Postdoctoral Researcher in Statistics, School of Computer Science & Statistics, Trinity College Dublin
Room number: LB1.05
Email: tobinjo@tcd.ie
Links: LinkedIn, Google Scholar, Github
Current Research
I am the Trinity postdoctoral research associate for the AIM4Health project, exploring the influence of nutrition, lifestyle, genetic, and socio-economic factors on mental health problems in older people. As part of my work, I develop novel statistical approaches to allow a team of multi-disciplinary researchers to gain new insights from large complex datasets. AIM4Health is a North-South partnership project between Trinity College Dublin and Ulster University.
Publications
Conference Papers
‘Reinforced EM Algorithm for Clustering with Gaussian Mixture Models’ – Joshua Tobin, Chin Pang Ho & Mimi Zhang. Presented at the SIAM International Conference on Data Mining 2023.
‘DCF: An Efficient and Robust Density-Based Clustering Method’ – Joshua Tobin & Mimi Zhang. Presented at the IEEE International Conference on Data Mining 2021.
Journal Articles
‘A Theoretical Analysis of Density Peaks Clustering and the CPF Algorithm’ – Joshua Tobin & Mimi Zhang. Submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence.
Theses
‘Consistent Mode-Finding for Parametric and Non-Parametric Clustering’ – Joshua Tobin. PhD Thesis submitted in October 2022.
Ongoing Projects
‘Robust Modal Clustering’ – Joshua Tobin & Mimi Zhang.
‘Multi-view Co-clustering with the Latent Block Model’ – Joshua Tobin, James Ng & Mimi Zhang.
‘Cluster Analysis of Multi-Variate Functional Data through Non-linear Representation Learning’ – Joshua Tobin & Mimi Zhang.
Software
PyPi – CPFCluster, an implementation of the Component-wise Peak-Finding (CPF) clustering method.
Further projects are available on my Github repository.
Presentations
On March 22nd 2023, I presented the work completed to date as part of the AIM4Health project at Ulster University’s IRSC seminar series. A recording of the talk is available here.
On July 20th 2022, I presented at the International Federation of Classification Society’s Bi-Annual Conference in Porto. The talk introduced ongoing work related to the development of a reinforced EM algorithm for clustering with GMMs.
On July 14th 2021, I presented my contribution to the Smart Dublin Active Travel Challenge 2021. My project aimed to develop an algorithm which could automatically detect broken or unusable bikes in Dublin’s dockless bikesharing networks. In collaboration with Dublin’s two largest bikeshare operators, I developed the first attempt at such an algorithm. My project was one of twelve finalists and was showcased in a webinar attended by public officials from multiple EU countries and active travel industry leaders. A recording of the event is available here. The application of machine learning methods for dockless active travel networks remains an area of interest, and I hope to further develop and improve my method in the near future.
Blog Posts
- Seminar 11/05: ‘Consistent Mode-Finding for Parametric and Non-Parametric Clustering’
- Seminar 13/04: ‘Model-Based Clustering of Flow and Mass Cytometry Data’
- Seminar 30/03: PhD Lightning Round
- Seminar 19/04: ‘Latent Factor Bayesian Multivariate INAR Models’
- Seminar 23/03: ‘A Bayesian approach for regression in the presence of covariate shift: an application to galaxies redshift estimation’
- Seminar 16/03: ‘A latent variable model to infer food intake using multiple metabolomic biomarkers’
- Seminar 07/3: ‘Bayesian analysis of diffusion-driven multi-type epidemic models with application to COVID-19‘
- Seminar 31/5: ‘Representation Costs of Linear Neural Networks: Analysis and Design’
- Seminar 27/4: ‘Bayesian Distributed Lag Regression and Model Selection Method for the Analysis of ANCA Vasculitis Flares.’