OpenUCT :: Browsing by Author "Bassett, Bruce"

Browsing by Author "Bassett, Bruce"

Now showing 1 - 20 of 22

Open Access
Aspects of Bayesian inference, classification and anomaly detection
(2021) Roberts, Ethan; Bassett, Bruce
The primary objective of this thesis is to develop rigorous Bayesian tools for common statistical challenges arising in modern science where there is a heightened demand for precise inference in the presence of large, known uncertainties. This thesis explores in detail two arenas where this manifests. The first is the development and testing of a unified Bayesian anomaly detection and classification framework (BADAC) which allows principled anomaly detection in the presence of measurement uncertainties, which are rarely incorporated into machine learning algorithms. BADAC deals with uncertainties by marginalising over the unknown, true value of the data. Using simulated data with Gaussian noise as an example, BADAC is shown to be superior to standard algorithms in both classification and anomaly detection performance in the presence of uncertainties. Additionally, BADAC provides well-calibrated classification probabilities, valuable for use in scientific pipelines. BADAC is therefore ideal where computational cost is not a limiting factor and statistical rigour is important. We discuss approximations to speed up BADAC, such as the use of Gaussian processes, and finally introduce a new metric, the Rank-Weighted Score (RWS), that is particularly suited to evaluating an algorithm's ability to detect anomalies. The second major exploration in this thesis presents methods for rigorous statistical inference in the presence of classification uncertainties and errors. Although this is explored specifically through supernova cosmology, the context is general. Supernova cosmology without spectra will be an important component of future surveys due to massive increases in data volumes in next-generation surveys such as from the Vera C. Rubin Observatory. This lack of supernova spectra results both in uncertainty in the redshifts and type of the supernova, which if ignored, leads to significantly biased estimates of cosmological parameters. We present a hierarchical Bayesian formalism, zBEAMS, which addresses this problem by marginalising over the unknown or uncertain supernova redshifts and types to produce unbiased cosmological estimates that are competitive with supernova data with fully spectroscopically confirmed redshifts. zBEAMS thus provides a unified treatment of both photometric redshifts, classification uncertainty and host galaxy misidentification, effectively correcting the inevitable contamination in the Hubble diagram with little or no loss of statistical power.
Open Access
Challenges in the hunt for dark energy dynamics
(2008) Hlozek, Renée; Bassett, Bruce
One of the greatest challenges in modern cosmology is determining the origin of the observed acceleration of the Universe. The 'dark energy' believed to supply the negative pressure responsible for this cosmic acceleration remains elusive despite over a decade of investigation. Hunting for deviation from the 'vanilla' cosmological model, ACDM, and detecting dynamics with redshift in the equation of state remains a key research area, with many challenges. We introduce some of the challenges in the search for such dark energy dynamics. We illustrate that under the assumption of well-motivated scaling models for dark energy dynamics early universe constraints on the dark energy density imply that these models will be essentially indistinguishable from ACDM for the next decade. After introducing the Fisher Matrix formalism, we derive the Fisher Flex test as a measure of whether the assumption of Gaussianity in the likelihood is incorrect for parameter estimation. This formalism is general for any cosmological survey. Lastly, we study the degeneracies between dark energy and curvature and matter in a non-parametric approach, and show that incorrectly assuming values of cosmological components can exactly mimic dark energy dynamics. We connect to the parametric approach by showing how these uncertainties also degrade constraints on the dark energy parameters in an assumed functional form for w. Improving the accuracy of surveys and experiments to search for possible signatures of dark energy dynamics is the focus of much attention in contemporary cosmology; we highlight challenges in the hunt for dark energy dynamics.
