arXiv subjects

Wentao Zhang

Publications and source records attributed to Wentao Zhang.

At least 433 records · Page 24Linked to original sources

arXiv2022

Hyper-Tune: Towards Efficient Hyper-parameter Tuning at Scale

The ever-growing demand and complexity of machine learning are putting pressure on hyper-parameter tuning systems: while the evaluation cost of models continues to increase, the scalability of state-of-the-arts starts to become a crucial bottleneck. In this paper, inspired by our experience when deploying hyper-parameter tuning in a real-world application in production and the limitations of existing systems, we propose Hyper-Tune, an efficient and robust distributed hyper-parameter tuning framework. Compared with existing systems, Hyper-Tune highlights multiple system optimizations, including (1) automatic resource allocation, (2) asynchronous scheduling, and (3) multi-fidelity optimizer. We conduct extensive evaluations on benchmark datasets and a large-scale real-world dataset in production. Empirically, with the aid of these optimizations, Hyper-Tune outperforms competitive hyper-parameter tuning systems on a wide range of scenarios, including XGBoost, CNN, RNN, and some architectural hyper-parameters for neural networks. Compared with the state-of-the-art BOHB and A-BOHB, Hyper-Tune achieves up to 11.2x and 5.1x speedups, respectively.

cs.LG↗

arXiv2021

Efficient Diversity-Driven Ensemble for Deep Neural Networks

The ensemble of deep neural networks has been shown, both theoretically and empirically, to improve generalization accuracy on the unseen test set. However, the high training cost hinders its efficiency since we need a sufficient number of base models and each one in the ensemble has to be separately trained. Lots of methods are proposed to tackle this problem, and most of them are based on the feature that a pre-trained network can transfer its knowledge to the next base model and then accelerate the training process. However, these methods suffer a severe problem that all of them transfer knowledge without selection and thus lead to low diversity. As the effect of ensemble learning is more pronounced if ensemble members are accurate and diverse, we propose a method named Efficient Diversity-Driven Ensemble (EDDE) to address both the diversity and the efficiency of an ensemble. To accelerate the training process, we propose a novel knowledge transfer method which can selectively transfer the previous generic knowledge. To enhance diversity, we first propose a new diversity measure, then use it to define a diversity-driven loss function for optimization. At last, we adopt a Boosting-based framework to combine the above operations, such a method can also further improve diversity. We evaluate EDDE on Computer Vision (CV) and Natural Language Processing (NLP) tasks. Compared with other well-known ensemble methods, EDDE can get highest ensemble accuracy with the lowest training cost, which means it is efficient in the ensemble of neural networks.

cs.LG↗

arXiv2021

K-Core Decomposition on Super Large Graphs with Limited Resources

K-core decomposition is a commonly used metric to analyze graph structure or study the relative importance of nodes in complex graphs. Recent years have seen rapid growth in the scale of the graph, especially in industrial settings. For example, our industrial partner runs popular social applications with billions of users and is able to gather a rich set of user data. As a result, applying K-core decomposition on large graphs has attracted more and more attention from academics and the industry. A simple but effective method to deal with large graphs is to train them in the distributed settings, and some distributed K-core decomposition algorithms are also proposed. Despite their effectiveness, we experimentally and theoretically observe that these algorithms consume too many resources and become unstable on super-large-scale graphs, especially when the given resources are limited. In this paper, we deal with those super-large-scale graphs and propose a divide-and-conquer strategy on top of the distributed K-core decomposition algorithm. We evaluate our approach on three large graphs. The experimental results show that the consumption of resources can be significantly reduced, and the calculation on large-scale graphs becomes more stable than the existing methods. For example, the distributed K-core decomposition algorithm can scale to a large graph with 136 billion edges without losing correctness with our divide-and-conquer technique.

cs.DC↗

arXiv2021

Coupling-assisted quasi-bound states in the continuum in heterogeneous metasurfaces

In this paper, we present a Bound states in the continuum (BIC) metamaterial in heterogeneous structures based on the universal coupled mode theory. We find the more general physical parameters to represent BIC, which are the resonant frequencies and corresponding phases of metamaterial structures. Therefore, BIC metamaterial comes from the equal value of the resonant frequencies and phases of metamaterial structures which are not only for homogeneous structures. Meanwhile if slightly vary one of metamaterial structure's resonant frequency and phase by varying geometry, we can obtain the quasi-BIC instead of broken symmetry of homogeneous structures. In this paper, we provide the BIC and quasi-BIC with one example of two heterogeneous structures which are cut wire (CW) and Split-Ring Resonator (SRR), to widely extends the metamaterial BIC beyond common sense. Furthermore, we demonstrate the simulation results and experimental results to proof our idea.

