Aerial Imagery for Roof Segmentation: A Large-Scale Dataset towards Automatic Mapping of Buildings
arXiv admin note: This version has been removed as the user did not have the right to agree to the license at the time of submission
cs.CV↗
arXiv subjects
Lei Wang
Publications and source records attributed to Lei Wang.
Maunakea Spectroscopic Explorer (MSE): a preliminary design of multi-object high resolution spectrograph
The Maunakea Spectroscopic Explorer (MSE) project will transform the CFHT 3.6m optical telescope to a 10m class dedicated multi-object spectroscopic facility, with an ability to measure thousands of objects with three spectral resolution modes respectively low resolution of R~3,000, moderate resolution of R~6,000 and high resolution of R~40,000. Two identical multi-object high resolution spectrographs are expected to simultaneously produce 1084 spectra with high resolution of 40,000 at Blue (401-416nm) and Green (472-489nm) channels, and 20,000 at Red (626-674nm) channel. At the Conceptual Design Phase (CoDP), different optical schemes were proposed to meet the challenging requirements, especially a unique design with a novel transmission image slicer array, and another conventional design with oversize Volume Phase Holographic (VPH) gratings. It became clear during the CoDP that both designs presented problems of complexity or feasibility of manufacture, especially high line density disperser (general name for all kinds of grating, grism, prism). At the present, a new design scheme is proposed for investigating the optimal way to reduce technical risk and get more reliable estimation of cost and timescale. It contains new dispersers, F/2 fast collimator and so on. Therein, the disperser takes advantage of a special grism and a prism to reduce line density on grating surface, keep wide opening angle of optical path, and get the similar spectrum layout in all three spectral channels. For the fast collimator, it carefully compares on-axis and off-axis designs in throughput, interface to fiber assembly and technical risks. The current progress is more competitive and credible than the previous design, but it also indicates more challenging work will be done to improve its accessibility in engineering.
astro-ph.IM↗
Unsupervised Generative Modeling Using Matrix Product States
Generative modeling, which learns joint probability distribution from data and generates samples according to it, is an important task in machine learning and artificial intelligence. Inspired by probabilistic interpretation of quantum physics, we propose a generative model using matrix product states, which is a tensor network originally proposed for describing (particularly one-dimensional) entangled quantum states. Our model enjoys efficient learning analogous to the density matrix renormalization group method, which allows dynamically adjusting dimensions of the tensors and offers an efficient direct sampling approach for generative tasks. We apply our method to generative modeling of several standard datasets including the Bars and Stripes, random binary patterns and the MNIST handwritten digits to illustrate the abilities, features and drawbacks of our model over popular generative models such as Hopfield model, Boltzmann machines and generative adversarial networks. Our work sheds light on many interesting directions of future exploration on the development of quantum-inspired algorithms for unsupervised machine learning, which are promisingly possible to be realized on quantum devices.
cond-mat.stat-mech↗
A Simple and Space Efficient Segment Tree Implementation
The segment tree is an extremely versatile data structure. In this paper, a new heap based implementation of segment trees is proposed. In such an implementation of segment tree, the structural information associated with the tree nodes can be removed completely. Some primary computational geometry problems such as stabbing counting queries, measure of union of intervals, and maximum clique size of Intervals are used to demonstrate the efficiency of the new heap based segment tree implementation. Each interval in a set $S=\{I_1 ,I_2 ,\cdots,I_n\}$ of $n$ intervals can be insert into or delete from the heap based segment tree in $O(\log n)$ time. All the primary computational geometry problems can be solved efficiently.