Open Access
Classification and visualisation of text documents using networks
(2018) Phaweni, Thembani; Durbach, Ian; Varughese, Melvin; Bassett, Bruce
In both the areas of text classification and text visualisation graph/network theoretic methods can be applied effectively. For text classification we assessed the effectiveness of graph/network summary statistics to develop weighting schemes and features to improve test accuracy. For text visualisation we developed a framework using established visual cues from the graph visualisation literature to communicate information intuitively. The final output of the visualisation component of the dissertation was a tool that would allow members of the public to produce a visualisation from a text document. We represented a text document as a graph/network. The words were nodes and the edges were created when a pair of words appeared within a pre-specified distance (window) of words from each other. The text document model is a matrix representation of a document collection such that it can be integrated into a machine or statistical learning algorithm. The entries of this matrix can be weighting according to various schemes. We used the graph/network representation of a text document to create features and weighting schemes that could be applied to the text document model. This approach was not well developed for text classification therefore we applied different edge weighting methods, window sizes, weighting schemes and features. We also applied three machine learning algorithms, naïve Bayes, neural networks and support vector machines. We compared our various graph/network approaches to the traditional document model with term frequency inverse-document-frequency. We were interested in establishing whether or not the use of graph weighting schemes and graph features could increase test accuracy for text classification tasks. As far as we can tell from the literature, this is the first attempt to use graph features to weight bag-of-words features for text classification. These methods had been applied to information retrieval (Blanco & Lioma, 2012). It seemed they could also be applied to text classification. The text visualisation field seemed divorced from the text summarisation and information retrieval fields, in that text co-occurrence relationships were not treated with equal importance. Developments in the graph/network visualisation literature could be taken advantage of for the purposes of text visualisation. We created a framework for text visualisation using the graph/network representation of a text document. We used force directed algorithms to visualise the document. We used established visual cues like, colour, size and proximity in space to convey information through the visualisation. We also applied clustering and part-of-speech tagging to allow for filtering and isolating of specific information within the visualised document. We demonstrated this framework with four example texts. We found that total degree, a graph weighting scheme, outperformed term frequency on average. The effect of graph features depended heavily on the machine learning method used: for the problems we considered graph features increased accuracy for SVM classifiers, had little effect for neural networks and decreased accuracy for naïve Bayes classifiers Therefore the impact on test accuracy of adding graph features to the document model is dependent on the machine learning algorithm used. The visualisation of text graphs is able to convey meaningful information regarding the text at a glance through established visual cues. Related words are close together in visual space and often connected by thick edges. Large nodes often represent important words. Modularity clustering is able to extract thematically consistent clusters from text graphs. This allows for the clusters to be isolated and investigated individually to understand specific themes within a document. The use of part-of-speech tagging is effective in both reducing the number of words being displayed but also increasing the relevance of words being displayed. This was made clear through the use of part-of-speech tags applied to the Internal Resistance of Apartheid Wikipedia webpage. The webpage was reduced to its proper nouns which contained much of the important information in the text. Training accuracy is important in text classification which is a task that can often be performed on vast amounts of documents. Much of the research in text classification is aimed at increasing classification accuracy either through feature engineering, or optimising machine learning methods. The finding that total degree outperformed term frequency on average provides an alternative avenue for achieving higher test accuracy. The finding that the addition of graph features can increase test accuracy when matched with the right machine learning algorithm suggests some new research should be conducted regarding the role that graph features can have in text classification. Text visualisation is used as an exploratory tool and as a means of quickly and easily conveying text information. The framework we developed is able to create automated text visualisations that intuitively convey information for short and long text documents. This can greatly reduce the amount of time it takes to assess the content of a document which can increase general access to information.