physics.optics↗

arXiv2021

OpenBox: A Generalized Black-box Optimization Service

Black-box optimization (BBO) has a broad range of applications, including automatic machine learning, engineering, physics, and experimental design. However, it remains a challenge for users to apply BBO methods to their problems at hand with existing software packages, in terms of applicability, performance, and efficiency. In this paper, we build OpenBox, an open-source and general-purpose BBO service with improved usability. The modular design behind OpenBox also facilitates flexible abstraction and optimization of basic BBO components that are common in other existing systems. OpenBox is distributed, fault-tolerant, and scalable. To improve efficiency, OpenBox further utilizes "algorithm agnostic" parallelization and transfer learning. Our experimental results demonstrate the effectiveness and efficiency of OpenBox compared to existing systems.

cs.LG↗

arXiv2021

RIM: Reliable Influence-based Active Learning on Graphs

Message passing is the core of most graph models such as Graph Convolutional Network (GCN) and Label Propagation (LP), which usually require a large number of clean labeled data to smooth out the neighborhood over the graph. However, the labeling process can be tedious, costly, and error-prone in practice. In this paper, we propose to unify active learning (AL) and message passing towards minimizing labeling costs, e.g., making use of few and unreliable labels that can be obtained cheaply. We make two contributions towards that end. First, we open up a perspective by drawing a connection between AL enforcing message passing and social influence maximization, ensuring that the selected samples effectively improve the model performance. Second, we propose an extension to the influence model that incorporates an explicit quality factor to model label noise. In this way, we derive a fundamentally new AL selection criterion for GCN and LP--reliable influence maximization (RIM)--by considering quantity and quality of influence simultaneously. Empirical studies on public datasets show that RIM significantly outperforms current AL methods in terms of accuracy and efficiency.

cs.LG↗

arXiv2021

Node Dependent Local Smoothing for Scalable Graph Learning

Recent works reveal that feature or label smoothing lies at the core of Graph Neural Networks (GNNs). Concretely, they show feature smoothing combined with simple linear regression achieves comparable performance with the carefully designed GNNs, and a simple MLP model with label smoothing of its prediction can outperform the vanilla GCN. Though an interesting finding, smoothing has not been well understood, especially regarding how to control the extent of smoothness. Intuitively, too small or too large smoothing iterations may cause under-smoothing or over-smoothing and can lead to sub-optimal performance. Moreover, the extent of smoothness is node-specific, depending on its degree and local structure. To this end, we propose a novel algorithm called node-dependent local smoothing (NDLS), which aims to control the smoothness of every node by setting a node-specific smoothing iteration. Specifically, NDLS computes influence scores based on the adjacency matrix and selects the iteration number by setting a threshold on the scores. Once selected, the iteration number can be applied to both feature smoothing and label smoothing. Experimental results demonstrate that NDLS enjoys high accuracy -- state-of-the-art performance on node classifications tasks, flexibility -- can be incorporated with any models, scalability and efficiency -- can support large scale graphs with fast training.

cs.LG↗

arXiv2021

The Betti Number of the Independence Complex of Ternary Graphs

Given a graph $G$, the \textit{independence complex} $I(G)$ is the simplicial complex whose faces are the independent sets of $V(G)$. Let $\tilde{b}_i$ denote the $i$-th reduced Betti number of $I(G)$, and let $b(G)$ denote the sum of $\tilde{b}_i(G)$'s. A graph is ternary if it does not contain induced cycles with length divisible by three. G. Kalai and K. Meshulam conjectured that $b(G)\le 1$ whenever $G$ is ternary. We prove this conjecture. This extends a recent results proved by Chudnovsky, Scott, Seymour and Spirkl that for any ternary graph $G$, the number of independent sets with even cardinality and the independent sets with odd cardinality differ by at most 1.

math.CO↗

arXiv2021

Coordination on Time-Varying Antagonistic Networks

This paper studies coordination problem for time-varying networks suffering from antagonistic information, quantified by scaling parameters. By such a manner, interacting property of the participating individuals and antagonistic information can be quantified in a fully decoupled perspective, thus benefiting from merely directed spanning tree hypothesis is needed, in the sense of usual algebraic graph theory. We start with rigorous argument on the existence of weighted gain, and then derive relation among weighted gain, scaling parameter and Laplacian matrix guaranteeing antagonistic information cannot diverge system state. Based on these arguments, we devise coordination algorithm constrained by topology-dependent average time condition, thus relaxing the examination of directed spanning tree requirement for the union graph that is usually intractable. Moreover, the induced theoretical results are applied to time-varying networks with several mutually uninfluenced agents, in accompanying with some discussions and comparisons with respect to the existing developments.

