In this paper we present methods for reducing the computation requirements of joint segmentation and recognition of phones using the stochastic segment model (SSM). Our approach to the problem is twofold: first, we pr...
详细信息
In this paper we present methods for reducing the computation requirements of joint segmentation and recognition of phones using the stochastic segment model (SSM). Our approach to the problem is twofold: first, we present a fast segment classification method that reduces computation by a factor of two to four, depending on the confidence of choosing the most probable model. Second, we propose a split-and-merge segmentation algorithm as an alternative to the typical dynamic-programming solution of the segmentation and recognition problem, with computation savings increasing proportionally with model complexity. Although our current recognizer uses context-independent phone models, the results that we report on the TIMIT data base for speaker-independent joint segmentation and recognition are comparable to those of systems that use context information.
Evolutionary algorithms (EAs) are increasingly popular approaches to multi-objective optimization. One of their significant advantages is that they can directly optimize the Pareto front by evolving a population of so...
详细信息
Evolutionary algorithms (EAs) are increasingly popular approaches to multi-objective optimization. One of their significant advantages is that they can directly optimize the Pareto front by evolving a population of solutions, where the recombination (also called crossover) operators are usually employed to reproduce new and potentially better solutions by mixing up solutions in the population. Recombination in multi-objective evolutionary algorithms is, however, mostly applied heuristically. In this paper, we investigate how from a theoretical viewpoint a recombination operator will affect a multi-objective EA. First, we employ artificial benchmark problems: the weighted LPTNO problem (a generalization of the well-studied LOTZ problem), and the well-studied Cocz problem, for studying the effect of recombination. Our analysis discloses that recombination may accelerate the filling of the Pareto front by recombining diverse solutions and thus help solve multi-objective optimization. Because of this, for these two problems, we find that a multi-objective EA with recombination enabled achieves a better expected running time than any known EAs with recombination disabled. We further examine the effect of recombination on solving the multi-objective minimum spanning tree problem, which is an NP-hard problem. Following our finding on the artificial problems, our analysis shows that recombination also helps accelerate filling the Pareto front and thus helps find approximate solutions faster. (C) 2013 Elsevier B.V. All rights reserved.
We consider a simplified version of NOT-ALL-EQUAL 3-Sat, a variation of the famous SATISFIABILITY problem, where each clause is made up of exactly three distinct literals and the question is whether there exists a tru...
详细信息
We consider a simplified version of NOT-ALL-EQUAL 3-Sat, a variation of the famous SATISFIABILITY problem, where each clause is made up of exactly three distinct literals and the question is whether there exists a truth assignment such that for each clause at least one literal is set to true and at least one is set to false. We show that NOT-ALL-EQUAL 3-Sat remains NP-complete if (1) each variable appears exactly four times, (2) there are no negations in the formula, and (3) the formula is linear, i.e., each pair of distinct clauses shares at most one variable. Therewith, we improve upon two results in the literature. (C) 2020 Elsevier B.V. All rights reserved.
Correctness tools can improve software-development by supplementing human shortcomings-detecting programming errors such as null-pointer references, API usage errors, and failing to close file descriptors. Microsoft R...
详细信息
Correctness tools can improve software-development by supplementing human shortcomings-detecting programming errors such as null-pointer references, API usage errors, and failing to close file descriptors. Microsoft Research has developed two generations of correctness tools. Together, these tools offer a systematic approach for finding and fixing errors early in the development process.
Being motivated by John Tantalo's Planarity Game, we consider straight line plane drawings of a planar graph G with edge crossings and wonder how obfuscated such drawings can be. We define obf (G), the obfuscation...
详细信息
Being motivated by John Tantalo's Planarity Game, we consider straight line plane drawings of a planar graph G with edge crossings and wonder how obfuscated such drawings can be. We define obf (G), the obfuscation complexity of G, to be the maximum number of edge crossings in a drawing of G. Relating obf (G) to the distribution of vertex degrees in G, we show an efficient way of constructing a drawing of G with at least obf (G)/3 edge crossings. We prove bounds (delta(G)(2)/24 - o(1)) n(2) <= obf(G) < 3 n(2) for an n-vertex planar graph G with minimum vertex degree delta(G) >= 2. The shift complexity of G, denoted by shift(G), is the minimum number of vertex shifts sufficient to eliminate all edge crossings in an arbitrarily obfuscated drawing of G (after shifting a vertex, all incident edges are supposed to be redrawn correspondingly). If delta (G) >= 3, then shift(G) is linear in the number of vertices due to the known fact that the matching number of G is linear. However, in the case delta (G) >= 2 we notice that shift(G) can be linear even if the matching number is bounded. As for computational complexity, we show that, given a drawing D of a planar graph, it is NP-hard to find an optimum sequence of shifts making D crossing-free. (C) 2008 Elsevier B.V. All rights reserved.
The quadratic matrix equation (QME)Q(X)= AX(2 )+ BX +C = 0occurs in the branches such as the quadratic eigenvalue problems and quasi-birth-death processes. Also, the numerical solution of QMEs is an essential step in ...
