|
|
Ockham's Razor from a Fully Probabilistic Design Perspective
Hoffmann, A. ; Quinn, Anthony
This research report investigates an approach to the design of an Ockham prior penalising parametric complexity in the Hierarchical Fully Probabilistic Design (HFPD) [1] setting. We identify a term which penalises the introduction of an additional parameter in the Wold decomposition. We also derive the objective Ockham Parameter Prior (OPI) in this context, based on earlier work [2], and we show that the two are, in fact, closely related. This confers validity on the HFPD Ockham term.
|
|
|
Transferring Improved Local Kernel Design in Multi-Source Bayesian Transfer Learning, with an application in Air Pollution Monitoring in India
Nugent, Sh. ; Quinn, Anthony
Existing frameworks for multi-task learning [1],[2] often rely on completely modelled relationships between tasks, which may not be available. Recent work [3], [4] has been undertaken on approaches to fully probabilistic methods for transfer learning between two Gaussian Process (GP) tasks. There, the target algorithm accepts source knowledge in the form of a probabilistic prior from a source algorithm, without requiring the target to model their interaction with the source. These strategies have offered robust improvements on current state of the art algorithms, such as the Intrinsic Coregionalization Model. The Bayesian Transfer Learning algorithm proposed in [4], was found to provide robust, positive\ntransfer. This algorithm was then extended to accommodate knowledge transfer from multiple source modellers [5]. Improved predictive performance was observed from increases in the number of sources. This report reviews the multi-source transfer findings in [5] and applies it to a real world problem of pollution modelling in India, using public-domain data.
|
|
| |
|
|
Bayesian Selective Transfer Learning for Patient-Specific Inference in Thyroid Radiotherapy
Murray, Sean Ernest ; Quinn, Anthony
This research report outlines a selective transfer approach for Bayesian estimation of patient-specific levels of radioiodine activity in the thyroid during the treatment of differentiated thyroid carcinoma. The work seeks to address some limitations of previous approaches [4] which involve generic, non-selective transfer of archival data. It is proposed that improvements in patient-specific inferences may be achieved via transferring external population knowledge selectively. This involves matching the patient to a similar sub-population based on available metadata, generating a Gaussian Mixture Model within the partitioned data, and optimally transferring a data predictive distribution from the sub-population to the specific patient. Additionally, a performance evaluation method is proposed and early-stage results presented.
|
|
|
Identifikace aktivity štítné žlázy a pravděpodobnostní odhadování absorbovaných dávek v nukleární medicíně
Jirsa, Ladislav ; Quinn, A. ; Varga, F.
The Bayesian identification of a linear regression model (called the biphasic model) for time dependence of thyroid gland activity in 131I radiotherapy is presented. Prior knowledge is elicited via hard parameter constraints and via the merging of external information from an archive of patient records. This prior regularization is shown to be crucial in the reported context, where data typically comprise only two or three high-noise measurements. The posterior distribution is simulated via a Langevin diffusion algorithm, whose optimization for the thyroid activity application is explained. Excellent patient-specific predictions of thyroid activity are reported. The posterior inference of the patient-specific total radiation dose is computed, allowing the uncertainty of the dose to be quantified in a consistent form. The relevance of this work in clinical practice is explained.
|
|
|
Bayesovský odhad optimálních okamžiků snímání lymfoscintigrafických obrazů
Doyle, S. ; Quinn, A. ; Gebouský, Petr
Recent research has pioneered quantitative Bayesian methods for diagnosis of lymphedema. The effectiveness of this approach depends on the selection of optimal lymphoscintigraphic imaging times for the entire limb. This paper develops a multichannel parametric model of the arm and derives an expression for the optimal lymphoscintigraphic sampling times. The Bayesian inference algorithm is applied to an over-sampled scintigraphic image sequence, and an optimal subret of such images is inferred.
|
|
| |
guest ::
