Instance Based Sparse Classifier Fusion for Speaker Verification
Journal Title: Journal of Information Systems and Telecommunication - Year 2016, Vol 4, Issue 3
Abstract
This paper focuses on the problem of ensemble classification for text-independent speaker verification. Ensemble classification is an efficient method to improve the performance of the classification system. This method gains the advantage of a set of expert classifiers. A speaker verification system gets an input utterance and an identity claim, then verifies the claim in terms of a matching score. This score determines the resemblance of the input utterance and pre-enrolled target speakers. Since there is a variety of information in a speech signal, state-of-the-art speaker verification systems use a set of complementary classifiers to provide a reliable decision about the verification. Such a system receives some scores as input and takes a binary decision: accept or reject the claimed identity. Most of the recent studies on the classifier fusion for speaker verification used a weighted linear combination of the base classifiers. The corresponding weights are estimated using logistic regression. Additional researches have been performed on ensemble classification by adding different regularization terms to the logistic regression formulae. However, there are missing points in this type of ensemble classification, which are the correlation of the base classifiers and the superiority of some base classifiers for each test instance. We address both problems, by an instance based classifier ensemble selection and weight determination method. Our extensive studies on NIST 2004 speaker recognition evaluation (SRE) corpus in terms of EER, minDCF and minCLLR show the effectiveness of the proposed method.
Authors and Affiliations
Mohammad Hasheminejad, Hassan Farsi
Keywords
Related Articles
- EP ID EP184067
- DOI 10.7508/jist.2016.03.007
- Views 98
- Downloads 0
Optimal Sensor Scheduling Algorithms for Distributed Sensor Networks
In this paper, a sensor network is used to estimate the dynamic states of a system. At each time step, one (or multiple) sensors are available that can send its measured data to a central node, in which all of processing...
A Novel Ultra-Broad Band, High Gain, and Low Noise Distributed Amplifier Using Modified Regulated Cascode Configuration (MRGC) Gain-Cell
In this paper, an ultra-broad bandwidth, low noise, and high gain-flatness CMOS distributed amplifier (CMOS-DA) based on a novel gain-cell is presented. The new gain-cell that enhances the output impedance as a result th...
Towards Accelerating IP Lookups on Commodity PC Routers using Bloom Filter: Proposal of Bloom-Bird
Nowadays, routers are the main backbone of computer networks specifically the Internet. Moreover, the need for high-performance and high-speed routers has become a fundamental issue due to significant growth of informati...
A New Switched-beam Setup for Adaptive Antenna Array Beamforming
In this paper, a new spatio-temporal based approach is proposed which improves the speed and performance of temporal-based algorithms, conventional Least Mean Square (LMS), Normalized LMS (NLMS) and Variable Step-size LM...
A New Finite Field Multiplication Algorithm to Improve Elliptic Curve Cryptosystem Implementations
This paper presents a new and efficient implementation approach for the elliptic curve cryptosystem (ECC) based on a novel finite field multiplication in GF(2m) and an efficient scalar multiplication algorithm. This new...