arXiv subjects

Yu Liu

Publications and source records attributed to Yu Liu.

At least 757 records · Page 42Linked to original sources

arXiv2022

Optimizing Replacement Policies for Content Delivery Network Caching: Beyond Belady to Attain A Seemingly Unattainable Byte Miss Ratio

When facing objects/files of differing sizes in content delivery networks (CDNs) caches, pursuing an optimal object miss ratio (OMR) by approximating Belady no longer ensures an optimal byte miss ratio (BMR), creating confusion about how to achieve a superior BMR in CDNs. To address this issue, we experimentally observe that there exists a time window to delay the eviction of the object with the longest reuse distance to improve BMR without increasing OMR. As a result, we introduce a deep reinforcement learning (RL) model to capture this time window by dynamically monitoring the changes in OMR and BMR, and implementing a BMR-friendly policy in the time window. Based on this policy, we propose a Belady and Size Eviction (LRU-BaSE) algorithm, reducing BMR while maintaining OMR. To make LRU-BaSE efficient and practical, we address the feedback delay problem of RL with a two-pronged approach. On the one hand, our observation of a rear section of the LRU cache queue containing most of the eviction candidates allows LRU-BaSE to shorten the decision region. On the other hand, the request distribution on CDNs makes it feasible to divide the learning region into multiple sub-regions that are each learned with reduced time and increased accuracy. In real CDN systems, compared to LRU, LRU-BaSE can reduce "backing to OS" traffic and access latency by 30.05\% and 17.07\%, respectively, on average. The results on the simulator confirm that LRU-BaSE outperforms the state-of-the-art cache replacement policies, where LRU-BaSE's BMR is 0.63\% and 0.33\% less than that of Belady and Practical Flow-based Offline Optimal (PFOO), respectively, on average. In addition, compared to Learning Relaxed Belady (LRB), LRU-BaSE can yield relatively stable performance when facing workload drift.

cs.NI↗

arXiv2022

Directed search for continuous gravitational waves from the possible kilonova remnant G4.8+6.2

G4.8+6.2 was proposed as a possible kilonova remnant associated with the Korean guest star of AD 1163 in our Milky Way galaxy. Its age is about 860 years according to the historical record. If a neutron star was left in the center of G4.8+6.2, this young neutron star may radiate strong continuous gravitational waves, which could beat the indirect age-based upper limit with current LIGO sensitivity. In this work, we searched such continuous gravitational waves in the frequency band $20-1500 \mathrm{~Hz}$. This search used two days of LIGO O3b data from the Hanford and Livingston detectors. While no signal was found, we placed upper limits on the gravitational wave strain. For comparison, we also showed the latest results of all-sky searches obtained with various search pipelines. With upgrading of the gravitational wave detectors, it will provide the opportunity to see whether a black hole or a neutron star is harbored inside G4.8+6.2.

astro-ph.HE↗

arXiv2022

Hyperbolic Hierarchical Contrastive Hashing

Hierarchical semantic structures, naturally existing in real-world datasets, can assist in capturing the latent distribution of data to learn robust hash codes for retrieval systems. Although hierarchical semantic structures can be simply expressed by integrating semantically relevant data into a high-level taxon with coarser-grained semantics, the construction, embedding, and exploitation of the structures remain tricky for unsupervised hash learning. To tackle these problems, we propose a novel unsupervised hashing method named Hyperbolic Hierarchical Contrastive Hashing (HHCH). We propose to embed continuous hash codes into hyperbolic space for accurate semantic expression since embedding hierarchies in hyperbolic space generates less distortion than in hyper-sphere space and Euclidean space. In addition, we extend the K-Means algorithm to hyperbolic space and perform the proposed hierarchical hyperbolic K-Means algorithm to construct hierarchical semantic structures adaptively. To exploit the hierarchical semantic structures in hyperbolic space, we designed the hierarchical contrastive learning algorithm, including hierarchical instance-wise and hierarchical prototype-wise contrastive learning. Extensive experiments on four benchmark datasets demonstrate that the proposed method outperforms the state-of-the-art unsupervised hashing methods. Codes will be released.