Open Access
Classification of multiwavelength transients with machine learning
(2019) Sooknunan, Kimeel; Lochner, Michelle; Bassett, Bruce
With the advent of powerful telescopes such as the Square Kilometre Array (SKA), its precursor MeerKAT and the Large Synoptic Survey Telescope (LSST), we are entering a golden era of multiwavelength transient astronomy. The large MeerKAT science project ThunderKAT may dramatically increase the detected number of radio transients. Currently radio transient datasets are still very small, allowing spectroscopic classification of all objects of interest. As the event rate increases, follow-up resources must be prioritised by making use of early classification of the radio data. Machine learning algorithms have proven themselves invaluable in the context of optical astronomy, however it has yet to be applied to radio transients. In the burgeoning era of multimessenger astronomy, incorporating data from different telescopes such as MeerLICHT, Fermi, LSST and the gravitational wave observatory LIGO could significantly improve classification of events. Here we present MALT (Machine Learning for Transients): a general machine learning pipeline for multiwavelength transient classification. In order to make use of most machine learning algorithms, "features" must be extracted from complex and often high dimensional datasets. In our approach, we first interpolate the data onto a uniform grid using Gaussian processes, we then perform a wavelet decomposition and finally reduce the dimensionality using principal component analysis. We then classify the light curves with the popular machine learning algorithm random forests. For the first time, we apply machine learning to the classification of radio transients. Unfortunately publicly available radio transient data is scarce and our dataset consists of just 87 light curves, with several classes only consisting of a single example. However machine learning is often applied to such small datasets by making use of data augmentation. We develop a novel data augmentation technique based on Gaussian processes, able to generate new data statistically consistent with the original. As the dataset is currently small, three studies were done on the effect of the training set. The classifier was trained on a non-representative training set, achieving an overall accuracy of 77.8% over all 11 classes with the known 87 lightcurves with just eight hours of observations. The expected increase in performance, as more training data are acquired, is shown by training the classifier on a simulated representative training set, achieving an average accuracy of 95.8% across all 11 classes. Finally, the effectiveness of including multiwavelength data for general transient classification is demonstrated. First the classifier is trained on wavelet features and a contextual feature, achieving an average accuracy of 72.9%. The classifier was then trained on wavelet features and a contextual feature, together with a single optical flux feature. This addition improves the overall accuracy to 94.7%. This work provides a general approach for multiwavelength transient classification and shows that machine learning can be highly effective at classifying the influx of radio transients anticipated with MeerKAT and other radio telescopes.
Open Access
Cosmic acceleration and the coincidence problem
(2009) Kubwimana, Jean Claude; Bassett, Bruce
In the standard model of the Universe, the cosmos has only accelerated once since decoupling and only recently, at around a redshift of z ̃ 0.5 as supported by different observations including Type Ia Supernovae (SNIa), the Cosmic Microwave Background (CMB), Large Scale Structure (LSS), and Weak Lensing (WL). This confirmation however, lacks a fundamental physics explanation. The hypothetical form of energy termed 'dark energy' (DE) assumed to account for that acceleration behavior, is still mysterious and why its dominance only occurred recently is a profound problem widely known as the coincidence problem. So far all attempts for resolving the coincidence the problem have been unsatisfactory. Here we investigate a possible solution to the coincidence problem in the form of multiples phases of acceleration (MPA). If there were more than one phase of acceleration between now and decoupling, then the current phase of acceleration would be much less special, alleviating the coincidence problem. We use a modified Markov Chain Monte Carlo (MCMC) technique together with the WMAP five year TT data to search for parameters allowing a second phase of acceleration. Despite extensive search we find no models that simultaneously fit the WMAP data and yield a second phase of acceleration, ruling out this particular set of models as the solution to the coincidence problem.
Open Access
Cosmological constraints from the SDSS luminous red galaxies
(2006) Tegmark, Max; Eisenstein, Daniel J; Strauss, Michael A; Weinberg, David H; Blanton, Michael R; Frieman, Joshua A; Fukugita, Masataka; Gunn, James E; Hamilton, Andrew J S; Knapp, Gillian R; Nichol, Robert C; Ostriker, Jeremiah P; Padmanabhan, Nikhil; Percival, Will J; Schlegel, David J; Schneider, Donald P; Scoccimarro, Roman; Seljak, Uroš; Seo, Hee-Jong; Swanson, Molly; Szalay, Alexander S; Vogeley, Michael S; Yoo, Jaiyul; Zehavi, Idit; Abazajian, Kevork; Anderson, Scott F; Annis, James; Bahcall, Neta A; Bassett, Bruce; Berlind, Andreas; Brinkmann, Jon; Budavári, Tamás
No abstract prepared.
