Hock Gan

Speaker Identification Using Data-Driven Score Classification

Image Processing and Communications, Jun 1, 2016

We present a comparative evaluation of different classification algorithms for a fusion engine th... more We present a comparative evaluation of different classification algorithms for a fusion engine that is used in a speaker identity selection task. The fusion engine combines the scores from a number of classifiers, which uses the GMM-UBM approach to match speaker identity. The performances of the evaluated classification algorithms were examined in both the text-dependent and text-independent operation modes. The experimental results indicated a significant improvement in terms of speaker identification accuracy, which was approximately 7% and 14.5% for the textdependent and the text-independent scenarios, respectively. We suggest the use of fusion with a discriminative algorithm such as a Support Vector Machine in a real-world speaker identification application where the text-independent scenario predominates based on the findings.

Download

Fraud Detection in Voice-Based Identity Authentication Applications and Services

Keeping track of the multiple passwords, PINs, memorable dates and other authentication details n... more Keeping track of the multiple passwords, PINs, memorable dates and other authentication details needed to gain remote access to accounts is one of modern life's less appealing challenges. The employment of a voice-based verification as a biometric technology for both children and adults could be a good replacement to the old fashioned memory dependent procedure. Using voice for authentication could be beneficial in several application areas, including, security, protection, education, call-based and web-based services. Voice-based biometric applications are subject to different types of spoofing attacks. The most accessible and affordable type of spoofing for a voice-based biometrics system is a replay attack. Replay, which is to playback a pre-recorded speech sample, presents a genuine risk to automatic speaker verification technology. This work presents two architectures for detecting frauds caused by replay attacks in a voice-based biometrics authentication systems. Experimental results confirmed that obtained performances from both methods could further improve by applying a machine learning algorithm for performing fusion at the score level. The performance of both methods further improved by fusion using independent sources of scores in different architectures.

Download

Using Multi-Modal Bio-Digital Technologies to Support the Assessment of Cognitive Abilities of Children with Physical and Neurological Impairments

Variation of streak size distributions between PNI children and non-PNI children ... 8.5.3. Effec... more

Download

A Holistic Systems Security Approach Featuring Thin Secure Elements for Resilient IoT Deployments

Sensors, Sep 14, 2020

IoT systems differ from traditional Internet systems in that they are different in scale, footpri... more IoT systems differ from traditional Internet systems in that they are different in scale, footprint, power requirements, cost and security concerns that are often overlooked. IoT systems inherently present different fail-safe capabilities than traditional computing environments while their threat landscapes constantly evolve. Further, IoT devices have limited collective security measures in place. Therefore, there is a need for different approaches in threat assessments to incorporate the interdependencies between different IoT devices. In this paper, we run through the design cycle to provide a security-focused approach to the design of IoT systems using a use case, namely, an intelligent solar-panel project called Daedalus. We utilise STRIDE/DREAD approaches to identify vulnerabilities using a thin secure element that is an embedded, tamper proof microprocessor chip that allows the storage and processing of sensitive data. It benefits from low power demand and small footprint as a crypto processor as well as is compatible with IoT requirements. Subsequently, a key agreement based on an asymmetric cryptographic scheme, namely B-SPEKE was used to validate and authenticate the source. We find that end-to-end and independent stand-alone procedures used for validation and encryption of the source data originating from the solar panel are cost-effective in that the validation is carried out once and not several times in the chain as is often the case. The threat model proved useful not so much as a panacea for all threats but provided the framework for the consideration of known threats, and therefore appropriate mitigation plans to be deployed.

Download

Improving speaker verification performance under spoofing attacks by fusion of different operational modes

In this paper, we propose a methodology for the fusion of different modes of speaker verification... more In this paper, we propose a methodology for the fusion of different modes of speaker verification (SV) operation (fixed-passphrase, text-dependent and text-independent mode), using regression fusion models. The experimental results with and without spoofing attack conditions and using different single mode speaker verification engines, GMM-UBM, HMM-UBM and i-vector, indicated improvement in all the experiments. The 6.75 % in terms of EER is achieved as the best speaker verification performance, when using fusion of scores from three modes of operation of HMM-UBM based speaker verification systems. Relative improvement of 22.32 % achieved compare to the best performing single mode engine.

Embedded Trusted Monitoring and Management Modules for Smart Solar Panels

This paper investigates developing a prototype of smart solar panels. This architecture consists ... more This paper investigates developing a prototype of smart solar panels. This architecture consists of a panel monitoring module and the central management unit. The monitoring module is to be embedded inside each PV panel making it secure to transfer the trusted data via Wi-Fi to the central Management unit (which can accommodate an array of PV panels in an installation). This module is required for data storage and provides the ability to upload secure data to the cloud. This platform presents the ability to securely manage large numbers of rooftop solar panels in a distributed ledger by implementing block chain algorithm. For achieving this purpose, Module 400 is envisaged to be turned into a Blockchain node as it provides the infrastructure to implement this technology.

