arXiv subjects

Feng Liu

Publications and source records attributed to Feng Liu.

At least 667 records · Page 37Linked to original sources

arXiv2018

Learning Environmental Sounds with Multi-scale Convolutional Neural Network

Deep learning has dramatically improved the performance of sounds recognition. However, learning acoustic models directly from the raw waveform is still challenging. Current waveform-based models generally use time-domain convolutional layers to extract features. The features extracted by single size filters are insufficient for building discriminative representation of audios. In this paper, we propose multi-scale convolution operation, which can get better audio representation by improving the frequency resolution and learning filters cross all frequency area. For leveraging the waveform-based features and spectrogram-based features in a single model, we introduce two-phase method to fuse the different features. Finally, we propose a novel end-to-end network called WaveMsNet based on the multi-scale convolution operation and two-phase method. On the environmental sounds classification datasets ESC-10 and ESC-50, the classification accuracies of our WaveMsNet achieve 93.75% and 79.10% respectively, which improve significantly from the previous methods.

cs.SD↗

arXiv2018

Prediction of two-dimensional nodal-line semimetal in a carbon nitride covalent network

Carbon nitride compounds have emerged recently as a prominent member of 2D materials beyond graphene. The experimental realizations of 2D graphitic carbon nitride g-C$_3$N$_4$, nitrogenated holey grahpene C$_2$N, polyaniline C$_3$N have shown their promising potential in energy and environmental applications. In this work, we predict a new type of carbon nitride network with a C$_9$N$_4$ stoichiometry from first principle calculations. Unlike common C-N compounds and covalent organic frameworks (COFs), which are typically insulating, surprisingly C$_9$N$_4$ is found to be a 2D nodal-line semimetal (NLSM). The nodal line in C$_9$N$_4$ forms a closed ring centered at $Γ$ point, which originates from the pz orbitals of both C and N. The linear crossing happens right at Fermi level contributed by two sets of dispersive Kagome and Dirac bands, which is robust due to negligible spin-orbital-coupling (SOC) in C and N. Besides, it is revealed that the formation of nodal ring is of accidental band degeneracy in nature induced by the chemical potential difference of C and N, as validated by a single orbital tight-binding model, rather than protected by crystal in-plane mirror symmetry or band topology. Interestingly, a new structure of nodal line, i.e., nodal-cylinder, is found in momentum space for AA-stacking C$_9$N$_4$. Our results imply possible functionalization for a novel metal-free C-N covalent network with interesting semimetallic properties.

cond-mat.mtrl-sci↗

arXiv2018

iVQA: Inverse Visual Question Answering

We propose the inverse problem of Visual question answering (iVQA), and explore its suitability as a benchmark for visuo-linguistic understanding. The iVQA task is to generate a question that corresponds to a given image and answer pair. Since the answers are less informative than the questions, and the questions have less learnable bias, an iVQA model needs to better understand the image to be successful than a VQA model. We pose question generation as a multi-modal dynamic inference process and propose an iVQA model that can gradually adjust its focus of attention guided by both a partially generated question and the answer. For evaluation, apart from existing linguistic metrics, we propose a new ranking metric. This metric compares the ground truth question's rank among a list of distractors, which allows the drawbacks of different algorithms and sources of error to be studied. Experimental results show that our model can generate diverse, grammatically correct and content correlated questions that match the given answer.

cs.CV↗

arXiv2018

City Brain, a New Architecture of Smart City Based on the Internet Brain

In the ten years after the Smart City was put forward, there are still problems like unclear concept, lack of top-down design and information island. With the further development of the Internet, the brain-like architecture of the Internet is becoming clearer and clearer. As a product of combination of city buildings and the Internet, the Smart City will also have a new architecture, and the city brain thus appears. Based on the Internet Brain, this paper describes how to construct the Smart City in the form of brain-like tissue, and how to evaluate the construction level of the Smart City (City IQ) relying on the Big SNS (city neural networks) and city cloud reflex arcs.

