Searcharxiv⌕ Search

arXiv subjects

Lei Chen

Publications and source records attributed to Lei Chen.

At least 559 records · Page 31Linked to original sources
arXiv2020

AutoSF: Searching Scoring Functions for Knowledge Graph Embedding

Scoring functions (SFs), which measure the plausibility of triplets in knowledge graph (KG), have become the crux of KG embedding. Lots of SFs, which target at capturing different kinds of relations in KGs, have been designed by humans in recent years. However, as relations can exhibit complex patterns that are hard to infer before training, none of them can consistently perform better than others on existing benchmark data sets. In this paper, inspired by the recent success of automated machine learning (AutoML), we propose to automatically design SFs (AutoSF) for distinct KGs by the AutoML techniques. However, it is non-trivial to explore domain-specific information here to make AutoSF efficient and effective. We firstly identify a unified representation over popularly used SFs, which helps to set up a search space for AutoSF. Then, we propose a greedy algorithm to search in such a space efficiently. The algorithm is further sped up by a filter and a predictor, which can avoid repeatedly training SFs with same expressive ability and help removing bad candidates during the search before model training. Finally, we perform extensive experiments on benchmark data sets. Results on link prediction and triplets classification show that the searched SFs by AutoSF, are KG dependent, new to the literature, and outperform the state-of-the-art SFs designed by humans.

cs.LG↗
arXiv2020

2-Entity RANSAC for robust visual localization in changing environment

Visual localization has attracted considerable attention due to its low-cost and stable sensor, which is desired in many applications, such as autonomous driving, inspection robots and unmanned aerial vehicles. However, current visual localization methods still struggle with environmental changes across weathers and seasons, as there is significant appearance variation between the map and the query image. The crucial challenge in this situation is that the percentage of outliers, i.e. incorrect feature matches, is high. In this paper, we derive minimal closed form solutions for 3D-2D localization with the aid of inertial measurements, using only 2 pairs of point matches or 1 pair of point match and 1 pair of line match. These solutions are further utilized in the proposed 2-entity RANSAC, which is more robust to outliers as both line and point features can be used simultaneously and the number of matches required for pose calculation is reduced. Furthermore, we introduce three feature sampling strategies with different advantages, enabling an automatic selection mechanism. With the mechanism, our 2-entity RANSAC can be adaptive to the environments with different distribution of feature types in different segments. Finally, we evaluate the method on both synthetic and real-world datasets, validating its performance and effectiveness in inter-session scenarios.

cs.RO↗
arXiv2020

Block Hankel Tensor ARIMA for Multiple Short Time Series Forecasting

This work proposes a novel approach for multiple time series forecasting. At first, multi-way delay embedding transform (MDT) is employed to represent time series as low-rank block Hankel tensors (BHT). Then, the higher-order tensors are projected to compressed core tensors by applying Tucker decomposition. At the same time, the generalized tensor Autoregressive Integrated Moving Average (ARIMA) is explicitly used on consecutive core tensors to predict future samples. In this manner, the proposed approach tactically incorporates the unique advantages of MDT tensorization (to exploit mutual correlations) and tensor ARIMA coupled with low-rank Tucker decomposition into a unified framework. This framework exploits the low-rank structure of block Hankel tensors in the embedded space and captures the intrinsic correlations among multiple TS, which thus can improve the forecasting results, especially for multiple short time series. Experiments conducted on three public datasets and two industrial datasets verify that the proposed BHT-ARIMA effectively improves forecasting accuracy and reduces computational cost compared with the state-of-the-art methods.

cs.LG↗
arXiv2020

Non-realizability of the pure braid group as area-preserving homeomorphisms

Let $\text{Homeo}_+(D^2_n)$ be the group of orientation-preserving homeomorphisms of $D^2$ fixing the boundary pointwise and $n$ marked points as a set. Nielsen realization problem for the braid group asks whether the natural projection $p_n:\text{Homeo}_+(D^2_n)\to B_n:=π_0(\text{Homeo}_+(D^2_n))$ has a section over subgroups of $B_n$. All of the previous methods either use torsions or Thurston stability, which do not apply to the pure braid group $PB_n$, the subgroup of $B_n$ that fixes $n$ marked points pointwise. In this paper, we show that the pure braid group has no realization inside the area-preserving homeomorphisms using rotation numbers.