cs.CV↗

arXiv2022

Safety-Enhanced Autonomous Driving Using Interpretable Sensor Fusion Transformer

Large-scale deployment of autonomous vehicles has been continually delayed due to safety concerns. On the one hand, comprehensive scene understanding is indispensable, a lack of which would result in vulnerability to rare but complex traffic situations, such as the sudden emergence of unknown objects. However, reasoning from a global context requires access to sensors of multiple types and adequate fusion of multi-modal sensor signals, which is difficult to achieve. On the other hand, the lack of interpretability in learning models also hampers the safety with unverifiable failure causes. In this paper, we propose a safety-enhanced autonomous driving framework, named Interpretable Sensor Fusion Transformer(InterFuser), to fully process and fuse information from multi-modal multi-view sensors for achieving comprehensive scene understanding and adversarial event detection. Besides, intermediate interpretable features are generated from our framework, which provide more semantics and are exploited to better constrain actions to be within the safe sets. We conducted extensive experiments on CARLA benchmarks, where our model outperforms prior methods, ranking the first on the public CARLA Leaderboard. Our code will be made available at https://github.com/opendilab/InterFuser

cs.CV↗

arXiv2022

Node-line Dirac semimetal manipulated by Kondo mechanism in nonsymmorphic CePt$_2$Si$_2$

Dirac node lines (DNLs) are characterized by Dirac-type linear crossings between valence and conduction bands along one-dimensional node lines in the Brillouin zone (BZ). Spin-orbit coupling (SOC) usually shifts the degeneracy at the crossings thus destroys DNLs, and so far the reported DNLs in a few materials are non-interacting type, making the search for robust interacting DNLs in real materials appealing. Here, via first-principle calculations, we reveal that Kondo interaction together with nonsymmorphic lattice symmetries can drive a robust interacting DNLs in a Kondo semimetal CePt_2Si_2, and the feature of DNLs can be significantly manipulated by Kondo behavior in different temperature regions. Based on the density function theory combining dynamical mean-field theory (DFT+DMFT), we predict a transition to Kondo-coherent state at coherent temperature T_coh= 80 K upon cooling, verified by temperature dependence of Ce-4f self-energy, Kondo resonance peak, magnetic susceptibility and momentum-resolved spectral. Below T_coh, well-resolved narrow heavy-fermion bands emerge near the Fermi level, constructing clearly visualized interacting DNLs locating at the BZ boundary, in which the Dirac fermions have strongly enhanced effective mass and reduced velocity. In contrast, above a crossover temperature T_KS =600 K, the destruction of local Kondo screening drives non-interacting DNLs which are comprised by light conduction electrons at the same location. These DNLs are protected by lattice nonsymmorphic symmetries thus robust under intrinsic strong SOC. Our proposal of DNLs which can be significantly manipulated according to Kondo behavior provides an unique realization of interacting Dirac semimetals in real strongly correlated materials, and serves as a convenient platform to investigate the effect of electronic correlations on topological materials.

cond-mat.str-el↗

arXiv2022

ACE: Cooperative Multi-agent Q-learning with Bidirectional Action-Dependency

Multi-agent reinforcement learning (MARL) suffers from the non-stationarity problem, which is the ever-changing targets at every iteration when multiple agents update their policies at the same time. Starting from first principle, in this paper, we manage to solve the non-stationarity problem by proposing bidirectional action-dependent Q-learning (ACE). Central to the development of ACE is the sequential decision-making process wherein only one agent is allowed to take action at one time. Within this process, each agent maximizes its value function given the actions taken by the preceding agents at the inference stage. In the learning phase, each agent minimizes the TD error that is dependent on how the subsequent agents have reacted to their chosen action. Given the design of bidirectional dependency, ACE effectively turns a multiagent MDP into a single-agent MDP. We implement the ACE framework by identifying the proper network representation to formulate the action dependency, so that the sequential decision process is computed implicitly in one forward pass. To validate ACE, we compare it with strong baselines on two MARL benchmarks. Empirical experiments demonstrate that ACE outperforms the state-of-the-art algorithms on Google Research Football and StarCraft Multi-Agent Challenge by a large margin. In particular, on SMAC tasks, ACE achieves 100% success rate on almost all the hard and super-hard maps. We further study extensive research problems regarding ACE, including extension, generalization, and practicability. Code is made available to facilitate further research.