cs.CY↗

arXiv2018

Inverse Visual Question Answering: A New Benchmark and VQA Diagnosis Tool

In recent years, visual question answering (VQA) has become topical. The premise of VQA's significance as a benchmark in AI, is that both the image and textual question need to be well understood and mutually grounded in order to infer the correct answer. However, current VQA models perhaps `understand' less than initially hoped, and instead master the easier task of exploiting cues given away in the question and biases in the answer distribution. In this paper we propose the inverse problem of VQA (iVQA). The iVQA task is to generate a question that corresponds to a given image and answer pair. We propose a variational iVQA model that can generate diverse, grammatically correct and content correlated questions that match the given answer. Based on this model, we show that iVQA is an interesting benchmark for visuo-linguistic understanding, and a more challenging alternative to VQA because an iVQA model needs to understand the image better to be successful. As a second contribution, we show how to use iVQA in a novel reinforcement learning framework to diagnose any existing VQA model by way of exposing its belief set: the set of question-answer pairs that the VQA model would predict true for a given image. This provides a completely new window into what VQA models `believe' about images. We show that existing VQA models have more erroneous beliefs than previously thought, revealing their intrinsic weaknesses. Suggestions are then made on how to address these weaknesses going forward.

cs.CV↗

arXiv2018

Alloy engineering of topological semimetal phase transition in MgTa$_{2-x}$Nb$_x$N$_3$

Dirac, triple-point and Weyl fermions represent three topological semimetal phases, characterized with a descending degree of band degeneracy, which have been realized separately in specific crystalline materials with different lattice symmetries. Here we demonstrate an alloy engineering approach to realize all three types of fermions in one single material system of MgTa$_{2-x}$Nb$_x$N$_3$. Based on symmetry analysis and first-principles calculations, we map out a phase diagram of topological order in the parameter space of alloy concentration and crystalline symmetry, where the intrinsic MgTa$_2$N$_3$ with the highest symmetry hosts the Dirac semimetal phase which transforms into the triple-point and then the Weyl semimetal phase with the increasing Nb concentration that lowers the crystalline symmetries. Therefore, alloy engineering affords a unique approach for experimental investigation of topological transitions of semimetallic phases manifesting different fermionic behaviors.

cond-mat.mtrl-sci↗

arXiv2018

Locating All Real Solutions of Power Flow Equations: A Convex Optimization Based Method

This paper proposes a convex optimization based method that either locates all real roots of a set of power flow equations or declares no real solution exists in the given area. In the proposed method, solving the power flow equations is reformulated as a global optimization problem (GPF for short) that minimizes the sum of slack variables. All the global minima of GPF with a zero objective value have a one-to-one correspondence to the real roots of power flow equations. By solving a relaxed version of GPF over a hypercube, if the optimal value is strictly positive, there is no solution in this area and the hypercube is discarded. Otherwise the hypercube is further divided into smaller ones. This procedure repeats recursively until all the real roots are located in small enough hypercubes through the successive refinement of the feasible region embedded in a bisection paradigm. This method is desired in a number of power system security assessment applications, for instance, the transient stability analysis as well as voltage stability analysis, where the closest unstable equilibrium and all Type-I unstable equilibrium is required, respectively. The effectiveness of the proposed method is verified by analyzing several test systems.

math.OC↗

arXiv2018

Novel Approaches to Accelerating the Convergence Rate of Markov Decision Process for Search Result Diversification