cs.DS↗
Interactive Medical Image Segmentation via Point-Based Interaction and Sequential Patch Learning
Due to low tissue contrast, irregular object appearance, and unpredictable location variation, segmenting the objects from different medical imaging modalities (e.g., CT, MR) is considered as an important yet challenging task. In this paper, we present a novel method for interactive medical image segmentation with the following merits. (1) Our design is fundamentally different from previous pure patch-based and image-based segmentation methods. We observe that during delineation, the physician repeatedly check the inside-outside intensity changing to determine the boundary, which indicates that comparison in an inside-outside manner is extremely important. Thus, we innovatively model our segmentation task as learning the representation of the bi-directional sequential patches, starting from (or ending in) the given central point of the object. This can be realized by our proposed ConvRNN network embedded with a gated memory propagation unit. (2) Unlike previous interactive methods (requiring bounding box or seed points), we only ask the physician to merely click on the rough central point of the object before segmentation, which could simultaneously enhance the performance and reduce the segmentation time. (3) We utilize our method in a multi-level framework for better performance. We systematically evaluate our method in three different segmentation tasks including CT kidney tumor, MR prostate, and PROMISE12 challenge, showing promising results compared with state-of-the-art methods. The code is available here: \href{https://github.com/sunalbert/Sequential-patch-based-segmentation}{Sequential-patch-based-segmentation}.
cs.CV↗
Data Dwarfs: A Lens Towards Fully Understanding Big Data and AI Workloads
The complexity and diversity of big data and AI workloads make understanding them difficult and challenging. This paper proposes a new approach to characterizing big data and AI workloads. We consider each big data and AI workload as a pipeline of one or more classes of unit of computations performed on different initial or intermediate data inputs. Each class of unit of computation captures the common requirements while being reasonably divorced from individual implementations, and hence we call it a data dwarf. For the first time, among a wide variety of big data and AI workloads, we identify eight data dwarfs that takes up most of run time, including Matrix, Sampling, Logic, Transform, Set, Graph, Sort and Statistic. We implement the eight data dwarfs on different software stacks as the micro benchmarks of an open-source big data and AI benchmark suite, and perform comprehensive characterization of those data dwarfs from perspective of data sizes, types, sources, and patterns as a lens towards fully understanding big data and AI workloads.
cs.DC↗
Optimal Load-Balancing for High-Density Wireless Networks with Flow-Level Dynamics
We consider the load-balancing design for forwarding incoming flows to access points (APs) in high-density wireless networks with both channel fading and flow-level dynamics, where each incoming flow has a certain amount of service demand and leaves the system once its service request is complete (referred as flow-level dynamic model). The efficient load-balancing design is strongly needed for supporting high-quality wireless connections in high-density areas. Despite the presence of a variety of earlier works on the design and analysis of the load-balancing schemes in wireless networks, there does not exist a work on the load-balancing design in the realistic flow-level dynamic model. In this work, we propose a Join-the-Least-Workload (JLW) Algorithm that always forwards the incoming flows to the AP with the smallest workload in the presence of flow-level dynamics. However, our considered flow-level dynamic model differs from traditional queueing model for wireless networks in the following two aspects: (1) the dynamics of the flows is short-term and flows will leave the network once they received the desired amount of service; (2) each individual flow faces an independent channel fading. These differences pose significant challenges on the system performance analysis. To tackle these challenges, we perform Lyapunov-drift-based analysis of the stochastic network taking into account sharp flow-level dynamics. Our analysis reveals that our proposed JLW Algorithm not only achieves maximum system throughput, but also minimizes the total system workload in heavy-traffic regimes. Moreover, we observe from both our theoretical and simulation results that the mean total workload performance under the proposed JLW Algorithm does not degrade as the number of APs increases, which is strongly desirable in high-density wireless networks.
