Sub-vector based biometric speaker verification using MLLR super-vector

In this paper, we propose a sub-vector based speaker characterization method for biometric speaker verification, where speakers are represented by uniform segmentation of their maximum likelihood linear regression (MLLR) super-vectors called m-vectors. The MLLR transformation is estimated with respect to universal background model (UBM) without any speech/phonetic information. We introduce two strategies for segmentation of MLLR super-vector: one is called disjoint and other is an overlapped window technique. During test phase, m-vectors of the test utterance are scored against the claimant speaker. Before scoring, m-vectors are post-processed to compensate the session variability. In addition, we propose a clustering algorithm for multiple-class wise MLLR transformation, where Gaussian components of the UBM are clustered into different groups using the concept of expectation maximization (EM) and maximum likelihood (ML). In this case, MLLR transformations are estimated with respect to each class using the sufficient statistics accumulated from the Gaussian components belonging to the particular class, which are then used for m-vector system. The proposed method needs only once alignment of the data with respect to the UBM for multiple MLLR transformations. We first show that the proposed multi-class m-vector system shows promising speaker verification performance when compared to the conventional i-vector based speaker verification system. Secondly, the proposed EM based clustering technique is robust to the random initialization in-contrast to the conventional K-means algorithm and yields system performance better/equal which is best obtained by the K-means. Finally, we show that the fusion of the m-vector with the i-vector further improves the performance of the speaker verification in both score as well as feature domain. The experimental results are shown on various tasks of NIST 2008 speaker recognition evaluation (SRE) core condition.     

A study on the roles of total variability space and session variability modeling in speaker recognition

Speaker verification (SV) using i-vector concept becomes state-of-the-art. In this technique, speakers are projected onto the total variability space and represented by vectors called i-vectors. During testing, the i-vectors of the test speech segment and claimant are conditioned to compensate for the session variability before scoring. So, i-vector system can be viewed as two processing blocks: one is total variability space and the other is post-processing module. Several questions arise, such as, (i) which part of the i-vector system plays a major role in speaker verification: total variability space or post-processing task; (ii) is the post-processing module intrinsic to the total variability space? The motivation of this paper is to partially answer these questions by proposing several simpler speaker characterization systems for speaker verification, where speakers are represented by their speaker characterization vectors (SCVs). The SCVs are obtained by uniform segmentation of the speakers gaussian mixture models (GMMs)- and maximum likelihood linear regression (MLLR) super-vectors. We consider two adaptation approaches for GMM super-vector: one is maximum a posteriori and other is MLLR. Similarly to the i-vector, SCVs are post-processed for session variability compensation during testing. The proposed system shows promising performance when compared to the classical i-vector system which indicates that the post-processing task plays an major role in i-vector based SV system and is not intrinsic to the total variability space. All experimental results are shown on NIST 2008 SRE core condition.     

I-vector based speaker recognition using advanced channel compensation techniques

This paper investigates advanced channel compensation techniques for the purpose of improving i-vector speaker verification performance in the presence of high intersession variability using the NIST 2008 and 2010 SRE corpora. The performance of four channel compensation techniques: (a) weighted maximum margin criterion (WMMC), (b) source-normalized WMMC (SN-WMMC), (c) weighted linear discriminant analysis (WLDA) and (d) source-normalized WLDA (SN-WLDA) have been investigated. We show that, by extracting the discriminatory information between pairs of speakers as well as capturing the source variation information in the development i-vector space, the SN-WLDA based cosine similarity scoring (CSS) i-vector system is shown to provide over 20% improvement in EER for NIST 2008 interview and microphone verification and over 10% improvement in EER for NIST 2008 telephone verification, when compared to SN-LDA based CSS i-vector system. Further, score-level fusion techniques are analyzed to combine the best channel compensation approaches, to provide over 8% improvement in DCF over the best single approach, SN-WLDA, for NIST 2008 interview/telephone enrolment-verification condition. Finally, we demonstrate that the improvements found in the context of CSS also generalize to state-of-the-art GPLDA with up to 14% relative improvement in EER for NIST SRE 2010 interview and microphone verification and over 7% relative improvement in EER for NIST SRE 2010 telephone verification.     