Download

Fraud Detection in Voice-Based Identity Authentication Applications and Services

2016 IEEE 16th International Conference on Data Mining Workshops (ICDMW), 2016

Keeping track of the multiple passwords, PINs, memorable dates and other authentication details n... more Keeping track of the multiple passwords, PINs, memorable dates and other authentication details needed to gain remote access to accounts is one of modern life's less appealing challenges. The employment of a voice-based verification as a biometric technology for both children and adults could be a good replacement to the old fashioned memory dependent procedure. Using voice for authentication could be beneficial in several application areas, including, security, protection, education, call-based and web-based services. Voice-based biometric applications are subject to different types of spoofing attacks. The most accessible and affordable type of spoofing for a voice-based biometrics system is a replay attack. Replay, which is to playback a pre-recorded speech sample, presents a genuine risk to automatic speaker verification technology. This work presents two architectures for detecting frauds caused by replay attacks in a voice-based biometrics authentication systems. Experimental results confirmed that obtained performances from both methods could further improve by applying a machine learning algorithm for performing fusion at the score level. The performance of both methods further improved by fusion using independent sources of scores in different architectures.

Download

Speaker Identification Using Data-Driven Score Classification

Image Processing & Communications, 2016

We present a comparative evaluation of different classification algorithms for a fusion engine th... more We present a comparative evaluation of different classification algorithms for a fusion engine that is used in a speaker identity selection task. The fusion engine combines the scores from a number of classifiers, which uses the GMM-UBM approach to match speaker identity. The performances of the evaluated classification algorithms were examined in both the text-dependent and text-independent operation modes. The experimental results indicated a significant improvement in terms of speaker identification accuracy, which was approximately 7% and 14.5% for the text-dependent and the text-independent scenarios, respectively. We suggest the use of fusion with a discriminative algorithm such as a Support Vector Machine in a real-world speaker identification application where the text-independent scenario predominates based on the findings.

Download

Development of the maximum-streak measure for evaluating the suitability of non-handheld devices in cognitive tests of Physically and Neurologically Impaired(PNI) children

This study examines an objective measure for discriminating between different degrees of performa... more This study examines an objective measure for discriminating between different degrees of performance in a child’s use of multiple non-hand-held devices when undertaking a simple competence test: the maximum streak of successfully, making a binary (YES / NO) response in the test. The paper reports encouraging results for the maximum consecutive success (streak) measure in its ability to inform on device suitability for a particular child. Our results show that different devices influence success streak, while performance of children vary when using different devices, based on severity of impairment. While this is in line with our main hypothesis, that a suitable match between device and impairment is necessary, our results provide a first candidate measure that can be potentially useful to personalizing a device to a particular user and his/her capabilities. Keywords— disabled children, non-hand held device, motorskill test, cognitive test, streak

Bio-digital device impact on a constant load cognitive test of children with physical and neurological impairments

In this paper we extend on our earlier work utilizing maximum consecutive success as a measure us... more In this paper we extend on our earlier work utilizing maximum consecutive success as a measure used in the context of testing motor skills, for children with physical and neurological impairment. Here we verify a change in the "ceiling" of performance that takes place as the cognitive load is increased. In the results we see two groups establish themselves in terms of the ones capable of achieving the test objectives and the ones that engage but are not successful. In this study, we compare results for two different tests, one of motor skills and the second of categorization assessment and find agreement between the two using our approach. Working on our main hypothesis that it is possible to provide better fit of devices to PNI children, the results provide initial suggestions that cognitive test success depend on a combination of motor skill and cognitive ability which is not completely separable. Ranking of bio-digital devices ideally need to be done using the target ap...

Improving speaker verification performance under spoofing attacks by fusion of different operational modes

2017 IEEE 13th International Colloquium on Signal Processing & its Applications (CSPA)

—In this paper, we propose a methodology for the fusion of different modes of speaker verificatio... more —In this paper, we propose a methodology for the fusion of different modes of speaker verification (SV) operation (fixed-passphrase, text-dependent and text-independent mode), using regression fusion models. The experimental results with and without spoofing attack conditions and using different single mode speaker verification engines, GMM-UBM, HMM-UBM and i-vector, indicated improvement in all the experiments. The 6.75 % in terms of EER is achieved as the best speaker verification performance, when using fusion of scores from three modes of operation of HMM-UBM based speaker verification systems. Relative improvement of 22.32 % achieved compare to the best performing single mode engine.

Increasing accessibility of a cognitive test for children with motor and speech disability

Bio-digital device impact on a constant load cognitive test of children with physical and neurological impairments

Development of the maximum-streak measure for evaluating the suitability of non-handheld devices in cognitive tests of Physically and Neurologically Impaired (PNI) children

A Holistic Systems Security Approach Featuring Thin Secure Elements for Resilient IoT Deployments

Sensors

IoT systems differ from traditional Internet systems in that they are different in scale, footpri... more IoT systems differ from traditional Internet systems in that they are different in scale, footprint, power requirements, cost and security concerns that are often overlooked. IoT systems inherently present different fail-safe capabilities than traditional computing environments while their threat landscapes constantly evolve. Further, IoT devices have limited collective security measures in place. Therefore, there is a need for different approaches in threat assessments to incorporate the interdependencies between different IoT devices. In this paper, we run through the design cycle to provide a security-focused approach to the design of IoT systems using a use case, namely, an intelligent solar-panel project called Daedalus. We utilise STRIDE/DREAD approaches to identify vulnerabilities using a thin secure element that is an embedded, tamper proof microprocessor chip that allows the storage and processing of sensitive data. It benefits from low power demand and small footprint as a...

Download

Disaster recovery for very large GSM/UMTS HLR databases

HLR data migration

HLR mated-pair auto cutover

by Hock Gan and Alan Darbyshire

Cellular Communications System with Screening of Unauthorized Services

Using Multi-Modal Bio-Digital Technologies to Support the Assessment of Cognitive Abilities of Children with Physical and Neurological Impairments

Uploads

Papers by Hock Gan

Log In