A New Unsupervised Short-Utterance based Speaker Identification Approach with Parametric t-SNE Dimensionality Reduction

Elnaggar, Omar; Arelhi, Roselina

doi:10.1109/icaiic.2019.8669051

Cited by 7 publications

(4 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such 3D visualisation of joint kinematic timeseries facilitated the design of the informative prior which was found to enhance the Bayesian inference, and could potentially be useful for medical screening and/or diagnosis. Dimension reduction has been successfully applied to visualise data across different domains, from wearable sensing 11 through speech processing 36 to knowledge exchange 37 . The following text elaborates on the two DR methods employed in this paper.…”

Section: Joint Kinematics Preprocessingmentioning

confidence: 99%

“…All 1,350 axes of rotation were written to a comma-separated values (CSV) file for subsequent import into MATLAB © (The MathWorks, Massachusetts, US). Therein, an orientation dataset generator script concatenated each synthetic axis of rotation with each angle of rotation from the closed set { 36 36 π, 35 36 π, 34 36 π, ..., 1 36 π}. The output synthetic axis-angle dataset contained a total of 48,600 orientations.…”

Section: Data Availabilitymentioning

confidence: 99%

See 1 more Smart Citation

KIDS: kinematics-based (in)activity detection and segmentation in a sleep case study

Elnaggar¹,

Arelhi²,

Coenen³

et al. 2023

Preprint

View full text Add to dashboard Cite

Sleep behaviour and in-bed movements contain rich information on the neurophysiological health of people, and have a direct link to the general well-being and quality of life. Standard clinical practices rely on polysomnography for sleep assessment; however, it is intrusive, performed in unfamiliar environments and requires trained personnel. Progress has been made on less invasive sensor technologies, such as actigraphy, but clinical validation raises concerns over their reliability and precision. Additionally, the field lacks a widely acceptable algorithm, with proposed approaches ranging from raw signal or feature thresholding to data-hungry classification models, many of which are unfamiliar to medical staff. This paper proposes an online Bayesian probabilistic framework for objective (in)activity detection and segmentation based on clinically meaningful joint kinematics, measured by a custom-made wearable sensor. Intuitive three-dimensional visualisations of kinematic timeseries were accomplished through dimension reduction based preprocessing, offering out-of-the-box framework explainability potentially useful for clinical monitoring and diagnosis. The proposed framework attained up to 99.2% F 1 -score and 0.96 Pearson's correlation coefficient in, respectively, the posture change detection and inactivity segmentation tasks. The work paves the way for a reliable home-based analysis of movements during sleep which would serve patient-centred longitudinal care plans.

show abstract

Section: Joint Kinematics Preprocessingmentioning

confidence: 99%

Section: Data Availabilitymentioning

confidence: 99%

KIDS: kinematics-based (in)activity detection and segmentation in a sleep case study

Elnaggar¹,

Arelhi²,

Coenen³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The first visualisation-based approach utilises uniform manifold approximation and projection (UMAP) [60] to produce a two-dimensional (2D) force-directed graph U of high-dimensional datasets, such as Ψ(τ ) and * Ψ(•). Dimensionality reduction has been successfully applied to visualise data in many domains, including human motion analysis [36], pedagogical research [61] and speaker recognition [62]. In this work, UMAP facilitates the visualisation of postural observations of the same posture (intra-class distribution) as well as across different postures (inter-class distribution).…”

Section: Performance Interpretationmentioning

confidence: 99%

Sleep Posture One-Shot Learning Framework Using Kinematic Data Augmentation: In-Silico and In-Vivo Case Studies

Elnaggar¹,

Coenen²,

Hopkinson³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Sleep posture is linked to several health conditions such as nocturnal cramps and more serious musculoskeletal issues. However, in-clinic sleep assessments are often limited to vital signs (e.g. brain waves). Wearable sensors with embedded inertial measurement units have been used for sleep posture classification; nonetheless, previous works consider only few (commonly four) postures, which are inadequate for advanced clinical assessments. Moreover, posture learning algorithms typically require longitudinal data collection to function reliably, and often operate on raw inertial sensor readings unfamiliar to clinicians. This paper proposes a new framework for sleep posture classification based on a minimal set of joint angle measurements. The proposed framework is validated on a rich set of twelve postures in two experimental pipelines: computer animation to obtain synthetic postural data, and human participant pilot study using custom-made miniature wearable sensors. Through fusing raw geo-inertial sensor measurements to compute a filtered estimate of relative segment orientations across the wrist and ankle joints, the body posture can be characterised in a way comprehensible to medical experts. The proposed sleep posture learning framework offers plug-and-play posture classification by capitalising on a novel kinematic data augmentation method that requires only one training example per posture. Additionally, a new metric together with data visualisations are employed to extract meaningful insights from the postures dataset, demonstrate the added value of the data augmentation method, and explain the classification performance. The proposed framework attained promising overall accuracy as high as 100% on synthetic data and 92.7% on real data, on par with state of the art data-hungry algorithms available in the literature.

show abstract

“…The well-known t-distributed stochastic neighbourhood embedding (t-SNE) (Van Der Maaten et al, 2008) demonstrated significant success at producing 2D/3D visualisation of highdimensional datasets. t-SNE was employed in several domains, such as speaker identification (Elnaggar et al, 2019), handwritten digits and images (Abdelmoula et al, 2016) and animal behaviour mapping (Todd et al, 2017). Its main function is to capture the local data structures, while retaining the global data distribution.…”

Section: Visualisation Of the Ke Processmentioning

confidence: 99%

Quantification of Knowledge Exchange Within Classrooms: An AI-based Approach

Elnaggar¹,

Arelhi²

2021

ECE Official Conference Proceedings

Self Cite

View full text Add to dashboard Cite

Knowledge management improves efficiency and productivity of a company. A typical knowledge transfer pipeline, an enabler of knowledge management, consists of academics, higher education institutions, research funding bodies and companies. While the knowledge exchange mainstream sheds light on research collaborations, the evaluation of in-classroom knowledge exchange is often omitted, underestimating the impact this would have on the student employability. Current work on knowledge exchange at higher education institutions primarily focuses on: (i) collaborations with external parties, and (ii) identifying factors that affect knowledge sharing behaviours. This paper extends knowledge exchange to classroom teaching through: (i) formulating a framework among undergraduate Engineering students, and (ii) proposing an Artificial Intelligence based approach for evaluating the knowledge exchange process. The framework comprises of two group coursework with an intermediate handover event emulating an industrial workplace scenario in which knowledge exchange plays a key role. Then, an artificial intelligence-based visualisation technique processes data from two coursework-based surveys, completed before and after the abrupt handover event, to assess the change in the student intellectual backgrounds using two-dimensional maps embedding students as datapoints. The results interestingly reveal correlations between standard student evaluation metrics (for example grades, peer review and survey scores) and the formation of datapoint clusters. It is argued in the paper that the proposed artificial intelligence tool lends educators with tools to better understand the individual student performance in ways that are not captured by conventional academic assessments.

show abstract

A New Unsupervised Short-Utterance based Speaker Identification Approach with Parametric t-SNE Dimensionality Reduction

Cited by 7 publications

References 54 publications

KIDS: kinematics-based (in)activity detection and segmentation in a sleep case study

KIDS: kinematics-based (in)activity detection and segmentation in a sleep case study

Sleep Posture One-Shot Learning Framework Using Kinematic Data Augmentation: In-Silico and In-Vivo Case Studies

Quantification of Knowledge Exchange Within Classrooms: An AI-based Approach

Contact Info

Product

Resources

About