eess.SY↗

arXiv2021

Evaluating Deep Graph Neural Networks

Graph Neural Networks (GNNs) have already been widely applied in various graph mining tasks. However, they suffer from the shallow architecture issue, which is the key impediment that hinders the model performance improvement. Although several relevant approaches have been proposed, none of the existing studies provides an in-depth understanding of the root causes of performance degradation in deep GNNs. In this paper, we conduct the first systematic experimental evaluation to present the fundamental limitations of shallow architectures. Based on the experimental results, we answer the following two essential questions: (1) what actually leads to the compromised performance of deep GNNs; (2) when we need and how to build deep GNNs. The answers to the above questions provide empirical insights and guidelines for researchers to design deep and well-performed GNNs. To show the effectiveness of our proposed guidelines, we present Deep Graph Multi-Layer Perceptron (DGMLP), a powerful approach (a paradigm in its own right) that helps guide deep GNN designs. Experimental results demonstrate three advantages of DGMLP: 1) high accuracy -- it achieves state-of-the-art node classification performance on various datasets; 2) high flexibility -- it can flexibly choose different propagation and transformation depths according to graph size and sparsity; 3) high scalability and efficiency -- it supports fast training on large-scale graphs. Our code is available in https://github.com/zwt233/DGMLP.

cs.LG↗

arXiv2021

Grain: Improving Data Efficiency of Graph Neural Networks via Diversified Influence Maximization

Data selection methods, such as active learning and core-set selection, are useful tools for improving the data efficiency of deep learning models on large-scale datasets. However, recent deep learning models have moved forward from independent and identically distributed data to graph-structured data, such as social networks, e-commerce user-item graphs, and knowledge graphs. This evolution has led to the emergence of Graph Neural Networks (GNNs) that go beyond the models existing data selection methods are designed for. Therefore, we present Grain, an efficient framework that opens up a new perspective through connecting data selection in GNNs with social influence maximization. By exploiting the common patterns of GNNs, Grain introduces a novel feature propagation concept, a diversified influence maximization objective with novel influence and diversity functions, and a greedy algorithm with an approximation guarantee into a unified framework. Empirical studies on public datasets demonstrate that Grain significantly improves both the performance and efficiency of data selection (including active learning and core-set selection) for GNNs. To the best of our knowledge, this is the first attempt to bridge two largely parallel threads of research, data selection, and social influence maximization, in the setting of GNNs, paving new ways for improving data efficiency.

cs.LG↗

arXiv2021

ROD: Reception-aware Online Distillation for Sparse Graphs

Graph neural networks (GNNs) have been widely used in many graph-based tasks such as node classification, link prediction, and node clustering. However, GNNs gain their performance benefits mainly from performing the feature propagation and smoothing across the edges of the graph, thus requiring sufficient connectivity and label information for effective propagation. Unfortunately, many real-world networks are sparse in terms of both edges and labels, leading to sub-optimal performance of GNNs. Recent interest in this sparse problem has focused on the self-training approach, which expands supervised signals with pseudo labels. Nevertheless, the self-training approach inherently cannot realize the full potential of refining the learning performance on sparse graphs due to the unsatisfactory quality and quantity of pseudo labels. In this paper, we propose ROD, a novel reception-aware online knowledge distillation approach for sparse graph learning. We design three supervision signals for ROD: multi-scale reception-aware graph knowledge, task-based supervision, and rich distilled knowledge, allowing online knowledge transfer in a peer-teaching manner. To extract knowledge concealed in the multi-scale reception fields, ROD explicitly requires individual student models to preserve different levels of locality information. For a given task, each student would predict based on its reception-scale knowledge, while simultaneously a strong teacher is established on-the-fly by combining multi-scale knowledge. Our approach has been extensively evaluated on 9 datasets and a variety of graph-based tasks, including node classification, link prediction, and node clustering. The result demonstrates that ROD achieves state-of-art performance and is more robust for the graph sparsity.

cs.LG↗

arXiv2021

VolcanoML: Speeding up End-to-End AutoML via Scalable Search Space Decomposition

End-to-end AutoML has attracted intensive interests from both academia and industry, which automatically searches for ML pipelines in a space induced by feature engineering, algorithm/model selection, and hyper-parameter tuning. Existing AutoML systems, however, suffer from scalability issues when applying to application domains with large, high-dimensional search spaces. We present VolcanoML, a scalable and extensible framework that facilitates systematic exploration of large AutoML search spaces. VolcanoML introduces and implements basic building blocks that decompose a large search space into smaller ones, and allows users to utilize these building blocks to compose an execution plan for the AutoML problem at hand. VolcanoML further supports a Volcano-style execution model - akin to the one supported by modern database systems - to execute the plan constructed. Our evaluation demonstrates that, not only does VolcanoML raise the level of expressiveness for search space decomposition in AutoML, it also leads to actual findings of decomposition strategies that are significantly more efficient than the ones employed by state-of-the-art AutoML systems such as auto-sklearn.