Recently, some studies have utilized the Markov Decision Process for diversifying (MDP-DIV) the search results in information retrieval. Though promising performances can be delivered, MDP-DIV suffers from a very slow convergence, which hinders its usability in real applications. In this paper, we aim to promote the performance of MDP-DIV by speeding up the convergence rate without much accuracy sacrifice. The slow convergence is incurred by two main reasons: the large action space and data scarcity. On the one hand, the sequential decision making at each position needs to evaluate the query-document relevance for all the candidate set, which results in a huge searching space for MDP; on the other hand, due to the data scarcity, the agent has to proceed more "trial and error" interactions with the environment. To tackle this problem, we propose MDP-DIV-kNN and MDP-DIV-NTN methods. The MDP-DIV-kNN method adopts a $k$ nearest neighbor strategy, i.e., discarding the $k$ nearest neighbors of the recently-selected action (document), to reduce the diversification searching space. The MDP-DIV-NTN employs a pre-trained diversification neural tensor network (NTN-DIV) as the evaluation model, and combines the results with MDP to produce the final ranking solution. The experiment results demonstrate that the two proposed methods indeed accelerate the convergence rate of the MDP-DIV, which is 3x faster, while the accuracies produced barely degrade, or even are better.

cs.IR↗

arXiv2018

Distributed Optimal Frequency Control Considering a Nonlinear Network-Preserving Model

This paper addresses the distributed optimal frequency control of power systems considering a network-preserving model with nonlinear power flows and excitation voltage dynamics. Salient features of the proposed distributed control strategy are fourfold: i) nonlinearity is considered to cope with large disturbances; ii) only a part of generators are controllable; iii) no load measurement is required; iv) communication connectivity is required only for the controllable generators. To this end, benefiting from the concept of 'virtual load demand', we first design the distributed controller for the controllable generators by leveraging the primal-dual decomposition technique. We then propose a method to estimate the virtual load demand of each controllable generator based on local frequencies. We derive incremental passivity conditions for the uncontrollable generators. Finally, we prove that the closed-loop system is asymptotically stable and its equilibrium attains the optimal solution to the associated economic dispatch problem. Simulations, including small and large-disturbance scenarios, are carried on the New England system, demonstrating the effectiveness of our design.

eess.SY↗

arXiv2018

Regularity and continuity of the multilinear strong maximal operators

Let $m\ge 1$, in this paper, our object of investigation is the regularity and and continuity properties of the following multilinear strong maximal operator $${\mathscr{M}}_{\mathcal{R}}(\vec{f})(x)=\sup_{\substack{R \ni x R\in\mathcal{R}}}\prod\limits_{i=1}^m\frac{1}{|R|}\int_{R}|f_i(y)|dy,$$ where $x\in\mathbb{R}^d$ and $\mathcal{R}$ denotes the family of all rectangles in $\mathbb{R}^d$ with sides parallel to the axes. When $m=1$, denote $\mathscr{M}_{\mathcal{R}}$ by $\mathcal {M}_{\mathcal{R}}$.Then, $\mathcal {M}_{\mathcal{R}}$ coincides with the classical strong maximal function initially studied by Jessen, Marcinkiewicz and Zygmund. We showed that ${\mathscr{M}}_{\mathcal{R}}$ is bounded and continuous from the Sobolev spaces $W^{1,p_1}(\mathbb{R}^d)\times \cdots\times W^{1,p_m}(\mathbb{R}^d)$ to $W^{1,p} (\mathbb{R}^d)$, from the Besov spaces $B_{s}^{p_1,q} (\mathbb{R}^d)\times\cdots\times B_s^{p_m,q}(\mathbb{R}^d)$ to $B_s^{p,q}(\mathbb{R}^d)$, from the Triebel-Lizorkin spaces $F_{s}^{p_1,q}(\mathbb{R}^d)\times\cdots\times F_s^{p_m,q}(\mathbb{R}^d)$ to $F_s^{p,q}(\mathbb{R}^d)$. As a consequence, we further showed that ${\mathscr{M}}_{\mathcal{R}}$ is bounded and continuous from the fractional Sobolev spaces $W^{s,p_1}(\mathbb{R}^d)\times \cdots\times W^{s,p_m}(\mathbb{R}^d)$ to $W^{s,p}(\mathbb{R}^d)$ for $0<s\leq 1$ and $1<p<\infty$. As an application, we obtain a weak type inequality for the Sobolev capacity, which can be used to prove the $p$-quasicontinuity of $\mathscr{M}_{\mathcal{R}}$. The discrete type of the strong maximal operators has also been considered. We showed that this discrete type of the maximal operators enjoys somewhat unexpected regularity properties.

