arXiv subjects

Zhen Huang

Publications and source records attributed to Zhen Huang.

171 records · Page 10Linked to original sources

arXiv2015

Maximum a Posteriori Adaptation of Network Parameters in Deep Models

We present a Bayesian approach to adapting parameters of a well-trained context-dependent, deep-neural-network, hidden Markov model (CD-DNN-HMM) to improve automatic speech recognition performance. Given an abundance of DNN parameters but with only a limited amount of data, the effectiveness of the adapted DNN model can often be compromised. We formulate maximum a posteriori (MAP) adaptation of parameters of a specially designed CD-DNN-HMM with an augmented linear hidden networks connected to the output tied states, or senones, and compare it to feature space MAP linear regression previously proposed. Experimental evidences on the 20,000-word open vocabulary Wall Street Journal task demonstrate the feasibility of the proposed framework. In supervised adaptation, the proposed MAP adaptation approach provides more than 10% relative error reduction and consistently outperforms the conventional transformation based methods. Furthermore, we present an initial attempt to generate hierarchical priors to improve adaptation efficiency and effectiveness with limited adaptation data by exploiting similarities among senones.

cs.LG↗

arXiv2015

Feature Selection with Redundancy-complementariness Dispersion

Feature selection has attracted significant attention in data mining and machine learning in the past decades. Many existing feature selection methods eliminate redundancy by measuring pairwise inter-correlation of features, whereas the complementariness of features and higher inter-correlation among more than two features are ignored. In this study, a modification item concerning the complementariness of features is introduced in the evaluation criterion of features. Additionally, in order to identify the interference effect of already-selected False Positives (FPs), the redundancy-complementariness dispersion is also taken into account to adjust the measurement of pairwise inter-correlation of features. To illustrate the effectiveness of proposed method, classification experiments are applied with four frequently used classifiers on ten datasets. Classification results verify the superiority of proposed method compared with five representative feature selection methods.

cs.LG↗

arXiv2013

Effect of growth oxygen pressure on anisotropic-strain-induced phase separation in epitaxial La$_{0.67}$Ca$_{0.33}$MnO$_{3}$/NdGaO$_{3}$(001) films

The effect of deposition oxygen pressure (P$_{O}$) on phase separation (PS) induced in epitaxial La$_{0.67}$Ca$_{0.33}$MnO$_{3}$/NdGaO$_{3}$(001) films was investigated. Fully oxygenated films grown at high P$_{O}$ are anisotropically strained. They exhibit PS over a wide temperature range, because of the large orthorhombicity of NdGaO$_{3}$ substrates. The paramagnetic insulator-to-ferromagnetic metal (FM) and FM-to-antiferromagnetic insulator (AFI) transitions gradually shift to lower temperatures with decreasing PO. The AFI state is initially weakened (P$_{O}$ >= 30 Pa), but then becomes more robust against the magnetic field (P$_{O}$ < 30 Pa). The out-of-plane film lattice parameter increases with decreasing P$_{O}$. For films grown at P$_{O}$>= 30 Pa, the slight oxygen deficiency may enlarge the lattice unit cell, reduce the anisotropic strain and suppress the AFI state. Films deposited at P$_{O}$ < 30 Pa instead experience an average compressive strain. The enhanced compressive strain and structural defects in the films may lead to the robust AFI state. These results aid our understanding of PS in manganite films.

cond-mat.str-el↗

arXiv2011

Data-centric Misbehavior Detection in VANETs

Detecting misbehavior (such as transmissions of false information) in vehicular ad hoc networks (VANETs) is very important problem with wide range of implications including safety related and congestion avoidance applications. We discuss several limitations of existing misbehavior detection schemes (MDS) designed for VANETs. Most MDS are concerned with detection of malicious nodes. In most situations, vehicles would send wrong information because of selfish reasons of their owners, e.g. for gaining access to a particular lane. Because of this (\emph{rational behavior}), it is more important to detect false information than to identify misbehaving nodes. We introduce the concept of data-centric misbehavior detection and propose algorithms which detect false alert messages and misbehaving nodes by observing their actions after sending out the alert messages. With the data-centric MDS, each node can independently decide whether an information received is correct or false. The decision is based on the consistency of recent messages and new alert with reported and estimated vehicle positions. No voting or majority decisions is needed, making our MDS resilient to Sybil attacks. Instead of revoking all the secret credentials of misbehaving nodes, as done in most schemes, we impose fines on misbehaving nodes (administered by the certification authority), discouraging them to act selfishly. This reduces the computation and communication costs involved in revoking all the secret credentials of misbehaving nodes.