Open Access
Deep adaptive anomaly detection using an active learning framework
(2022) Sekyi, Emmanuel; Bassett, Bruce
Anomaly detection is the process of finding unusual events in a given dataset. Anomaly detection is often performed on datasets with a fixed set of predefined features. As a result of this, if the normal features bear a close resemblance to the anomalous features, most anomaly detection algorithms exhibit poor performance. This work seeks to answer the question, can we deform these features so as to make the anomalies standout and hence improve the anomaly detection outcome? We employ a Deep Learning and an Active Learning framework to learn features for anomaly detection. In Active Learning, an Oracle (usually a domain expert) labels a small amount of data over a series of training rounds. The deep neural network is trained after each round to incorporate the feedback from the Oracle into the model. Results on the MNIST, CIFAR-10 and Galaxy Zoo datasets show that our algorithm, Ahunt, significantly outperforms other anomaly detection algorithms used on a fixed, static, set of features. Ahunt can therefore overcome a poor choice of features that happen to be suboptimal for detecting anomalies in the data, learning more appropriate features. We also explore the role of the loss function and Active Learning query strategy, showing these are important, especially when there is a significant variation in the anomalies.
Open Access
Deep learning for supernovae detection
(2017) Amar, Gilad; Bassett, Bruce
In future astronomical sky surveys it will be humanly impossible to classify the tens of thousands of candidate transients detected per night. This thesis explores the potential of using state-of-the-art machine learning algorithms to handle this burden more accurately and quickly than trained astronomers. To this end Deep Learning methods are applied to classify transients using real-world data from the Sloan Digital Sky Survey. Using cutting-edge training techniques several Convolutional Neural networks are trained and hyper-parameters tuned to outperform previous approaches and find that human labelling errors are the primary obstacle to further improvement. The tuning and optimisation of the deep models took in excess of 700 hours on a 4-Titan X GPU cluster.
Open Access
Evolutionary algorithms for optimising reinforcement learning policy approximation
(2019) Cuningham, Blake; Bassett, Bruce
Reinforcement learning methods have become more efficient in recent years. In particular, the A3C (asynchronous advantage actor critic) approach demonstrated in Mnih et al. (2016) was able to halve the training time of the existing state-of-the-art approaches. However, these methods still require relatively large amounts of training resources due to the fundamental exploratory nature of reinforcement learning. Other machine learning approaches are able to improve the ability to train reinforcement learning agents by better processing input information to help map states to actions - convolutional and recurrent neural networks are helpful when input data is in image form that does not satisfy the Markov property. The specific required architecture of these convolutional and recurrent neural network models is not obvious given infinite possible permutations. There is very limited research giving clear guidance on neural network structure in a RL (reinforcement learning) context, and grid search-like approaches require too many resources and do not always find good optima. In order to address these, and other, challenges associated with traditional parameter optimization methods, an evolutionary approach similar to that taken by Dufourq and Bassett (2017) for image classification tasks was used to find the optimal model architecture when training an agent that learns to play Atari Pong. The approach found models that were able to train reinforcement learning agents faster, and with fewer parameters than that found by OpenAI’s model in Blackwell et al. (2018) - a superhuman level of performance.
Open Access
FireFly: A Bayesian Approach to Source Finding in Astronomical Data
(2019) Moloko, Oarabile Hope; Lochner, Michelle; Bassett, Bruce
Efficient and rigorous source finding techniques are needed for the upcoming large data sets from telescopes like MeerKAT, LSST and the SKA. Most of the current source-finding algorithms lack full statistical rigor. Typically these algorithms use some form of thresholding to find sources, which leads to contamination and missed sources. Ideally we would like to use all the available information when performing source detection, including any prior knowledge we may have. Bayesian statistics is the obvious approach as it allows precise statistical interrogations of the data and the inclusion of all available information. In this thesis, we implement nested sampling and Monte Carlo Markov Chain (MCMC) techniques to develop a new Bayesian source finding technique called FireFly. FireFly employs a technique of switching ‘on’ and ‘off’ sources during sampling to deal with the fact that we don’t know how many true sources are present. It therefore tackles one of the critical questions in source finding, which is estimating the number of real sources in the image. We compare FireFly against a Bayesian evidence-based search method and show on simulated astronomical images that FireFly outperforms the evidence-based approach. We further investigate two implementations of FireFly: the first with nested sampling and the second with MCMC. Our results show that MCMC FireFly has better computational scaling than the nested sampling version FireFly but the nested sampling version of FireFly appears to perform somewhat better than MCMC FireFly. Future work should examine how best to quantify FireFly performance and extend the formalism developed here to deal with multiwavelength data.