cs.NI↗
Anchor-based Nearest Class Mean Loss for Convolutional Neural Networks
Discriminative features are critical for machine learning applications. Most existing deep learning approaches, however, rely on convolutional neural networks (CNNs) for learning features, whose discriminant power is not explicitly enforced. In this paper, we propose a novel approach to train deep CNNs by imposing the intra-class compactness and the inter-class separability, so as to enhance the learned features' discriminant power. To this end, we introduce anchors, which are predefined vectors regarded as the centers for each class and fixed during training. Discriminative features are obtained by constraining the deep CNNs to map training samples to the corresponding anchors as close as possible. We propose two principles to select the anchors, and measure the proximity of two points using the Euclidean and cosine distance metric functions, which results in two novel loss functions. These loss functions require no sample pairs or triplets and can be efficiently optimized by batch stochastic gradient descent. We test the proposed method on three benchmark image classification datasets and demonstrate its promising results.
cs.CV↗
Alignment between satellite and central galaxies in the SDSS DR7: dependence on large-scale environment
The alignment between satellites and central galaxies has been studied in detail both in observational and theoretical works. The widely accepted fact is that the satellites preferentially reside along the major axis of their central galaxy. However, the origin and large-scale environment dependence of this alignment are still unknown. In an attempt to figure out those, we use data constructed from SDSS DR7 to investigate the large-scale environmental dependence of this alignment with emphasis on examining the alignments' dependence on the colour of the central galaxy. We find a very strong large-scale environmental dependence of the satellite-central alignment in groups with blue centrals. Satellites of blue centrals in knots are preferentially located perpendicular to the major axis of the centrals, and the alignment angle decreases with environment namely when going from knots to voids. The alignment angle strongly depend on the ${}^{0.1}(g-r)$ colour of centrals. We suggest that the satellite-central alignment is the result of a competition between satellite accretion within large scale-structure and galaxy evolution inside host haloes. For groups containing red central galaxies, the satellite-central alignment is mainly determined by the evolution effect, while for blue central dominated groups, the effect of large-scale structure plays a more important role, especially in knots. Our results provide an explanation for how the satellite-central alignment forms within different large-scale environments. The perpendicular case in groups and knots with blue centrals may also provide insight into understanding similar polar arrangements such the formation of the Milky Way and Centaurus A's satellite system.
astro-ph.GA↗
Differentiable Learning of Quantum Circuit Born Machine
Quantum circuit Born machines are generative models which represent the probability distribution of classical dataset as quantum pure states. Computational complexity considerations of the quantum sampling problem suggest that the quantum circuits exhibit stronger expressibility compared to classical neural networks. One can efficiently draw samples from the quantum circuits via projective measurements on qubits. However, similar to the leading implicit generative models in deep learning, such as the generative adversarial networks, the quantum circuits cannot provide the likelihood of the generated samples, which poses a challenge to the training. We devise an efficient gradient-based learning algorithm for the quantum circuit Born machine by minimizing the kerneled maximum mean discrepancy loss. We simulated generative modeling of the Bars-and-Stripes dataset and Gaussian mixture distributions using deep quantum circuits. Our experiments show the importance of circuit depth and gradient-based optimization algorithm. The proposed learning algorithm is runnable on near-term quantum device and can exhibit quantum advantages for generative modeling.
quant-ph↗
Study of Isolated-photon and Jet Momentum Imbalance in $pp$ and $PbPb$ collisions
In this paper, we study the production of isolated-photon plus a jet in $pp$ and $PbPb$ collisions, which can be used as an important probe to the jet transport property in quark gluon plasma created in heavy ion collisions. Normally, there are two types of observables associated with the production of isolated-photon plus a jet, namely, the azimuthal angular correlation and the transverse momentum imbalance. To understand both observables in the full kinematical region, we need to employ the perturbative QCD calculation, which takes into account the hard splitting of partons, together with the Sudakov resummation formalism, which resums soft gluon splittings. Furthermore, by introducing energy-loss into the system, we calculate the enhancement of the momentum imbalance distribution for $AA$ as compared to $pp$ collisions and make predictions for future unfolded experimental data. In addition, in order to extract the jet transport coefficient more precisely in our numerical calculation, we also distinguish quark jets from gluon jets, since they interact with quark gluon plasma with different strengths. This work provides a reliable theoretical tool for the calculation of the gamma-jet correlation, which can lead us to a more precise extraction of the jet transport coefficient in relativistic heavy-ion collisions.