math.GT↗
arXiv2020

Topology Identification in Distribution Networks using Harmonic Synchrophasor Measurements

Topology identification (TI) in distribution networks is a challenging task due to the limited measurement resources and therefore the inevitable need to use pseudo-measurements that are often inaccurate. To address this issue, a new method is proposed in this paper to integrate harmonic synchrophasors into the TI problem in order to enhance TI accuracy in distribution networks. In this method, topology identification is done jointly based on both fundamental synchrophasor measurements and harmonic synchrophasor measurements. This is done by formulating and then solving a mixed-integer linear programming (MILP) problem. Furthermore, an analysis is provided to capture the number of and the location of harmonic sources and sensors that are needed to ensure full observability. The benefits of the proposed TI scheme are compared against those of the traditional scheme that utilizes only the fundamental measurements. Finally, through numerical simulations on the IEEE 33-Bus power system, it is shown that the proposed scheme is considerably accurate compared to the traditional scheme in topology identification.

eess.SY↗
arXiv2020

Gravitational waves detection with exceptional points in micro cavities

Here we propose a new gravitational waves(GWs) detector in broad frequency band, which is operated at exceptional points(EPs) in micro cavities. The detected signal is an eigenfrequency split of the mechanical modes caused by the spatial strain. Due to the complex square root topology near the EP, the splitting is greatly enhanced for sufficiently small perturbations. Compared to current strategies, it can be achieved at the room temperature and has advantages in micro device scale, wide frequency band and higher sensitivity.

quant-ph↗
arXiv2020

Revisiting Graph based Collaborative Filtering: A Linear Residual Graph Convolutional Network Approach

Graph Convolutional Networks (GCNs) are state-of-the-art graph based representation learning models by iteratively stacking multiple layers of convolution aggregation operations and non-linear activation operations. Recently, in Collaborative Filtering (CF) based Recommender Systems (RS), by treating the user-item interaction behavior as a bipartite graph, some researchers model higher-layer collaborative signals with GCNs. These GCN based recommender models show superior performance compared to traditional works. However, these models suffer from training difficulty with non-linear activations for large user-item graphs. Besides, most GCN based models could not model deeper layers due to the over smoothing effect with the graph convolution operation. In this paper, we revisit GCN based CF models from two aspects. First, we empirically show that removing non-linearities would enhance recommendation performance, which is consistent with the theories in simple graph convolutional networks. Second, we propose a residual network structure that is specifically designed for CF with user-item interaction modeling, which alleviates the over smoothing problem in graph convolution aggregation operation with sparse user-item interaction data. The proposed model is a linear model and it is easy to train, scale to large datasets, and yield better efficiency and effectiveness on two real datasets. We publish the source code at https://github.com/newlei/LRGCCF.

cs.IR↗
arXiv2020

Adapting Grad-CAM for Embedding Networks

The gradient-weighted class activation mapping (Grad-CAM) method can faithfully highlight important regions in images for deep model prediction in image classification, image captioning and many other tasks. It uses the gradients in back-propagation as weights (grad-weights) to explain network decisions. However, applying Grad-CAM to embedding networks raises significant challenges because embedding networks are trained by millions of dynamically paired examples (e.g. triplets). To overcome these challenges, we propose an adaptation of the Grad-CAM method for embedding networks. First, we aggregate grad-weights from multiple training examples to improve the stability of Grad-CAM. Then, we develop an efficient weight-transfer method to explain decisions for any image without back-propagation. We extensively validate the method on the standard CUB200 dataset in which our method produces more accurate visual attention than the original Grad-CAM method. We also apply the method to a house price estimation application using images. The method produces convincing qualitative results, showcasing the practicality of our approach.