Improving the performance of GPLDA speaker verification using unsupervised inter-dataset variability compensation approaches

In practical applications, speaker verification systems have to be developed and trained using data which is outside the domain of the intended application as the collection of significant amount of in-domain data could be difficult. Experimental studies have found that when a GPLDA system is trained using out-domain data, it significantly affects the speaker verification performance due to the mismatch between development data and evaluation data. This paper proposes several unsupervised inter-dataset variability compensation approaches for the purpose of improving the performance of GPLDA systems trained using out-domain data. We show that when GPLDA is trained using out-domain data, we can improve the performance by as much as 39% by using by score normalisation using small amounts of in-domain data. Also in situations where rich out-domain data and only limited in-domain data are available, a pooled-linear-weighted technique to estimate the GPLDA parameters shows 35% relative improvements in equal error rate (EER) on int–int conditions. We also propose a novel inter-dataset covariance normalization (IDCN) approach to overcome in- and out-domain data mismatch problem. Our unsupervised IDCN-compensated GPLDA system shows 14 and 25% improvement respectively in EER over out-domain GPLDA speaker verification on tel–tel and int–int training–testing conditions. We provide intuitive explanations as to why these inter-dataset variability compensation approaches provide improvements to speaker verification accuracy.     

Know your customer: computing <Emphasis Type="Italic">k</Emphasis>-most promising products for targeted marketing

The advancement of World Wide Web has revolutionized the way the manufacturers can do business. The manufacturers can collect customer preferences for products and product features from their sales and other product-related Web sites to enter and sustain in the global market. For example, the manufactures can make intelligent use of these customer preference data to decide on which products should be selected for targeted marketing. However, the selected products must attract as many customers as possible to increase the possibility of selling more than their respective competitors. This paper addresses this kind of product selection problem. That is, given a database of existing products P from the competitors, a set of company’s own products Q, a dataset C of customer preferences and a positive integer k, we want to find k-most promising products (k-MPP) from Q with maximum expected number of total customers for targeted marketing. We model k-MPP query and propose an algorithmic framework for processing such query and its variants. Our framework utilizes grid-based data partitioning scheme and parallel computing techniques to realize k-MPP query. The effectiveness and efficiency of the framework are demonstrated by conducting extensive experiments with real and synthetic datasets.     

Approximating the Sparsest k-Subgraph in Chordal Graphs

Given a simple undirected graph G = (V, E) and an integer k < |V|, the Sparsest k-Subgraph problem asks for a set of k vertices which induces the minimum number of edges. As a generalization of the classical independent set problem, Sparsest k-Subgraph is ????-hard and even not approximable unless ?????? in general graphs. Thus, we investigate Sparsest k-Subgraph in graph classes where independent set is polynomial-time solvable, such as subclasses of perfect graphs. Our two main results are the ????-hardness of Sparsest k-Subgraph on chordal graphs, and a greedy 2-approximation algorithm. Finally, we also show how to derive a P T A S for Sparsest k-Subgraph on proper interval graphs.     

TAPER: query-aware,partition-enhancement for large,heterogenous graphs

Graph partitioning has long been seen as a viable approach to addressing Graph DBMS scalability. A partitioning, however, may introduce extra query processing latency unless it is sensitive to a specific query workload, and optimised to minimise inter-partition traversals for that workload. Additionally, it should also be possible to incrementally adjust the partitioning in reaction to changes in the graph topology, the query workload, or both. Because of their complexity, current partitioning algorithms fall short of one or both of these requirements, as they are designed for offline use and as one-off operations. The TAPER system aims to address both requirements, whilst leveraging existing partitioning algorithms. TAPER takes any given initial partitioning as a starting point, and iteratively adjusts it by swapping chosen vertices across partitions, heuristically reducing the probability of inter-partition traversals for a given path queries workload. Iterations are inexpensive thanks to time and space optimisations in the underlying support data structures. We evaluate TAPER on two different large test graphs and over realistic query workloads. Our results indicate that, given a hash-based partitioning, TAPER reduces the number of inter-partition traversals by \(\sim \)80%; given an unweighted Metis partitioning, by \(\sim \)30%. These reductions are achieved within eight iterations and with the additional advantage of being workload-aware and usable online.     