cs.LG↗

arXiv2022

Dimensionality-Varying Diffusion Process

Diffusion models, which learn to reverse a signal destruction process to generate new data, typically require the signal at each step to have the same dimension. We argue that, considering the spatial redundancy in image signals, there is no need to maintain a high dimensionality in the evolution process, especially in the early generation phase. To this end, we make a theoretical generalization of the forward diffusion process via signal decomposition. Concretely, we manage to decompose an image into multiple orthogonal components and control the attenuation of each component when perturbing the image. That way, along with the noise strength increasing, we are able to diminish those inconsequential components and thus use a lower-dimensional signal to represent the source, barely losing information. Such a reformulation allows to vary dimensions in both training and inference of diffusion models. Extensive experiments on a range of datasets suggest that our approach substantially reduces the computational cost and achieves on-par or even better synthesis performance compared to baseline methods. We also show that our strategy facilitates high-resolution image synthesis and improves FID of diffusion model trained on FFHQ at $1024\times1024$ resolution from 52.40 to 10.46. Code and models will be made publicly available.

cs.LG↗

arXiv2022

Reciprocal phase transition-enabled electro-optic modulation

Electro-optic (EO) modulation is a well-known and essential topic in the field of communications and sensing. Its ultrahigh efficiency is unprecedentedly desired in the current green and data era. However, dramatically increasing the modulation efficiency is difficult due to the monotonic mapping relationship between the electrical signal and modulated optical signal. Here, a new mechanism termed phase-transition EO modulation is revealed from the reciprocal transition between two distinct phase planes arising from the bifurcation. Remarkably, a monolithically integrated mode-locked laser (MLL) is implemented as a prototype. A 24.8-GHz radio-frequency signal is generated and modulated, achieving a modulation energy efficiency of 3.06 fJ/bit improved by about four orders of magnitude and a contrast ratio exceeding 50 dB. Thus, MLL-based phase-transition EO modulation is characterised by ultrahigh modulation efficiency and ultrahigh contrast ratio, as experimentally proved in radio-over-fibre and underwater acoustic-sensing systems. This phase-transition EO modulation opens a new avenue for green communication and ubiquitous connections.

physics.optics↗

arXiv2022

Teach-DETR: Better Training DETR with Teachers

In this paper, we present a novel training scheme, namely Teach-DETR, to learn better DETR-based detectors from versatile teacher detectors. We show that the predicted boxes from teacher detectors are effective medium to transfer knowledge of teacher detectors, which could be either RCNN-based or DETR-based detectors, to train a more accurate and robust DETR model. This new training scheme can easily incorporate the predicted boxes from multiple teacher detectors, each of which provides parallel supervisions to the student DETR. Our strategy introduces no additional parameters and adds negligible computational cost to the original detector during training. During inference, Teach-DETR brings zero additional overhead and maintains the merit of requiring no non-maximum suppression. Extensive experiments show that our method leads to consistent improvement for various DETR-based detectors. Specifically, we improve the state-of-the-art detector DINO with Swin-Large backbone, 4 scales of feature maps and 36-epoch training schedule, from 57.8% to 58.9% in terms of mean average precision on MSCOCO 2017 validation set. Code will be available at https://github.com/LeonHLJ/Teach-DETR.

cs.CV↗

arXiv2022

Experimental demonstration of topological bounds in quantum metrology

Quantum metrology is deeply connected to quantum geometry, through the fundamental notion of quantum Fisher information. Inspired by advances in topological matter, it was recently suggested that the Berry curvature and Chern numbers of band structures can dictate strict lower bounds on metrological properties, hence establishing a strong connection between topology and quantum metrology. In this work, we provide a first experimental verification of such topological bounds, by performing optimal quantum multi-parameter estimation and achieving the best possible measurement precision. By emulating the band structure of a Chern insulator, we experimentally determine the metrological potential across a topological phase transition, and demonstrate strong enhancement in the topologically non-trivial regime. Our work opens the door to metrological applications empowered by topology, with potential implications for quantum many-body systems.

quant-ph↗

arXiv2022

Constructing multi-level urban clusters based on population distributions and interactions

A city (or an urban cluster) is not an isolated spatial unit, but a combination of areas with closely linked socio-economic activities. However, so far, we lack a consistent and quantitative approach to define multi-level urban clusters through these socio-economic connections. Here, using granular population distribution and flow data from China, we propose a bottom-up aggregation approach to quantify urban clusters at multiple spatial scales. We reveal six 'phases' (i.e., levels) in the population density-flow diagram, each of which corresponds to a spatial configuration of urban clusters from large to small. Besides, our results show that Zipf's law appears only after the fifth level, confirming the spatially dependent nature of urban laws. Our approach does not need pre-defined administrative boundaries and can be applied effectively on a global scale.