cs.CV↗
arXiv2020

Personalized Multimedia Item and Key Frame Recommendation

When recommending or advertising items to users, an emerging trend is to present each multimedia item with a key frame image (e.g., the poster of a movie). As each multimedia item can be represented as multiple fine-grained visual images (e.g., related images of the movie), personalized key frame recommendation is necessary in these applications to attract users' unique visual preferences. However, previous personalized key frame recommendation models relied on users' fine-grained image behavior of multimedia items (e.g., user-image interaction behavior), which is often not available in real scenarios. In this paper, we study the general problem of joint multimedia item and key frame recommendation in the absence of the fine-grained user-image behavior. We argue that the key challenge of this problem lies in discovering users' visual profiles for key frame recommendation, as most recommendation models would fail without any users' fine-grained image behavior. To tackle this challenge, we leverage users' item behavior by projecting users (items) in two latent spaces: a collaborative latent space and a visual latent space. We further design a model to discern both the collaborative and visual dimensions of users, and model how users make decisive item preferences from these two spaces. As a result, the learned user visual profiles could be directly applied for key frame recommendation. Finally, experimental results on a real-world dataset clearly show the effectiveness of our proposed model on the two recommendation tasks.

cs.IR↗
arXiv2019

Variable warm dust around the Herbig Ae star HD~169142: Birth of a ring?

The Herbig Ae star HD 169142 is known to have a gaseous disk with a large inner hole, and also a photometrically variable inner dust component in the sub-au region. Following up our previous analysis, we further studied the temporal evolution of inner dust around HD 169142, which may provide information on the evolution from late-stage protoplanetary disks to debris disks. We used near-infrared interferometric observations obtained with VLTI/PIONIER to constrain the dust distribution at three epochs spanning six years. We also studied the photometric variability of HD 169142 using our optical-infrared observations and archival data. Our results indicate that a dust ring at ~0.3 au formed at some time between 2013 and 2018, and then faded (but did not completely disappear) by 2019. The short-term variability resembles that observed in extreme debris disks, and is likely related to short-lived dust of secondary origin, though variable shadowing from the inner ring could be an alternative interpretation. If confirmed, this is the first direct detection of secondary dust production inside a protoplanetary disk.

astro-ph.EP↗
arXiv2019

Enhanced sensing of non-Newtonian effects at ultrashort range with exceptional points in optomechanical systems

We propose an optomechanical nano-gravimeter based on exceptional points. The system is a coupled cavity optomechanical system, in which the gain and loss are applied by driving the cavities with a blue detuned and red detuned electromagnetic field, respectively. When the gain and loss reach a balance, the system will show the degeneracy of exceptional points, and any perturbation will cause an eigenfrequencies split, which is proportional to the square root of the perturbation strength. Compared with the traditional optomechanical sensors, the sensitivity is greatly enhanced. This work paves the way for the design of optomechanical ultrasensitive force sensors that can be applied to detect non-Newtonian effects, high-order weak interactions, and so on.

quant-ph↗
arXiv2019

Ultrasensitive optomechanical detection of an axion-mediated force based on a sharp peak emerging in probe absorption spectrum

Axion remains the most convincing solution to the strong-CP problem and a well-motivated dark matter candidate, causing the search for axions and axion-like particles(ALPs) to attract attention continually. The exchange of such particles may cause anomalous spin-dependent forces, inspiring many laboratory ALP searching experiments based on the detection of macroscopic monopole-dipole interactions between polarized electrons/nucleons and unpolarized nucleons. Since there is no exact proof of the existence of these interactions, to detect them is still of great significance. In the present paper, we study the electron-neucleon monopole-dipole interaction with a new method, in which a hybrid spin-nanocantilever optomechanical system consisting of a nitrogen-vacancy(NV) center and a nanocantilever resonator is used. With a static magnetic field and a pump microwave beam and a probe microwave beam applied, a probe absorption spectrum could be obtained. Through specific peaks appearing in the spectrum, we can identify this monopole-dipole interaction. And we also provide a prospective constraint to constrain the interaction. Furthermore, because our method can also be applied to the detection of some other spin-dependent interactions, this work provides new ideas for the experimental searches of the anomalous spin-dependent interactions.

astro-ph.CO↗
arXiv2019

Design and commissioning of a 600 L Time Projection Chamber with Microbulk Micromegas