Verifiable threshold quantum secret sharing with sequential communication

A (t, n) threshold quantum secret sharing (QSS) is proposed based on a single d-level quantum system. It enables the (t, n) threshold structure based on Shamir’s secret sharing and simply requires sequential communication in d-level quantum system to recover secret. Besides, the scheme provides a verification mechanism which employs an additional qudit to detect cheats and eavesdropping during secret reconstruction and allows a participant to use the share repeatedly. Analyses show that the proposed scheme is resistant to typical attacks. Moreover, the scheme is scalable in participant number and easier to realize compared to related schemes. More generally, our scheme also presents a generic method to construct new (t, n) threshold QSS schemes based on d-level quantum system from other classical threshold secret sharing.     

Shadow Prices in Territory Division

We consider a geographic optimization problem in which we are given a region R, a probability density function f(?) defined on R, and a collection of n utility density functions u _i (?) defined on R. Our objective is to divide R into n sub-regions R _i so as to “balance” the overall utilities on the regions, which are given by the integrals \(\iint _{R_{i}}f(x)u_{i}(x)\, dA\). Using a simple complementary slackness argument, we show that (depending on what we mean precisely by “balancing” the utility functions) the boundary curves between optimal sub-regions are level curves of either the difference function u _i (x) ? u _j (x) or the ratio u _i (x)/u _j (x). This allows us to solve the problem of optimally partitioning the region efficiently by reducing it to a low-dimensional convex optimization problem. This result generalizes, and gives very short and constructive proofs of, several existing results in the literature on equitable partitioning for particular forms of f(?) and u _i (?). We next give two economic applications of our results in which we show how to compute a market-clearing price vector in an aggregate demand system or a variation of the classical Fisher exchange market. Finally, we consider a dynamic problem in which the density function f(?) varies over time (simulating population migration or transport of a resource, for example) and derive a set of partial differential equations that describe the evolution of the optimal sub-regions over time. Numerical simulations for both static and dynamic problems confirm that such partitioning problems become tractable when using our methods.     

Speaker Verification Using Adapted Gaussian Mixture Models

Reynolds, Douglas A., Quatieri, Thomas F., and Dunn, Robert B., Speaker Verification Using Adapted Gaussian Mixture Models, Digital Signal Processing10(2000), 19–41.In this paper we describe the major elements of MIT Lincoln Laboratory's Gaussian mixture model (GMM)-based speaker verification system used successfully in several NIST Speaker Recognition Evaluations (SREs). The system is built around the likelihood ratio test for verification, using simple but effective GMMs for likelihood functions, a universal background model (UBM) for alternative speaker representation, and a form of Bayesian adaptation to derive speaker models from the UBM. The development and use of a handset detector and score normalization to greatly improve verification performance is also described and discussed. Finally, representative performance benchmarks and system behavior experiments on NIST SRE corpora are presented.     

Multiple background models for speaker verification using the concept of vocal tract length and MLLR super-vector

In this paper, we investigate the use of Multiple Background Models (M-BMs) in Speaker Verification (SV). We cluster the speakers using either their Vocal Tract Lengths (VTLs) or by using their speaker specific Maximum Likelihood Linear Regression (MLLR) super-vector, and build a separate Background Model (BM) for each such cluster. We show that the use of M-BMs provide improved performance when compared to the use of a single/gender wise Universal Background Model (UBM). While the computational complexity during test remains same for both M-BMs and UBM, M-BMs require switching of models depending on the claimant and also score-normalization becomes difficult. To overcome these problems, we propose a novel method which aggregates the information from Multiple Background Models into a single gender independent UBM and is inspired by conventional Feature Mapping (FM) technique. We show that using this approach, we get improvement over the conventional UBM method, and yet this approach also permits easy use of score-normalization techniques. The proposed method provides relative improvement in Equal-Error Rate (EER) by 13.65?% in the case of VTL clustering, and 15.43?% in the case of MLLR super-vector when compared to the conventional single UBM system. When AT-norm score-normalization is used then the proposed method provided a relative improvement in EER of 20.96?% for VTL clustering and 22.48?% for MLLR super-vector based clustering. Furthermore, the proposed method is compared with the gender dependent speaker verification system using Gaussian Mixture Model-Support Vector Machines (GMM-SVM) super-vector linear kernel. The experimental results show that the proposed method perform better than gender dependent speaker verification system.     