Open Access
Genetic programming applied to RFI mitigation in radio astronomy
(2016) Staats, Kai; Bassett, Bruce
Genetic Programming is a type of machine learning that employs a stochastic search of a solutions space, genetic operators, a fitness function, and multiple generations of evolved programs to resolve a user-defined task, such as the classification of data. At the time of this research, the application of machine learning to radio astronomy was relatively new, with a limited number of publications on the subject. Genetic Programming had never been applied, and as such, was a novel approach to this challenging arena. Foundational to this body of research, the application Karoo GP was developed in the programming language Python following the fundamentals of tree-based Genetic Programming described in "A Field Guide to Genetic Programming" by Poli, et al. Karoo GP was tasked with the classification of data points as signal or radio frequency interference (RFI) generated by instruments and machinery which makes challenging astronomers' ability to discern the desired targets. The training data was derived from the output of an observation run of the KAT-7 radio telescope array built by the South African Square Kilometre Array (SKA-SA). Karoo GP, kNN, and SVM were comparatively employed, the outcome of which provided noteworthy correlations between input parameters, the complexity of the evolved hypotheses, and performance of raw data versus engineered features. This dissertation includes description of novel approaches to GP, such as upper and lower limits to the size of syntax trees, an auto-scaling multiclass classifier, and a Numpy array element manager. In addition to the research conducted at the SKA-SA, it is described how Karoo GP was applied to fine-tuning parameters of a weather prediction model at the South African Astronomical Observatory (SAAO), to glitch classification at the Laser Interferometer Gravitational-wave Observatory (LIGO), and to astro-particle physics at The Ohio State University.
Open Access
Heuristic optimisation and parameter estimation methods for modern cosmological surveys
(2006) Elson, E C; Bassett, Bruce; Van der Heyden, Kurt; Vilakazi, Zeblon
Includes bibliographical references.
Open Access
Investigating the relationship between mobile network performance metrics and customer satisfaction
(2019) Labuschagne, Louwrens; Bassett, Bruce; Little, Francesca
Fixed and mobile communication service providers (CSPs) are facing fierce competition among each other. In a globally saturated market, the primary di↵erentiator between CSPs has become customer satisfaction, typically measured by the Net Promoter Score (NPS) for a subscriber. The NPS is the answer to the question: ”How likely is it that you will recommend this product/company to a friend or colleague?” The responses range from 0 representing not at all likely to 10 representing extremely likely. In this thesis, we aim to identify which, if any, network performance metrics contribute to subscriber satisfaction. In particular, we investigate the relationship between the NPS survey results and 11 network performance metrics of the respondents of a major mobile operator in South Africa. We identify the most influential performance metrics by fitting both linear and non-linear statistical models to the February 2018 survey dataset and test the models on the June 2018 dataset. We find that metrics such as Call Drop Rate, Call Setup Failure Rate, Call Duration and Server Setup Latency are consistently selected as significant features in models of NPS prediction. Nevertheless we find that all the tested statistical and machine learning models, whether linear or non-linear, are poor predictors of NPS scores in a month, when only the network performance metrics in the same month are provided. This suggests that either NPS is driven primarily by other factors (such as customer service interactions at branches and contact centres) or are determined by historical network performance over multiple months.
Open Access
Machine learning in astronomy
(2015) Du Buisson, Lise; Bassett, Bruce
The search to find answers to the deepest questions we have about the Universe has fueled the collection of data for ever larger volumes of our cosmos. The field of supernova cosmology, for example, is seeing continuous development with upcoming surveys set to produce a vast amount of data that will require new statistical inference and machine learning techniques for processing and analysis. Distinguishing between real objects and artefacts is one of the first steps in any transient science pipeline and, currently, is still carried out by humans - often leading to hand scanners having to sort hundreds or thousands of images per night. This is a time-consuming activity introducing human biases that are extremely hard to characterise. To succeed in the objectives of future transient surveys, the successful substitution of human hand scanners with machine learning techniques for the purpose of this artefact-transient classification therefore represents a vital frontier. In this thesis we test various machine learning algorithms and show that many of them can match the human hand scanner performance in classifying transient difference g, r and i-band imaging data from the SDSS-II SN Survey into real objects and artefacts. Using principal component analysis and linear discriminant analysis, we construct a grand total of 56 feature sets with which to train, optimise and test a Minimum Error Classifier (MEC), a naive Bayes classifier, a k-Nearest Neighbours (kNN) algorithm, a Support Vector Machine (SVM) and the SkyNet artificial neural network.
