BZPEERReading, trust and collaboration for research
Menu

Discover

GraphFollow connections around a paper, topic or saved view.

Workspaces

HomePick up where your research flow left off.InboxHandle requests, replies and notifications from one place.CollectionsCurate reading lists, evidence sets and shareable dossiers.ResearchSee your private provenance graph, trail and imports.CollaborationMove from discovery into focused discussion rooms.PublishDraft, preview and publish full-text research articles.

Network

ExpertsFind specialists worth following or asking for help.CommunitiesJoin research circles organized around shared work.PartnersSee external organizations active around the same topics.

Opportunities

ListingsBrowse roles, calls and collaborations with clearer fit signals.

Account

Log inReturn to your reading and collaboration workspace.Create accountStart saving work, following people and joining teams.
Log inCreate account

Researcher profile

Greg Taylor

Greg Taylor contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate
Applicationsastro-ph.HEastro-ph.IMastro-ph.SRq-fin.RMMachine LearningMethodologyq-fin.ST

Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
7works
0followers
8topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

7 published item(s)

preprint2025arXiv

On the use of case estimate and transactional payment data in neural networks for individual loss reserving

The use of neural networks trained on individual claims data has become increasingly popular in the actuarial reserving literature. We consider how to best input historical payment data in neural network models. Additionally, case estimates are also available in the format of a time series, and we extend our analysis to assessing their predictive power. In this paper, we compare a feed-forward neural network trained on summarised transactions to a recurrent neural network equipped to analyse a claim's entire payment history and/or case estimate development history. We draw conclusions from training and comparing the performance of the models on multiple, comparable highly complex datasets simulated from SPLICE (Avanzi, Taylor and Wang, 2023). We find evidence that case estimates will improve predictions significantly, but that equipping the neural network with memory only leads to meagre improvements. Although the case estimation process and quality will vary significantly between insurers, we provide a standardised methodology for assessing their value.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Bringing SOUL on sky

The SOUL project is upgrading the 4 SCAO systems of LBT, pushing the current guide star limits of about 2 magnitudes fainter thanks to Electron Multiplied CCD detector. This improvement will open the NGS SCAO correction to a wider number of scientific cases from high contrast imaging in the visible to extra-galactic source in the NIR. The SOUL systems are today the unique case where pyramid WFS, adaptive secondary and EMCCD are used together. This makes SOUL a pathfinder for most of the ELT SCAO systems like the one of GMT, MICADO and HARMONI of E-ELT, where the same key technologies will be employed. Today we have 3 SOUL systems installed on the telescope in commissioning phase. The 4th system will be installed in a few months. We will present here the results achieved during daytime testing and commissioning nights up to the present date.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Stochastic loss reserving with mixture density neural networks

Neural networks offer a versatile, flexible and accurate approach to loss reserving. However, such applications have focused primarily on the (important) problem of fitting accurate central estimates of the outstanding claims. In practice, properties regarding the variability of outstanding claims are equally important (e.g., quantiles for regulatory purposes). In this paper we fill this gap by applying a Mixture Density Network ("MDN") to loss reserving. The approach combines a neural network architecture with a mixture Gaussian distribution to achieve simultaneously an accurate central estimate along with flexible distributional choice. Model fitting is done using a rolling-origin approach. Our approach consistently outperforms the classical over-dispersed model both for central estimates and quantiles of interest, when applied to a wide range of simulated environments of various complexity and specifications. We further extend the MDN approach by proposing two extensions. Firstly, we present a hybrid GLM-MDN approach called "ResMDN". This hybrid approach balances the tractability and ease of understanding of a traditional GLM model on one hand, with the additional accuracy and distributional flexibility provided by the MDN on the other. We show that it can successfully improve the errors of the baseline ccODP, although there is generally a loss of performance when compared to the MDN in the examples we considered. Secondly, we allow for explicit projection constraints, so that actuarial judgement can be directly incorporated in the modelling process. Throughout, we focus on aggregate loss triangles, and show that our methodologies are tractable, and that they out-perform traditional approaches even with relatively limited amounts of data. We use both simulated data -- to validate properties, and real data -- to illustrate and ascertain practicality of the approaches.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Detection of Echoes in PSR B1508+55 at frequencies below 100 MHz using the LWA1