Degree-Constrained Graph Orientation: Maximum Satisfaction and Minimum Violation

A degree-constrained graph orientation of an undirected graph G is an assignment of a direction to each edge in G such that the outdegree of every vertex in the resulting directed graph satisfies a specified lower and/or upper bound. Such graph orientations have been studied for a long time and various characterizations of their existence are known. In this paper, we consider four related optimization problems introduced in reference (Asahiro et al. LNCS 7422, 332–343 (2012)): For any fixed non-negative integer W, the problems MAX W-LIGHT, MIN W-LIGHT, MAX W-HEAVY, and MIN W-HEAVY take as input an undirected graph G and ask for an orientation of G that maximizes or minimizes the number of vertices with outdegree at most W or at least W. As shown in Asahiro et al. LNCS 7422, 332–343 (2012)).     

The formula of the solution for some classes of initial boundary value problems for the hyperbolic equation with two independent variables

A new representation is proved of the solutions of initial boundary value problems for the equation of the form u _xx (x, t) + r(x)u _x (x, t) ? q(x)u(x, t) = u _tt (x, t) + μ(x)u _t (x, t) in the section (under boundary conditions of the 1st, 2nd, or 3rd type in any combination). This representation has the form of the Riemann integral dependent on the x and t over the given section.     

Efficient attribute-based signature and signcryption realizing expressive access structures

This paper addresses the open problem of designing attribute-based signature (ABS) schemes with constant number of bilinear pairing operations for signature verification or short signatures for more general policies posed by Gagné et al. in Pairing 2012. Designing constant-size ABS for expressive access structures is a challenging task. We design two key-policy ABS schemes with constant-size signature for expressive linear secret-sharing scheme (LSSS)-realizable monotone access structures. Both the schemes utilize only 3 pairing operations in signature verification process. The first scheme is small universe construction, while the second scheme supports large universes of attributes. The signing key is computed according to LSSS-realizable access structure over signer’s attributes, and the message is signed with an attribute set satisfying the access structure. Our ABS schemes provide the existential unforgeability in selective attribute set security model and preserve signer privacy. We also propose a new attribute-based signcryption (ABSC) scheme for LSSS-realizable access structures utilizing only 6 pairings and making the ciphertext size constant. Our scheme is significantly more efficient than existing ABSC schemes. While the secret key (signing key or decryption key) size increases by a factor of number of attributes used in the system, the number of pairing evaluations is reduced to constant. Our protocol achieves (a) ciphertext indistinguishability under adaptive chosen ciphertext attacks assuming the hardness of decisional Bilinear Diffie–Hellman Exponent problem and (b) existential unforgeability under adaptive chosen message attack assuming the hardness of computational Diffie–Hellman Exponent problem. The security proofs are in selective attribute set security model without using any random oracle heuristic. In addition, our ABSC achieves public verifiability of the ciphertext, enabling any party to verify the integrity and validity of the ciphertext.     