physics.soc-ph↗

arXiv2022

PolyBuilding: Polygon Transformer for End-to-End Building Extraction

We present PolyBuilding, a fully end-to-end polygon Transformer for building extraction. PolyBuilding direct predicts vector representation of buildings from remote sensing images. It builds upon an encoder-decoder transformer architecture and simultaneously outputs building bounding boxes and polygons. Given a set of polygon queries, the model learns the relations among them and encodes context information from the image to predict the final set of building polygons with fixed vertex numbers. Corner classification is performed to distinguish the building corners from the sampled points, which can be used to remove redundant vertices along the building walls during inference. A 1-d non-maximum suppression (NMS) is further applied to reduce vertex redundancy near the building corners. With the refinement operations, polygons with regular shapes and low complexity can be effectively obtained. Comprehensive experiments are conducted on the CrowdAI dataset. Quantitative and qualitative results show that our approach outperforms prior polygonal building extraction methods by a large margin. It also achieves a new state-of-the-art in terms of pixel-level coverage, instance-level precision and recall, and geometry-level properties (including contour regularity and polygon complexity).

cs.CV↗

arXiv2022

SOM-Net: Unrolling the Subspace-based Optimization for Solving Full-wave Inverse Scattering Problems

In this paper, an unrolling algorithm of the iterative subspace-based optimization method (SOM) is proposed for solving full-wave inverse scattering problems (ISPs). The unrolling network, named SOM-Net, inherently embeds the Lippmann- Schwinger physical model into the design of network structures. The SOM-Net takes the deterministic induced current and the raw permittivity image obtained from back-propagation (BP) as the input. It then updates the induced current and the permittivity successively in sub-network blocks of the SOM- Net by imitating iterations of the SOM. The final output of the SOM-Net is the full predicted induced current, from which the scattered field and the permittivity image can also be deduced analytically. The parameters of the SOM-Net are optimized in a supervised manner with the total loss to simultaneously ensure the consistency of the induced current, the scattered field, and the permittivity in the governing equations. Numerical tests on both synthetic and experimental data verify the superior performance of the proposed SOM-Net over typical ones. The results on challenging examples like scatterers with tough profiles or high permittivity demonstrate the good generalization ability of the SOM-Net. With the use of deep unrolling technology, this work builds a bridge between traditional iterative methods and deep learning methods for solving ISPs.

eess.SP↗

arXiv2022

Deep Learning for Instance Retrieval: A Survey

In recent years a vast amount of visual content has been generated and shared from many fields, such as social media platforms, medical imaging, and robotics. This abundance of content creation and sharing has introduced new challenges, particularly that of searching databases for similar content-Content Based Image Retrieval (CBIR)-a long-established research area in which improved efficiency and accuracy are needed for real-time retrieval. Artificial intelligence has made progress in CBIR and has significantly facilitated the process of instance search. In this survey we review recent instance retrieval works that are developed based on deep learning algorithms and techniques, with the survey organized by deep network architecture types, deep features, feature embedding and aggregation methods, and network fine-tuning strategies. Our survey considers a wide variety of recent methods, whereby we identify milestone work, reveal connections among various methods and present the commonly used benchmarks, evaluation results, common challenges, and propose promising future directions.

cs.CV↗

arXiv2022

Thermal transport properties of IrSbSe

We report a thermal transport study of IrSbSe, which crystallizes in a noncentrosymmetric cubic structure with the $P2_13$ space group and shows a narrow-gap semiconducting behavior. The large discrepancy between the activation energy for conductivity [$E_ρ$ = 128(2) meV] and for thermopower [$E_S$ = 17.7(9) meV] from 200 to 300 K indicates the polaronic transport mechanism. Electrical resistivity varies as $exp(T_0/T)^{1/4}$ and thermopower varies as $T^{1/2}$ at low temperatures, indicating that it evolves into the Mott's variable-range hopping dominant conduction with decreasing temperature. IrSbSe shows relatively low value of thermal conductivity ($\sim$ 1.65 W/K$\cdot$m) and thermopower of about 0.24 mV/K around 100 K, yet poor electrical conductivity. On the other hand, high vacancy defect concentration on both Ir and Sb atomic sites of up to 15\%, suggests high defect tolerance and points to possibility of future improvement of carrier density by chemical substitution or defect optimization.

