arXiv subjects

David Dean

Publications and source records attributed to David Dean.

11 recordsLinked to original sources

arXiv2021

Creep-fatigue interactions in a polycrystalline structural material under typical high-temperature power plant operating conditions

A micromechanical model at the microscale within a crystal plasticity self-consistent model, is used to analyse loading histories in Type 316H stainless steel, common to structural components in high-temperature power plants. The study compares the SCM predictions on changes in mechanical behaviour and creep properties to analyses via the UK R5 structural assessment code. Plant-relevant cyclic-creep histories in Type 316H stainless steel at 550C are examined with focus on the estimation of accumulated creep strain. This aims to quantify how better understanding of creep deformation under cyclic loading can inform R5 code life assessment methods, which are strain based. The effect of cyclic plasticity on primary and secondary creep is quantified. The levels of creep strain accumulated during different dwells, following loading from different macroscopic stress states, is also evaluated and link is made to experimental observations in explaining the response. The impact of displacement- and load-controlled dwells on the evolution of the cyclic hysteresis loop is examined which revealed how different assumptions for creep strain accumulation could employ information from SCM results to reduce uncertainty. It also provides guidance in the development of interpolation frameworks for structural parameters, based on the SCM results which could be used directly in structural assessment and design approaches to estimate the accumulation of inelastic strain with a lower degree of conservatism.

cond-mat.mtrl-sci↗

arXiv2020

Meta Transfer Learning for Emotion Recognition

Deep learning has been widely adopted in automatic emotion recognition and has lead to significant progress in the field. However, due to insufficient annotated emotion datasets, pre-trained models are limited in their generalization capability and thus lead to poor performance on novel test sets. To mitigate this challenge, transfer learning performing fine-tuning on pre-trained models has been applied. However, the fine-tuned knowledge may overwrite and/or discard important knowledge learned from pre-trained models. In this paper, we address this issue by proposing a PathNet-based transfer learning method that is able to transfer emotional knowledge learned from one visual/audio emotion domain to another visual/audio emotion domain, and transfer the emotional knowledge learned from multiple audio emotion domains into one another to improve overall emotion recognition accuracy. To show the robustness of our proposed system, various sets of experiments for facial expression recognition and speech emotion recognition task on three emotion datasets: SAVEE, EMODB, and eNTERFACE have been carried out. The experimental results indicate that our proposed system is capable of improving the performance of emotion recognition, making its performance substantially superior to the recent proposed fine-tuning/pre-trained models based transfer learning methods.

cs.CV↗

arXiv2018

Meta Transfer Learning for Facial Emotion Recognition

The use of deep learning techniques for automatic facial expression recognition has recently attracted great interest but developed models are still unable to generalize well due to the lack of large emotion datasets for deep learning. To overcome this problem, in this paper, we propose utilizing a novel transfer learning approach relying on PathNet and investigate how knowledge can be accumulated within a given dataset and how the knowledge captured from one emotion dataset can be transferred into another in order to improve the overall performance. To evaluate the robustness of our system, we have conducted various sets of experiments on two emotion datasets: SAVEE and eNTERFACE. The experimental results demonstrate that our proposed system leads to improvement in performance of emotion recognition and performs significantly better than the recent state-of-the-art schemes adopting fine-\ tuning/pre-trained approaches.

cs.CV↗

arXiv2016

Anharmonic particles: suppressing van der Waals forces by an external field

We study the classical thermal component of Casimir, or van der Waals, forces between point particles with highly anharmonic dipole Hamiltonians when they are subjected to an external electric field. Using a model for which the individual dipole moments saturate in a strong field (a model that mimics the charges in a neutral, perfectly conducting sphere), we find that the resulting Casimir force depends strongly on the strength of the field, as demonstrated by analytical results. For a certain angle between external field and center to center axis, the fluctuation force can be tuned and suppressed to arbitrarily small values. We compare the forces between these anharmonic particles to those between harmonic ones, and also provide a simple formula for asymptotically large external fields, which we expect to be generally valid for the case of saturating dipole moments.

cond-mat.mes-hall↗

arXiv2016

Improving Short Utterance PLDA Speaker Verification using SUV Modelling and Utterance Partitioning Approach

This paper analyses the short utterance probabilistic linear discriminant analysis (PLDA) speaker verification with utterance partitioning and short utterance variance (SUV) modelling approaches. Experimental studies have found that instead of using single long-utterance as enrolment data, if long enrolled utterance is partitioned into multiple short utterances and average of short utterance i-vectors is used as enrolled data, that improves the Gaussian PLDA (GPLDA) speaker verification. This is because short utterance i-vectors have speaker, session and utterance variations, and utterance-partitioning approach compensates the utterance variation. Subsequently, SUV-PLDA is also studied with utterance partitioning approach, and utterance partitioning-based SUV-GPLDA system shows relative improvement of 9% and 16% in EER for NIST 2008 and NIST 2010 truncated 10sec-10sec evaluation condition as utterance partitioning approach compensates the utterance variation and SUV modelling approach compensates the mismatch between full-length development data and short-length evaluation data.