Improved i-vector extraction technique for speaker verification with short utterances

A major challenge in ASV is to improve performance with short speech segments for end-user convenience in real-world applications. In this paper, we present a detailed analysis of ASV systems to observe the duration variability effects on state-of-the-art i-vector and classical Gaussian mixture model-universal background model (GMM-UBM) based ASV systems. We observe an increase in uncertainty of model parameter estimation for i-vector based ASV with speech of shorter duration. In order to compensate the effect of duration variability in short utterances, we have proposed adaptation technique for Baum-Welch statistics estimation used to i-vector extraction. Information from pre-estimated background model parameters are used for adaptation method. The ASV performance with the proposed approach is considerably superior to the conventional i-vector based system. Furthermore, the fusion of proposed i-vector based system and GMM-UBM further improves the ASV performance, especially for short speech segments. Experiments conducted on two speech corpora, NIST SRE 2008 and 2010, have shown relative improvement in equal error rate (EER) in the range of 12–20%.     

Do Accelerating Turing Machines Compute the Uncomputable?

Accelerating Turing machines have attracted much attention in the last decade or so. They have been described as “the work-horse of hypercomputation” (Potgieter and Rosinger 2010: 853). But do they really compute beyond the “Turing limit”—e.g., compute the halting function? We argue that the answer depends on what you mean by an accelerating Turing machine, on what you mean by computation, and even on what you mean by a Turing machine. We show first that in the current literature the term “accelerating Turing machine” is used to refer to two very different species of accelerating machine, which we call end-stage-in and end-stage-out machines, respectively. We argue that end-stage-in accelerating machines are not Turing machines at all. We then present two differing conceptions of computation, the internal and the external, and introduce the notion of an epistemic embedding of a computation. We argue that no accelerating Turing machine computes the halting function in the internal sense. Finally, we distinguish between two very different conceptions of the Turing machine, the purist conception and the realist conception; and we argue that Turing himself was no subscriber to the purist conception. We conclude that under the realist conception, but not under the purist conception, an accelerating Turing machine is able to compute the halting function in the external sense. We adopt a relatively informal approach throughout, since we take the key issues to be philosophical rather than mathematical.     

On s-uniform property of compressing sequences derived from primitive sequences modulo odd prime powers

Let Z/(p^e) be the integer residue ring modulo p^e with p an odd prime and e ≥ 2. We consider the suniform property of compressing sequences derived from primitive sequences over Z/(p^e). We give necessary and sufficient conditions for two compressing sequences to be s-uniform with α provided that the compressing map is of the form ?(x₀, x₁,...,x_e?1) = g(x_e?1) + η(x₀, x₁,..., x_e?2), where g(x_e?1) is a permutation polynomial over Z/(p) and η is an (e ? 1)-variable polynomial over Z/(p).     

Almost cover-free codes

We say that an s-subset of codewords of a code X is (s, l)-bad if X contains l other codewords such that the conjunction of these l words is covered by the disjunction of the words of the s-subset. Otherwise, an s-subset of codewords of X is said to be (s, l)-bad. A binary code X is called a disjunctive (s, l) cover-free (CF) code if X does not contain (s, l)-bad subsets. We consider a probabilistic generalization of (s, l) CF codes: we say that a binary code is an (s, l) almost cover-free (ACF) code if almost all s-subsets of its codewords are (s, l)-good. The most interesting result is the proof of a lower and an upper bound for the capacity of (s, l) ACF codes; the ratio of these bounds tends as s→∞ to the limit value log₂ e/(le).     

On resol vability of Steine r systems S( v = 2m, 4, 3) of rank r ≤ v − m + 1 o ve r $$\mathbb{F}_2 $$

Two new constructions of Steiner quadruple systems S(v, 4, 3) are given. Both preserve resolvability of the original Steiner system and make it possible to control the rank of the resulting system. It is proved that any Steiner system S(v = 2 ^m , 4, 3) of rank r ≤ v ? m + 1 over F₂ is resolvable and that all systems of this rank can be constructed in this way. Thus, we find the number of all different Steiner systems of rank r = v ? m + 1.     

Upper and lower estimates for the number of some <Emphasis Type="Italic">k</Emphasis>-dimensional subspaces of a given weight over a finite field

A fast simulation method is proposed that makes it possible to construct upper and lower estimates for the number of k-measurable subspaces (of arbitrary weight ω) of an n-measurable vector space over a Galois field containing q elements. Numerical examples demonstrate the accuracy of the estimates.