math.CA↗

arXiv2018

Ubiquitous Ideal Spin-Orbit Coupling in a Screw Dislocation in Semiconductors

We theoretically demonstrate that screw dislocation (SD), a 1D topological defect widely present in semiconductors, exhibits ubiquitously a new form of spin-orbit coupling (SOC) effect. Differing from the widely known conventional 2D Rashba-Dresselhaus (RD) SOC effect that typically exists at surfaces/interfaces, the deep-level nature of SD-SOC states in semiconductors readily makes it an ideal SOC. Remarkably, the spin texture of 1D SD-SOC, pertaining to the inherent symmetry of SD, exhibits a significantly higher degree of spin coherency than the 2D RD-SOC. Moreover, the 1D SD-SOC can be tuned by ionicity in compound semiconductors to ideally suppress spin relaxation, as demonstrated by comparative first-principles calculations of SDs in Si/Ge, GaAs, and SiC. Our findings therefore open a new door to manipulating spin transport in semiconductors by taking advantage of an otherwise detrimental topological defect.

cond-mat.mtrl-sci↗

arXiv2018

Analysis of the Relation between Artificial Intelligence and the Internet from the Perspective of Brain Science

Artificial intelligence (AI) like deep learning, cloud AI computation has been advancing at a rapid pace since 2014. There is no doubt that the prosperity of AI is inseparable with the development of the Internet. However, there has been little attention to the link between AI and the internet. This paper explores them with brain insights mainly from four views:1) How is the general relation between artificial intelligence and Internet of Things, cloud computing, big data and Industrial Internet from the perspective of brain science. 2) Construction of a new AI system model with the Internet and brain science.

cs.CY↗

arXiv2017

Li doping kagome spin liquid compounds

Herbertsmithite and Zn-doped barlowite are two compounds for experimental realization of twodimensional gapped kagome spin liquid. Theoretically, it has been proposed that charge doping a quantum spin liquid gives rise to exotic metallic states, such as high-temperature superconductivity. However, one recent experiment about herbertsmithite with successful Li-doping shows surprisingly the insulating state even under the heavy doped scenario, which can hardly be explained by many-body physics. Using first-principles calculation, we performed a comprehensive study about the Li intercalated doping effect of these two compounds. For the Li-doped herbertsmithite, we identified the optimized Li position at the Cl-(OH)$_3$-Cl pentahedron site instead of previously speculated Cl-(OH)$_3$ tetrahedral site. With the increase of Li doping concentration, the saturation magnetization decreases linearly due to the charge transfer from Li to Cu ions. Moreover, we found that Li forms chemical bonds with the nearby (OH)$^-$ and Cl$^-$ ions, which lowers the surrounding chemical potential and traps the electron, as evidenced by the localized charge distribution, explaining the insulating behavior measured experimentally. Though with different structure from herbertsmithite, Zn-doped Barlowite shows the same features upon Li doping. We conclude that Li doping this family of kagome spin liquid cannot realize exotic metallic states, other methods should be further explored, such as element substitution with different valence electrons.

cond-mat.str-el↗

arXiv2017

Effects of mean-field and the softening of equation of state on elliptic flow in Au+Au collision at $\sqrt{s_{NN}}$ = 5 GeV from JAM model

