_k$ is increased, a phase transition from AF to Kondo paramagnetic (KP) phase occurs, without the intermediate phase coexisting AF order with Kondo screening found in square lattice systems. In the KP phase, the enhancement of frustration weakens the Kondo screening effect, resulting in a phase transition from KP to VBS. We also found a process to recover the AF order from VBS by increasing $J_k$ in a wide range of frustration strength. Our work may provide deeper understanding for the phase transitions in heavy-fermion materials, particularly for those exhibiting triangular frustration.

cond-mat.str-el↗

arXiv2016

Understanding doping, vacancy, lattice stability and superconductivity in KxFe2-ySe2

Metal-intercalated iron selenides are a class of superconductors that have received much attention but are less understood in comparison with their FeAs-based counterparts. Here, the controversial issues such as Fe vacancy, the real phase responsible for superconductivity, and lattice stability have been addressed based on first-principles calculations. New insights into the distinct features in terms of carrier doping have been revealed.

cond-mat.supr-con↗

arXiv2016

Large Range Manipulation of Exciton Species in Monolayer WS2

Unconventional emissions from exciton and trion in monolayer WS2 are studied by photoexcitation. Excited by 532nm laser beam, the carrier species in the monolayer WS2 are affected by the excess electrons escaping from photoionization of donor impurity, the concentration of which varies with different locations of the sample. Simply increasing the excitation power at room temperature, the excess electron and thus the intensity ratio of excited trion and exciton can be continuously tuned over a large range from 0.1 to 7.7. Furthermore, this intensity ratio can also be manipulated by varying temperature. However, in this way the resonance energy of the exciton and trion show red-shifts with increasing temperature due to electron-phonon coupling. The binding energy of the trion is determined to be ~23meV and independent to temperature, indicating strong Coulomb interaction of carriers in such 2D materials.

cond-mat.mes-hall↗