We report the design, construction, and initial commissioning results of a large high pressure gaseous Time Projection Chamber (TPC) with Micromegas modules for charge readout. The detector vessel has an inner volume of about 600 L and an active volume of 270 L. At 10 bar operating pressure, the active volume contains about 20 kg of xenon gas and can image charged particle tracks. Drift electrons are collected by the charge readout plane, which accommodates a tessellation of seven Micromegas modules. Each of the Micromegas covers a square of 20 cm by 20 cm. A new type of Microbulk Micromegas is chosen for this application due to its good gain uniformity and low radioactive contamination. Initial commissioning results with 1 Micromegas module running with 1 bar argon and isobutane gas mixture and 5 bar xenon and trimethylamine (TMA) gas mixture are reported. We also recorded extended background tracks from cosmic ray events and highlighted the unique tracking feature of this gaseous TPC.

physics.ins-det↗
arXiv2019

Precipitation Nowcasting with Star-Bridge Networks

Precipitation nowcasting, which aims to precisely predict the short-term rainfall intensity of a local region, is gaining increasing attention in the artificial intelligence community. Existing deep learning-based algorithms use a single network to process various rainfall intensities together, compromising the predictive accuracy. Therefore, this paper proposes a novel recurrent neural network (RNN) based star-bridge network (StarBriNet) for precipitation nowcasting. The novelty of this work lies in the following three aspects. First, the proposed network comprises multiple sub-networks to deal with different rainfall intensities and duration separately, which can significantly improve the model performance. Second, we propose a star-shaped information bridge to enhance the information flow across RNN layers. Third, we introduce a multi-sigmoid loss function to take the precipitation nowcasting criterion into account. Experimental results demonstrate superior performance for precipitation nowcasting over existing algorithms, including the state-of-the-art one, on a natural radar echo dataset.

cs.CV↗
arXiv2019

SpiralNet++: A Fast and Highly Efficient Mesh Convolution Operator

Intrinsic graph convolution operators with differentiable kernel functions play a crucial role in analyzing 3D shape meshes. In this paper, we present a fast and efficient intrinsic mesh convolution operator that does not rely on the intricate design of kernel function. We explicitly formulate the order of aggregating neighboring vertices, instead of learning weights between nodes, and then a fully connected layer follows to fuse local geometric structure information with vertex features. We provide extensive evidence showing that models based on this convolution operator are easier to train, and can efficiently learn invariant shape features. Specifically, we evaluate our method on three different types of tasks of dense shape correspondence, 3D facial expression classification, and 3D shape reconstruction, and show that it significantly outperforms state-of-the-art approaches while being significantly faster, without relying on shape descriptors. Our source code is available on GitHub.

cs.CV↗
arXiv2019

Spectral evolution and radial dust transport in the prototype young eruptive system EX Lup

EX Lup is the prototype of a class of pre-main sequence eruptive stars defined by their repetitive outbursts lasting several months. In 2008 January-September EX Lup underwent its historically largest outburst, brightening by about 4 magnitudes in visual light. In previous studies we discovered on-going silicate crystal formation in the inner disk during the outburst, but also noticed that the measured crystallinity fraction started decreasing after the source returned to the quiescent phase. Here we present new observations of the 10 $μ$m silicate feature, obtained with the MIDI and VISIR instruments at Paranal Observatory. The observations demonstrate that within five years practically all crystalline forsterite disappeared from the surface of the inner disk. We reconstruct this process by presenting a series of parametric axisymmetric radiative transfer models of an expanding dust cloud that transports the crystals from the terrestrial zone to outer disk regions where comets are supposed to form. Possibly the early Sun also experienced similar flare-ups, and the forming planetesimals might have incorporated crystalline silicate material produced by such outbursts. Finally, we discuss how far the location of the dust cloud could be constrained by future JWST observations.

astro-ph.SR↗
arXiv2019

A novel method of searching for an axial-vector mediated force

We present an atomic, molecular, and optical physics based method for the purpose of search for axial-vector mediated dipole-dipole interaction between electrons. In our scheme, applying a static magnetic field and a pump beam and a probe beam to a hybrid mechanical system composed of a nitrogen-vacancy center and a cantilever resonator, we could obtain a probe absorption spectrum. Based on the study of the relationship between this spectrum and the exotic dipole-dipole interaction, we put forward our detection principle and then provide a prospective constraint most stringent at a rough interaction range from $4\times{10}^{-8}$ to $2\times {10}^{-7}m$ . Our results indicate that this scheme could be put into consideration in relevant experimental searches .

quant-ph↗