cond-mat.str-el↗

arXiv2022

Anisotropic magnetotransport properties of the heavy-fermion superconductor CeRh$_2$As$_2$

We report anisotropic resistivity measurements of the heavy-fermion superconductor CeRh$_2$As$_2$ in magnetic fields up to 16 T and temperatures down to 0.35 K. The measured CeRh$_2$As$_2$ resistivity shows a signature corresponding to the suggested quadrupole density wave order state at $T_0 \sim$ 0.5 K for both measured directions. For a magnetic field applied along the tetragonal $a$ axis, $T_0$ is enhanced with magnetic field reaching $\sim$1.75 K at 16 T. Further, a magnetic field-induced transition occurs at $B_m \sim $ 8.1 T corresponding to a change to a new broken symmetry state. For a magnetic field applied along the $c$ axis, $T_0$ is suppressed below our base temperature $\sim$0.35 K by $B \sim$ 4.5 T, a field close to the previously reported field-induced transition within the superconducting state suggested to be from an even-parity to an odd-parity state. Our results indicate that the multiple superconducting phases in CeRh$_2$As$_2$ are intimately tied to the suppression of the proposed quadrupole-density-wave phase at $T_0$.

cond-mat.supr-con↗

arXiv2022

Large-batch Optimization for Dense Visual Predictions

Training a large-scale deep neural network in a large-scale dataset is challenging and time-consuming. The recent breakthrough of large-batch optimization is a promising way to tackle this challenge. However, although the current advanced algorithms such as LARS and LAMB succeed in classification models, the complicated pipelines of dense visual predictions such as object detection and segmentation still suffer from the heavy performance drop in the large-batch training regime. To address this challenge, we propose a simple yet effective algorithm, named Adaptive Gradient Variance Modulator (AGVM), which can train dense visual predictors with very large batch size, enabling several benefits more appealing than prior arts. Firstly, AGVM can align the gradient variances between different modules in the dense visual predictors, such as backbone, feature pyramid network (FPN), detection, and segmentation heads. We show that training with a large batch size can fail with the gradient variances misaligned among them, which is a phenomenon primarily overlooked in previous work. Secondly, AGVM is a plug-and-play module that generalizes well to many different architectures (e.g., CNNs and Transformers) and different tasks (e.g., object detection, instance segmentation, semantic segmentation, and panoptic segmentation). It is also compatible with different optimizers (e.g., SGD and AdamW). Thirdly, a theoretical analysis of AGVM is provided. Extensive experiments on the COCO and ADE20K datasets demonstrate the superiority of AGVM. For example, it can train Faster R-CNN+ResNet50 in 4 minutes without losing performance. AGVM enables training an object detector with one billion parameters in just 3.5 hours, reducing the training time by 20.9x, whilst achieving 62.2 mAP on COCO. The deliverables are released at https://github.com/Sense-X/AGVM.

cs.CV↗

arXiv2022

Single-Stage Open-world Instance Segmentation with Cross-task Consistency Regularization

Open-World Instance Segmentation (OWIS) is an emerging research topic that aims to segment class-agnostic object instances from images. The mainstream approaches use a two-stage segmentation framework, which first locates the candidate object bounding boxes and then performs instance segmentation. In this work, we instead promote a single-stage framework for OWIS. We argue that the end-to-end training process in the single-stage framework can be more convenient for directly regularizing the localization of class-agnostic object pixels. Based on the single-stage instance segmentation framework, we propose a regularization model to predict foreground pixels and use its relation to instance segmentation to construct a cross-task consistency loss. We show that such a consistency loss could alleviate the problem of incomplete instance annotation -- a common problem in the existing OWIS datasets. We also show that the proposed loss lends itself to an effective solution to semi-supervised OWIS that could be considered an extreme case that all object annotations are absent for some images. Our extensive experiments demonstrate that the proposed method achieves impressive results in both fully-supervised and semi-supervised settings. Compared to SOTA methods, the proposed method significantly improves the $AP_{100}$ score by 4.75\% in UVO$\rightarrow$UVO setting and 4.05\% in COCO$\rightarrow$UVO setting. In the case of semi-supervised learning, our model learned with only 30\% labeled data, even outperforms its fully-supervised counterpart with 50\% labeled data. The code will be released soon.

cs.CV↗