cs.LG↗

arXiv2021

Rigorous Signal Reconstruction in Terahertz Emission Spectroscopy

Terahertz (THz) emission spectroscopy is a powerful method that allows one to measure the ultrafast dynamics of polarization, current, or magnetization in a material based on THz emission from the material. However, the practical implementation of this method can be challenging, and can result in significant errors in the reconstruction of the quantity of interest. Here, we experimentally and theoretically demonstrate a rigorous method of signal reconstruction in THz emission spectroscopy, and describe the main experimental and theoretical sources of reconstruction error. We identify the linear line-of-sight geometry of the THz emission spectrometer as the optimal configuration for accurate, fully calibrated THz signal reconstruction. As an example, we apply our reconstruction method to ultrafast THz magnetometry experiment, where we recover the ultrafast magnetization dynamics in a photoexcited iron film, including both its temporal shape and absolute magnitude.

physics.optics↗

arXiv2021

Optical manipulation of electronic dimensionality in a quantum material

Exotic phenomenon can be achieved in quantum materials by confining electronic states into two dimensions. For example, relativistic fermions are realised in a single layer of carbon atoms, the quantized Hall effect can result from two-dimensional (2D) systems, and the superconducting transition temperature can be enhanced significantly in a one-atomic-layer material. Ordinarily, 2D electronic system can be obtained by exfoliating the layered materials, growing monolayer materials on substrates, or establishing interfaces between different materials. Herein, we use femtosecond infrared laser pulses to invert the periodic lattice distortion sectionally in a three-dimensional (3D) charge density wave material, creating macroscopic domain walls of transient 2D ordered electronic states with exotic properties. The corresponding ultrafast electronic and lattice dynamics are captured by time- and angle-resolved photoemission spectroscopy and MeV ultrafast electron diffraction. Surprisingly, a novel energy gap state, which might be a signature of light-induced superconductivity, is identified in the photoinduced 2D domain wall near the surface. Such optical modulation of atomic motion is a new path to realise 2D electronic states and will be a new platform for creating novel phases in quantum materials.

cond-mat.str-el↗

arXiv2021

Population transfer under local dephasing

Stimulated Raman adiabatic passage is a well-known technique for quantum population transfer due to its robustness again various sources of noises. Here we consider quantum population transfer from one spin to another via an intermediate spin which subjects to dephasing noise. We obtain an analytic expression for the transfer efficiency under a specific driving protocol, showing that dephasing could reduce the transfer efficiency, but the effect of dephasing could also be suppressed with a stronger laser coupling or a longer laser duration. We also consider another commonly used driving protocol, which shows that this analytic picture is still qualitatively correct.

quant-ph↗

arXiv2021

Coexistence of Ferroelectric-like Polarization and Dirac-like Surface State in TaNiTe5

By combining angle-resolved photoemission spectroscopy (ARPES), scanning tunneling microscopy (STM), piezoresponse force microscopy (PFM) and first-principles calculations, we have studied the low-energy band structure, atomic structure and charge polarization on the surface of a topological semimetal candidate TaNiTe5. Dirac-like surface states were observed on the (010) surface by ARPES, consistent with the first-principles calculations. On the other hand, PFM reveals a switchable ferroelectric-like polarization on the same surface. We propose that the noncentrosymmetric surface reconstruction observed by STM could be the origin of the observed ferroelectric-like state in this novel material. Our findings provide a new platform with the coexistence of ferroelectric-like surface charge distribution and novel surface states.

cond-mat.mtrl-sci↗

arXiv2021

Adiabatic following of terahertz surface plasmon-polaritons coupler via two waveguides structure

Most recently, two remarkable papers [New J. Phys. 21, 113004 (2019); IEEE J. Sel. Top. Quantum Electron 27, 1 (2020)] propose broadband complete transfer terahertz (THz) surface plasmon polaritons (SPPs) waveguide coupler by applying coherent quantum control -- Stimulated Raman adiabatic passage (STIRAP). However, previous researches request three SPPs waveguides coupler. In this paper, we propose a new design of a broadband complete transfer THz SPPs coupler with an innovative structure of two waveguides by employing two state adiabatic following. In order to realize this design, we introduce the detuning parameter into the coupling equation of SPPs waveguides for the first time. We believe that this finding will improve the THz communication domain.

physics.optics↗