hep-ph↗
Softening Theory of Matter Tuning Atomic Border to Make Soft Materials
Regulation of material softness has both theoretical and practical significances due to irreplaceable applications of soft matter in rather diverse areas. This article is dedicated to draft a theoretical category on interpreting the mechanisms lying behind all soft matters, which can be termed as Softening Theory. A technical strategy with generalized purpose was proposed for softening desired matter, i.e. the melting point of matter can be significantly reduced through tuning its interior boundaries in atomic level. This theory accords well with the classical nuclear droplet model that treats the nucleus as a droplet which had successfully explained many phenomena. It also explained the experimental fact that, the material's melting point is drastically reduced as the particles become smaller enough, in which situations effects of the atomic borders become much weaker. Along this direction, many phenomena existing in nature can be well understood. For example, if keeping in mind the fact that an atom consisting of nucleus and electronics can be regarded as fluid, all the matter consisted of atoms should maintain fluidic state in their macroscopic scale, according to the consistency between macro and micro worlds. However, many substances just cannot remain their original atomic fluidic behavior due to breaking of the consistency between macro and micro states. Based on the current softening theory, it is now easy to understand that the breaking of such consistency is just caused due to generated forces from the atomic interactions. To resolve such intrinsic confinement, a group of potential technical approaches can be developed to tune the atomic borders of the matter and thus make desired soft materials. This work provides a theoretical foundation to partially address the nature of the material which will aid to make future soft matters in the coming time.
cond-mat.soft↗
Ionic Modulation of Interfacial Magnetism through Electrostatic Doping in Pt/YIG bilayer heterostructure
Voltage modulation of yttrium iron garnet (YIG) with compactness, high speed response, energy efficiency and both practical/theoretical siginificances can be widely applied to various YIG based spintronics such as spin Hall, spin pumping, spin Seeback effects. Here we initial an ionic modulation of interfacial magnetism process on YIG/Pt bilayer heterostructures, where the Pt capping would influence the ferromagnetic (FMR) field position significantly, and realize a significant magnetism enhancement in bilayer system. A large voltage induced FMR field shifts of 690 Oe has been achieved in YIG (13 nm)/Pt (3 nm) multilayer heterostructures under a small voltage bias of 4.5 V. The remarkable ME tunability comes from voltage induced extra FM ordering in Pt metal layer near the Pt/YIG interface. The first-principle theoretical simulation reveal that the electrostatic doping induced Pt5+ ions have strong magnetic ordering due to uncompensated d orbit electrons. The large voltage control of FMR change pave a foundation towards novel voltage tunable YIG based spintronics.
cond-mat.mtrl-sci↗
Understanding Membership Inferences on Well-Generalized Learning Models
Membership Inference Attack (MIA) determines the presence of a record in a machine learning model's training data by querying the model. Prior work has shown that the attack is feasible when the model is overfitted to its training data or when the adversary controls the training algorithm. However, when the model is not overfitted and the adversary does not control the training algorithm, the threat is not well understood. In this paper, we report a study that discovers overfitting to be a sufficient but not a necessary condition for an MIA to succeed. More specifically, we demonstrate that even a well-generalized model contains vulnerable instances subject to a new generalized MIA (GMIA). In GMIA, we use novel techniques for selecting vulnerable instances and detecting their subtle influences ignored by overfitting metrics. Specifically, we successfully identify individual records with high precision in real-world datasets by querying black-box machine learning models. Further we show that a vulnerable record can even be indirectly attacked by querying other related records and existing generalization techniques are found to be less effective in protecting the vulnerable instances. Our findings sharpen the understanding of the fundamental cause of the problem: the unique influences the training instance may have on the model.