PSR B1508+55 is known to have a single component profile above 300 MHz. However, when we study it at frequencies below 100 MHz using the first station of the Long Wavelength Array, it shows multiple components. These include the main pulse, a precursor, a postcursor, and a trailing component. The separation of the trailing component from the main peak evolves over the course of a three year study. This evolution is likely an effect of the pulse signal getting refracted off an ionized gas cloud (acting as a lens) leading to what appears to be a trailing component in the profile as the pulsar signal traverses the interstellar medium. Using this interpretation, we identify the location and electron density of the lens affecting the pulse profile.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Modelling and understanding count processes through a Markov-modulated non-homogeneous Poisson process framework

The Markov-modulated Poisson process is utilised for count modelling in a variety of areas such as queueing, reliability, network and insurance claims analysis. In this paper, we extend the Markov-modulated Poisson process framework through the introduction of a flexible frequency perturbation measure. This contribution enables known information of observed event arrivals to be naturally incorporated in a tractable manner, while the hidden Markov chain captures the effect of unobservable drivers of the data. In addition to increases in accuracy and interpretability, this method supplements analysis of the latent factors. Further, this procedure naturally incorporates data features such as over-dispersion and autocorrelation. Additional insights can be generated to assist analysis, including a procedure for iterative model improvement. Implementation difficulties are also addressed with a focus on dealing with large data sets, where latent models are especially advantageous due the large number of observations facilitating identification of hidden factors. Namely, computational issues such as numerical underflow and high processing cost arise in this context and in this paper, we produce procedures to overcome these problems. This modelling framework is demonstrated using a large insurance data set to illustrate theoretical, practical and computational contributions and an empirical comparison to other count models highlight the advantages of the proposed approach.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Observing Flare Stars Below 100 MHz with the LWA

We observed the flare stars AD Leonis, Wolf 424, EQ Pegasi, EV Lacertae, and UV Ceti for nearly 135 hours. These stars were observed between 63 and 83 MHz using the interferometry mode of the Long Wavelength Array. Given that emission from flare stars is typically circularly polarized, we used the condition that any significant detection present in Stokes I must also be present in Stokes V at the same time in order for us to consider it a possible flare. Following this, we made one marginal flare detection for the star EQ Pegasi. This flare had a flux density of 5.91 Jy in Stokes I and 5.13 Jy in Stokes V, corresponding to a brightness temperature $1.75 \times 10^{16}(r/r_*)^{-2}$ K.

0 citations0 reviewsNo code linkedOpen access
preprint2019arXiv

Self-triggered radio detection and identification of cosmic air showers with the OVRO-LWA

A successful ground array Radio Frequency (RF)-only self-trigger on 10 high-energy cosmic ray events is demonstrated with 256 dual-polarization antennas of the Owens Valley Radio Observatory Long Wavelength Array (OVRO-LWA). This RF-only capability is predicated on novel techniques for Radio Frequency Interference (RFI) identification and mitigation with an analysis efficiency of 45\% for shower-driven events with a Signal-to-noise ratio $\gtrsim$ 5 against the galactic background noise power of individual antennas. This technique enables more efficient detection of cosmic rays over a wider range of zenith angles than possible via triggers from in-situ particle detectors and can be easily adapted by neutrino experiments relying on RF-only detection. This paper discusses the system design, RFI characterization and mitigation techniques, and initial results from 10 cosmic ray events identified within a 40-hour observing window. A design for a future optimized commensal cosmic-ray detector for the OVRO-LWA is presented, as well as recommendations for developing a similar capability for other experiments -- these designs either reduce data-rate or increase sensitivity by an order of magnitude for many configurations of radio instruments.

0 citations0 reviewsNo code linkedOpen access