cs.NI↗

arXiv2009

Simulated Quantum Computation of Global Minima

Finding the optimal solution to a complex optimization problem is of great importance in practically all fields of science, technology, technical design and econometrics. We demonstrate that a modified Grover's quantum algorithm can be applied to real problems of finding a global minimum using modest numbers of quantum bits. Calculations of the global minimum of simple test functions and Lennard-Jones clusters have been carried out on a quantum computer simulator using a modified Grover's algorithm. The number of function evaluations $N$ reduced from O(N) in classical simulation to $O(\sqrt{N})$ in quantum simulation. We also show how the Grover's quantum algorithm can be combined with the classical Pivot method for global optimization to treat larger systems.

quant-ph↗

arXiv2008

Entanglement, Berry Phases, and Level Crossings for the Atomic Breit-Rabi Hamiltonian

The relation between level crossings, entanglement, and Berry phases is investigated for the Breit-Rabi Hamiltonian of hydrogen and sodium atoms, describing a hyperfine interaction of electron and nuclear spins in a magnetic field. It is shown that the entanglement between nuclear and electron spins is maximum at avoided crossings. An entangled state encircling avoided crossings acquires a marginal Berry phase of a subsystem like an instantaneous eigenstate moving around real crossings accumulates a Berry phase. Especially, the nodal points of a marginal Berry phase correspond to the avoided crossing points.

quant-ph↗

arXiv2008

Nuclear-induced time evolution of entanglement of two-electron spins in anisotropically coupled quantum dot

We study the time evolution of entanglement of two spins in anisotropically coupled quantum dot interacting with the unpolarized nuclear spins environment. We assume that the exchange coupling strength in the z-direction $J_z$ is different from the lateral one $J_l$. We observe that the entanglement decays as a result of the coupling to the nuclear environment and reaches a saturation value, which depends on the value of the exchange interaction difference $J=\| J_l-J_z\|$ between the two spins and the strength of the applied external magnetic field. We find that the entanglement exhibits a critical behavior controlled by the competition between the exchange interaction $J$ and the external magnetic field. The entanglement shows a quasi-symmetric behavior above and below a critical value of the exchange interaction. It becomes more symmetric as the external magnetic field increases. The entanglement reaches a large saturation value, close to unity, when the exchange interaction is far above or below its critical value and a small one as it closely approaches the critical value. Furthermore, we find that the decay rate profile of entanglement is linear when the exchange interaction is much higher or lower than the critical value but converts to a power law and finally to a Gaussian as the critical value is approached from both directions. The dynamics of entanglement is found to be independent of the exchange interaction for isotropically coupled quantum dot.

cond-mat.mes-hall↗

arXiv2005

Entanglement as Measure of Electron-Electron Correlation in Quantum Chemistry Calculations

In quantum chemistry calculations, the correlation energy is defined as the difference between the Hartree-Fock limit energy and the exact solution of the nonrelativistic Schrodinger equation. With this definition, the electron correlation effects are not directly observable. In this report, we show that the entanglement can be used as an alternative measure of the electron correlation in quantum chemistry calculations. Entanglement is directly observable and it is one of the most striking properties of quantum mechanics. As an example we calculate the entanglement for He atom and H2 molecule with different basis sets.

quant-ph↗

arXiv2005

Dynamics of Entanglement for One-Dimensional Spin Systems in an External Time-Dependent Magnetic Field

We study the dynamics of entanglement for the XY-model, one-dimensional spin systems coupled through nearest neighbor exchange interaction and subject to an external time-dependent magnetic field. Using the two-site density matrix, we calculate the time-dependent entanglement of formation between nearest neighbor qubits. We investigate the effect of varying the temperature, the anisotropy parameter and the external time-dependent magnetic field on the entanglement. We have found that the entanglement can be localized between nearest neighbor qubits for certain values of the external time-dependent magnetic field. Moreover, as known for the magnetization of this model, the entanglement shows nonergodic behavior, it does not approach its equilibrium value at the infinite time limit.

quant-ph↗