cs.CR↗
Classification of proper holomorphic mappings between certain unbounded non-hyperbolic domains
The Fock-Bargmann-Hartogs domain $D_{n,m}(μ)$ ($μ>0$) in $\mathbb{C}^{n+m}$ is defined by the inequality $\|w\|^2<e^{-μ\|z\|^2},$ where $(z,w)\in \mathbb{C}^n\times \mathbb{C}^m$, which is an unbounded non-hyperbolic domain in $\mathbb{C}^{n+m}$. Recently, Tu-Wang obtained the rigidity result that proper holomorphic self-mappings of $D_{n,m}(μ)$ are automorphisms for $m\geq 2$, and found a counter-example to show that the rigidity result isn't true for $D_{n,1}(μ)$. In this article, we obtain a classification of proper holomorphic mappings between $D_{n,1}(μ)$ and $D_{N,1}(μ)$ with $N<2n$.
math.CV↗
A veracity preserving model for synthesizing scalable electricity load profiles
Electricity users are the major players of the electric systems, and electricity consumption is growing at an extraordinary rate. The research on electricity consumption behaviors is becoming increasingly important to design and deployment of the electric systems. Unfortunately, electricity load profiles are difficult to acquire. Data synthesis is one of the best approaches to solving the lack of data, and the key is the model that preserves the real electricity consumption behaviors. In this paper, we propose a hierarchical multi-matrices Markov Chain (HMMC) model to synthesize scalable electricity load profiles that preserve the real consumption behavior on three time scales: per day, per week, and per year. To promote the research on the electricity consumption behavior, we use the HMMC approach to model two distinctive raw electricity load profiles. One is collected from the resident sector, and the other is collected from the non-resident sectors, including different industries such as education, finance, and manufacturing. The experiments show our model performs much better than the classical Markov Chain model. We publish two trained models online, and researchers can directly use these trained models to synthesize scalable electricity load profiles for further researches.
cs.OH↗
Chemical Responsive Single Crystal Organic Magnet
Materials that change their magnetic properties in response to the external stimuli have long been of interest for their potential applicability in magnetic storage device, spintronics and smart magnetic materials. Organic materials are suitable candidates for such materials due to their chemical diversity, flexibility and designablity. However, most methods used for changing magnetism are inefficient or destructive to the magnetic material. Hence there is a need for innovation in this field. Here we report high-performance magnetic control of a gas-responsive single-molecule magnet (SMM). The results exhibit that the magnetic properties of the SMM can be significantly changed according to the gas environment it is in and some of the magnetic states can be reversibly transformed or coexistent in the SMM through artificial control. More importantly, the monocrystalline structure of the SMM remains unchanged during the transformation process except for slight change of the lattice constant. Thus, this work opens up new insights into the stimuli-responsive magnetic materials which have great prospects for application in artificial design magnetic network and also highlight their potential as smart materials.
cond-mat.mtrl-sci↗
Equivalence of restricted Boltzmann machines and tensor network states
The restricted Boltzmann machine (RBM) is one of the fundamental building blocks of deep learning. RBM finds wide applications in dimensional reduction, feature extraction, and recommender systems via modeling the probability distributions of a variety of input data including natural images, speech signals, and customer ratings, etc. We build a bridge between RBM and tensor network states (TNS) widely used in quantum many-body physics research. We devise efficient algorithms to translate an RBM into the commonly used TNS. Conversely, we give sufficient and necessary conditions to determine whether a TNS can be transformed into an RBM of given architectures. Revealing these general and constructive connections can cross-fertilize both deep learning and quantum many-body physics. Notably, by exploiting the entanglement entropy bound of TNS, we can rigorously quantify the expressive power of RBM on complex data sets. Insights into TNS and its entanglement capacity can guide the design of more powerful deep learning architectures. On the other hand, RBM can represent quantum many-body states with fewer parameters compared to TNS, which may allow more efficient classical simulations.
cond-mat.str-el↗