We perform a systematic study of elliptic flow ($v_2$) in Au+Au collision at $\sqrt{s_{NN}}=5$ GeV by using a microscopic transport model JAM. The centrality, pseudorapidity, transverse momentum and beam energy dependence of $v_2$ for charged as well as identified hadrons are studied. We investigate the effects of both the hadronic mean-field and the softening of equation of state (EoS) on elliptic flow. The softening of EoS is realized by imposing attractive orbits in two body scattering, which can reduce the pressure of the system. We found that the softening of EoS leads to the enhancement of $v_2$, while the hadronic mean-field suppresses $v_2$ relative to the cascade mode. It indicates that elliptic flow at high baryon density regions is highly sensitive to the EoS and the enhancement of $v_2$ may probe the signature of a first-order phase transition in heavy-ion collisions at beam energies of a strong baryon stopping region.

nucl-ex↗

arXiv2017

Three IQs of AI Systems and their Testing Methods

The rapid development of artificial intelligence has brought the artificial intelligence threat theory as well as the problem about how to evaluate the intelligence level of intelligent products. Both need to find a quantitative method to evaluate the intelligence level of intelligence systems, including human intelligence. Based on the standard intelligence system and the extended Von Neumann architecture, this paper proposes General IQ, Service IQ and Value IQ evaluation methods for intelligence systems, depending on different evaluation purposes. Among them, the General IQ of intelligence systems is to answer the question of whether the artificial intelligence can surpass the human intelligence, which is reflected in putting the intelligence systems on an equal status and conducting the unified evaluation. The Service IQ and Value IQ of intelligence systems are used to answer the question of how the intelligent products can better serve the human, reflecting the intelligence and required cost of each intelligence system as a product in the process of serving human.

cs.AI↗

arXiv2017

A Conditional Model of Wind Power Forecast Errors and Its Application in Scenario Generation

In power system operation, characterizing the stochastic nature of wind power is an important albeit challenging issue. It is well known that distributions of wind power forecast errors often exhibit significant variability with respect to different forecast values. Therefore, appropriate probabilistic models that can provide accurate information for conditional forecast error distributions are of great need. On the basis of Gaussian mixture model, this paper constructs analytical conditional distributions of forecast errors for multiple wind farms with respect to forecast values. The accuracy of the proposed probabilistic models is verified by using historical data. Thereafter, a fast sampling method is proposed to generate scenarios from the conditional distributions which are non-Gaussian and interdependent. The efficiency of the proposed sampling method is verified.

physics.data-an↗

arXiv2017

An Adjustable Chance-Constrained Approach for Flexible Ramping Capacity Allocation

With the fast growth of wind power penetration, power systems need additional flexibility to cope with wind power ramping. Several electricity markets have established requirements for flexible ramping capacity (FRC) reserves. This paper addresses two crucial issues that have rarely been discussed in the literature: 1) how to characterize wind power ramping under different forecast values and 2) how to achieve a reasonable trade-off between operational risks and FRC costs. Regarding the first issue, this paper proposes a concept of conditional distributions of wind power ramping, which is empirically verified by using simulation and real-world data. For the second issue, this paper develops an adjustable chance-constrained approach to optimally allocate FRC reserves. Equivalent tractable forms of the original problem are devised to improve computational efficiency. Tests carried out on a modified IEEE 118-bus system demonstrate the effectiveness and efficiency of the proposed method.

physics.soc-ph↗

arXiv2017

Actor-Critic Sequence Training for Image Captioning

Generating natural language descriptions of images is an important capability for a robot or other visual-intelligence driven AI agent that may need to communicate with human users about what it is seeing. Such image captioning methods are typically trained by maximising the likelihood of ground-truth annotated caption given the image. While simple and easy to implement, this approach does not directly maximise the language quality metrics we care about such as CIDEr. In this paper we investigate training image captioning methods based on actor-critic reinforcement learning in order to directly optimise non-differentiable quality metrics of interest. By formulating a per-token advantage and value computation strategy in this novel reinforcement learning based captioning model, we show that it is possible to achieve the state of the art performance on the widely used MSCOCO benchmark.

cs.CV↗