Open Access
New applications of statistics in astronomy and cosmology
(2014) Lochner, Michelle Aileen Anne; Bassett, Bruce
Over the last few decades, astronomy and cosmology have become data-driven fields. The parallel increase in computational power has naturally lead to the adoption of more sophisticated statistical techniques for data analysis in these fields, and in particular, Bayesian methods. As the next generation of instruments comes online, this trend should be continued since previously ignored effects must be considered rigorously in order to avoid biases and incorrect scientific conclusions being drawn from the ever-improving data. In the context of supernova cosmology, an example of this is the challenge from contamination as supernova datasets will become too large to spectroscopically confirm the types of all objects. The technique known as BEAMS (Bayesian Estimation Applied to Multiple Species) handles this contamination with a fully Bayesian mixture model approach, which allows unbiased estimates of the cosmological parameters. Here, we extend the original BEAMS formalism to deal with correlated systematics in supernovae data, which we test extensively on thousands of simulated datasets using numerical marginalization and Markov Chain Monte Carlo (MCMC) sampling over the unknown type of the supernova, showing that it recovers unbiased cosmological parameters with good coverage. We then apply Bayesian statistics to the field of radio interferometry. This is particularly relevant in light of the SKA telescope, where the data will be of such high quantity and quality that current techniques will not be adequate to fully exploit it. We show that the current approach to deconvolution of radio interferometric data is susceptible to biases induced by ignored and unknown instrumental effects such as pointing errors, which in general are correlated with the science parameters. We develop an alternative approach - Bayesian Inference for Radio Observations (BIRO) - which is able to determine the joint posterior for all scientific and instrumental parameters. We test BIRO on several simulated datasets and show that it is superior to the standard CLEAN and source extraction algorithms. BIRO fits all parameters simultaneously while providing unbiased estimates - and errors - for the noise, beam width, pointing errors and the fluxes and shapes of the sources.
Open Access
Optimising the Optimiser: Meta NeuroEvolution for Artificial Intelligence Problems
(2021) Hayes, Max Nieuwoudt; Bassett, Bruce; Clark, Allan
Since reinforcement learning algorithms have to fully solve a task in order to evaluate a set of hyperparameter values, conventional hyperparameter tuning methods can be highly sample inefficient and computationally expensive. Many widely used reinforcement learning architectures originate from scientific papers which include optimal hyperparameter values in the publications themselves, but do not indicate how the hyperparameter values were found. To address the issues related to hyperparameter tuning, three different experiments were investigated. In the first two experiments, Bayesian Optimisation and random search are compared. In the third and final experiment, the hyperparameter values found in second experiment are used to solve a more difficult reinforcement learning task, effectively performing hyperparameter transfer learning (later referred to as meta-transfer learning). The results from experiment 1 showed that there are certain scenarios in which Bayesian Optimisation outperforms random search for hyperparameter tuning, while the results of experiment 2 show that as more hyperparameters are simultaneously tuned, Bayesian Optimisation consistently finds better hyperparameter values than random search. However, BO took more than twice the amount of time to find these hyperparameter values than random search. Results from the third and final experiment indicate that hyperparameter values learned while tuning hyperparameters for a relatively easy to solve reinforcement learning task (Task A), can be used to solve a more complex task (Task B). With the available computing power for this thesis, hyperparameter optimisation was possible on the tasks in experiment 1 and experiment 2. This was not possible on the task in experiment 3, due to limited computing resources and the increased complexity of the reinforcement learning task in experiment 3, making the transfer of hyperparameters from one task (Task A) to the more difficult task (Task B) highly beneficial for solving the more computationally expensive task. The purpose of this work is to explore the effectiveness of Bayesian Optimisation as a hyperparameter tuning algorithm on the reinforcement learning algorithm NEAT's hyperparemters. An additional goal of this work is the experimental use of hyperparameter value transfer between reinforcement learning tasks, referred to in this work as Meta-Transfer Learning. This is introduced and discussed in greater detail in the Introduction chapter. All code used for this work is available in the repository: • https://github.com/maaxnaax/MSc_code