详细信息
The quadratic matrix equation (QME)Q(X)= AX(2 )+ BX +C = 0occurs in the branches such as the quadratic eigenvalue problems and quasi-birth-death processes. Also, the numerical solution of QMEs is an essential step in many computational methods for linear-quadratic and robust control, filtering, controller order reduction, inner-outer factorization, spectral factorization, and other applications. In this study, schemes are presented to solve the QME based on the Hermitian and skew-Hermitian splitting (HSS). It is shown that the proposed schemes converge to the solutions of the QME. Finally, some examples are solved to discover the application of the schemes in comparison with Newton's method.
作者：
Sun, WenhaoZheng, JianpingXidian Univ
State Key Lab Integrated Serv Network Collaborat Innovat Ctr Informat Sensing & Underst Sch Telecommun Engn Xian 710071 Peoples R China
The retention noise in MLC NAND flash memory has a great impact on reliability and lifetime of devices. How to recover information from the data corrupted by retention noise is a challenging problem. The joint process...
详细信息
The retention noise in MLC NAND flash memory has a great impact on reliability and lifetime of devices. How to recover information from the data corrupted by retention noise is a challenging problem. The joint processing of channel estimation, detection and low-density parity-check (LDPC) decoding is an effective method to handle this problem, while which is hindered by the limited computational resource of NAND flash memory devices. This paper proposes a discrete retention-failure recovery method, where the reading system only processes integers. The proposed method consists of two major components referred to binary tree searching (BTS) based channel updating and information bottleneck (IB) based LDPC decoding. In channel updating, the proposed learning based BTS algorithm can acquire reference voltages directly;thus it bypasses the intermediate stage of retention noise parameter estimation. The LDPC decoder is realized by IB based message passing, where node operation is achieved by lookup tables, and the messages passed between nodes are integers. Besides, a dynamic quantization level reduction strategy for IB based decoding is proposed to reduce the memory consumption of lookup tables. Finally, using PEG-constructed regular LDPC code (8000, 7200), the numerical results show that the proposed discrete method has much lower computational complexity and similar performance compared with the conventional method.
We show that the set of square numbers is the spectrum of an F Sigma(1,1)(1)-sentence (existential second order logic with quantification over unary function variables) with two unary function variables, but it is not...
详细信息
We show that the set of square numbers is the spectrum of an F Sigma(1,1)(1)-sentence (existential second order logic with quantification over unary function variables) with two unary function variables, but it is not the spectrum of F Sigma(1,1)(1)-sentences with only one unary function variable. (C) 1997 Elsevier Science B.V.
In the fifth-generation (5G) high-speed train (HST) system operating in the millimeter-wave (mmWave) band, a much higher Doppler shift occurs. Doppler shift severely degrades reception performance in orthogonal freque...
详细信息
In the fifth-generation (5G) high-speed train (HST) system operating in the millimeter-wave (mmWave) band, a much higher Doppler shift occurs. Doppler shift severely degrades reception performance in orthogonal frequency division multiplexing (OFDM)-based wireless communication systems. The performance of the Doppler shift estimator is directly related to safety in the HST because the 5G HST system is used for train control. Therefore, it is necessary to develop a fast and accurate Doppler shift estimator (DSE) with low complexity. In this paper, we propose a new machine learning-based DSE (MLDSE). Taking note of the fact that an HST travels the same path repeatedly, the MLDSE estimates the Doppler shift by using the reference signal received power (RSRP) values measured by the mobile receiver at all times. However, since there is a one-to-many mapping problem when the RSRP values reflecting the 5G beam sweeping and selection correspond to Doppler shifts, machine learning cannot be performed. To solve this problem, we design an RSRP ambiguity reducer (AR) for the machine learning input so that the pattern of RSRP values can be mapped and learned into corresponding Doppler shifts. As a result, MLDSE can estimate Doppler shift more accurately than any HST DSEs known to the authors. In addition, an MLDSE consisting of only three layers is superior to the conventional techniques in terms of computational complexity as well as estimation accuracy.
A differential evolution (DE)-based partial transmit sequence (PTS) scheme for peak-to-average power ratio (PAPR) reduction in orthogonal frequency division multiplexing (OFDM) systems has been proposed. PTS technique...
详细信息
A differential evolution (DE)-based partial transmit sequence (PTS) scheme for peak-to-average power ratio (PAPR) reduction in orthogonal frequency division multiplexing (OFDM) systems has been proposed. PTS techniques can improve the PAPR statistics of an OFDM signals, but the considerable computational complexity for the required search through a high-dimensional vector space is a potential problem for the implementation in practical systems. The DE is an efficient and powerful population-based stochastic search technique for solving optimisation problems over continuous space, which has been widely applied in many scientific and engineering fields. Thus, to reduce the complexity for searching phase weight vector and to improve the PAPR statistics, the authors introduce DE, to search the optimal phase weight factors. The simulation results show that the proposed DE-based PTS obtains an excellent PAPR performance with a low computational complexity.
暂无评论