cs.SD↗

arXiv2016

DNN based Speaker Recognition on Short Utterances

This paper investigates the effects of limited speech data in the context of speaker verification using deep neural network (DNN) approach. Being able to reduce the length of required speech data is important to the development of speaker verification system in real world applications. The experimental studies have found that DNN-senone-based Gaussian probabilistic linear discriminant analysis (GPLDA) system respectively achieves above 50% and 18% improvements in EER values over GMM-UBM GPLDA system on NIST 2010 coreext-coreext and truncated 15sec-15sec evaluation conditions. Further when GPLDA model is trained on short-length utterances (30sec) rather than full-length utterances (2min), DNN-senone GPLDA system achieves above 7% improvement in EER values on truncated 15sec-15sec condition. This is because short length development i-vectors have speaker, session and phonetic variation and GPLDA is able to robustly model those variations. For several real world applications, longer utterances (2min) can be used for enrollment and shorter utterances (15sec) are required for verification, and in those conditions, DNN-senone GPLDA system achieves above 26% improvement in EER values over GMM-UBM GPLDA systems.

cs.SD↗

arXiv2016

Domain adaptation based Speaker Recognition on Short Utterances

This paper explores how the in- and out-domain probabilistic linear discriminant analysis (PLDA) speaker verification behave when enrolment and verification lengths are reduced. Experiment studies have found that when full-length utterance is used for evaluation, in-domain PLDA approach shows more than 28% improvement in EER and DCF values over out-domain PLDA approach and when short utterances are used for evaluation, the performance gain of in-domain speaker verification reduces at an increasing rate. Novel modified inter dataset variability (IDV) compensation is used to compensate the mismatch between in- and out-domain data and IDV-compensated out-domain PLDA shows respectively 26% and 14% improvement over out-domain PLDA speaker verification when SWB and NIST data are respectively used for S normalization. When the evaluation utterance length is reduced, the performance gain by IDV also reduces as short utterance evaluation data i-vectors have more variations due to phonetic variations when compared to the dataset mismatch between in- and out-domain data.

cs.SD↗

arXiv2004

Neutrinoless double beta decay and direct searches for neutrino mass

Study of the neutrinoless double beta decay and searches for the manifestation of the neutrino mass in ordinary beta decay are the main sources of information about the absolute neutrino mass scale, and the only practical source of information about the charge conjugation properties of the neutrinos. Thus, these studies have a unique role in the plans for better understanding of the whole fast expanding field of neutrino physics.

hep-ph↗

arXiv2002

Asymptotic Behavior of the Wave Packet Propagation through a Barrier: the Green's Function Approach Revisited

To model the decay of a quasibound state we use the modified two-potential approach introduced by Gurvitz and Kalbermann. This method has proved itself useful in the past for calculating the decay width and the energy shift of an isolated quasistationary state. We follow the same approach in order to propagate the wave-packet in time with the ultimate goal of extracting the momentum-distribution of emitted particles. The advantage of the method is that it provides the time-dependent wave function in a simple semi-analytic form. We intend to apply this method to the modeling of metastable states for which no direct integration of the time-dependent Schroedinger equation is available today.

nucl-th↗

arXiv2002

Dynamic spin-glass behavior in a disorder-free, two-component model of quantum frustrated magnets

Motivated by the observation of a spin-glass transition in almost disorder-free Kagome antiferromagnets, and by the specific form of the effective low-energy model of the S=1/2, trimerized Kagome antiferromagnet, we investigate the possibility to obtain a spin-glass behavior in two-component, disorder-free models. We concentrate on a toy-model, a modified Ashkin-Teller model in a magnetic field that couples only to one species of spins, for which we prove that a dynamic spin-glass behavior occurs. The dynamics of the magnetization is closely related to that of the underlying Ising model in zero field in which spins and pseudo-spins are intimately coupled. The spin-glass like history dependence of the magnetization is a consequence of the ageing of the underlying Ising model.

cond-mat.str-el↗

arXiv1996

Site Disordered Spin Systems in the Gaussian Variational Approximation

We define a replica field theory describing finite dimensional site disordered spin systems by introducing the notion of grand canonical disorder, where the number of spins in the system is random but quenched. A general analysis of this field theory is made using the Gaussian variational or Hartree Fock method, and illustrated with several specific examples. Irrespective of the form of interaction between the spins this approximation predicts a spin glass phase. We discuss the replica symmetric phase at length, explicitly identifying the correlator that diverges at the spin glass transition. We also discuss the form of continuous replica symmetry breaking found just below the transition. Finally we show how an analysis of ferromagnetic ordering indicates a breakdown of the approximation.

cond-mat.dis-nn↗