Open Access
Probing primordial non-Gaussianity using large scale structure
(2009) Fantaye, Yabebal; Bassett, Bruce; Blake, Chris
Recent evidence from the WMAP satellite has lead to a tentative detection of non-Gaussianity. Using the bispectrum statistic applied to the MegaZ catalogue of over 600,000 luminous red galaxies we find new bounds on the large-scale nonGaussianity. We constrain the fNL parameter using a particular type of triangular configuration as well as the combination of all the possible triangles in harmonic space. The constraint on fNL from the combination of all possible triangular configurations is ffV'ial = 57 ± 52 with 68% confidence limit, which is consistent with vanishing primordial non-Gaussianity, although some triangular configurations on their own suggest a non-zero bispectrum which, if confirmed, would have a profound effect on modern cosmology.
Open Access
Radio Frequency Interference: Simulations for Radio Interferometry Arrays
(2021) Finlay, Chris; Bassett, Bruce
Radio Frequency Interference (RFI) is a massive problem for radio observatories around the world. Due to the growth of telecommunications and air travel RFI is increasing exactly when the world's radio telescopes are increasing significantly in sensitivity, making RFI one of the most pressing problems for astronomy in the era of the Square Kilometre Array (SKA). Traditionally RFI is dealt with through simple algorithms that remove unexpected rapid changes but the recent explosion of machine learning and artificial intelligence (AI) provides an exciting opportunity for pushing the state-of-the-art in RFI excision. Unfortunately, due to the lack of training data for which the true RFI contamination is known, it is impossible to reliably train and compare machine learning algorithms for RFI excision on radio telescope arrays currently. To address this stumbling block we present RFIsim, a radio interferometry simulator that includes the telescope properties of the MeerKAT array, a sky model based on previous radio surveys coupled with an RFI model designed to reproduce actual RFI seen at the MeerKAT site. We perform an indepth comparison of the simulator results with real observations using the MeerKAT telescope and show that RFIsim produces visibilities that mimic those produced by real observations very well. Finally, we describe how the data was key in the development of a new state-of-the-art deep learning RFI flagging algorithm in Vafaei et al. (2020.) [69] In particular, this work demonstrates that transfer learning from simulation to real data is an effective way to leverage the power of machine learning for RFI flagging in real-world observatories.
Open Access
Reinforcement learning for telescope optimisation
(2019) Blows, Curtly; Bassett, Bruce
Reinforcement learning is a relatively new and unexplored branch of machine learning with a wide variety of applications. This study investigates reinforcement learning and provides an overview of its application to a variety of different problems. We then explore the possible use of reinforcement learning for telescope target selection and scheduling in astronomy with the hope of effectively mimicking the choices made by professional astronomers. This is relevant as next-generation astronomy surveys will require near realtime decision making in response to high-speed transient discoveries. We experiment with and apply some of the leading approaches in reinforcement learning to simplified models of the target selection problem. We find that the methods used in this study show promise but do not generalise well. Hence while there are indications that reinforcement learning algorithms could work, more sophisticated algorithms and simulations are needed.
Open Access
A study of distances and the Baryon Acoustic Oscillations in curved spacetimes.
(2012) Clarke, Alan; Bassett, Bruce
The Baryon Acoustic Oscillations offer a powerful method of measuring cosmological distances and the expansion history of the Universe. Understanding of the BAO comes from linear physics and allows for accurate predictions of the BAO scale. This will result in accurate measurements of the parameters of the Universe. Currently, most BAO measurements assume a flat cosmology; this work seeks to investigate if the assumption of flatness provides inaccuracies in the measurement process.