Insight Centre for Data Analytics - Conference Items
Permanent URI for this collection
- ItemBayesian optimization with multi-objective acquisition function for bilevel problem(Springer, 2023-02-23) Dogan, Vedat; Prestwich, Steven D.; Science Foundation Ireland; European Regional Development FundA bilevel optimization problem consists of an upper-level and a lower-level optimization problem connected to each other hierarchically. Efficient methods exist for special cases, but in general solving these problems is difficult. Bayesian optimization methods are an interesting approach that speed up search using an acquisition function, and this paper proposes a modified Bayesian approach. It treats the upper-level problem as an expensive black-box function, and uses multiple acquisition functions in a multi-objective manner by exploring the Pareto-front. Experiments on popular bilevel benchmark problems show the advantage of the method.
- ItemCouRGe: Counterfactual reviews generator for sentiment analysis(Springer, 2022-02-23) Carraro, Diego; Brown, Kenneth N.; Science Foundation Ireland; European Regional Development FundPast literature in Natural Language Processing (NLP) has demonstrated that counterfactual data points are useful, for example, for increasing model generalisation, enhancing model interpretability, and as a data augmentation approach. However, obtaining counterfactual examples often requires human annotation effort, which is an expensive and highly skilled process. For these reasons, solutions that resort to transformer-based language models have been recently proposed to generate counterfactuals automatically, but such solutions show limitations. In this paper, we present CouRGe, a language model that, given a movie review (i.e. a seed review) and its sentiment label, generates a counterfactual review that is close (similar) to the seed review but of the opposite sentiment. CouRGe is trained by supervised fine-tuning of GPT-2 on a task-specific dataset of paired movie reviews, and its generation is prompt-based. The model does not require any modification to the network’s architecture or the design of a specific new task for fine-tuning. Experiments show that CouRGe’s generation is effective at flipping the seed sentiment and produces counterfactuals reasonably close to the seed review. This proves once again the great flexibility of language models towards downstream tasks as hard as counterfactual reasoning and opens up the use of CouRGe’s generated counterfactuals for the applications mentioned above.
- ItemFrequency dependence of loop antenna H-Field in free-space(Royal Irish Academy, RIA, 2022-10-26) Su, Zhen; O'Callaghan, Brendan; Kumar, Sanjeev; O'Flynn, Brendan; Barton, John; Bulja, Senad; O'Hare Daniel; Buckley, John L.; Enterprise Ireland; IDA Ireland; Science Foundation Ireland; European Regional Development FundThis paper investigates the effect of frequency on the magnetic field (H-field) strength at a specified distance from a single turn transmitting (TX) loop antenna in free-space. The H-field of a TX loop antenna in the Near Field (NF) region can be estimated using Biot-Savartâ s Law (BSL). However, the BSL method is only valid for DC current flow and does not consider the effects of frequency on the H-field strength. This work introduces the frequency-dependent Antenna Near Field (ANF) method reported in the literature and compares this method against the BSL method and EM simulation results. It is shown that the ANF method can estimate the H-field magnitude with an error of less than 5% when compared with finite-element methods (FEM). In addition, the ANF method computes in a fraction of a second, whereas the FEM method takes several minutes to compute.
- ItemPredicting recurrence-free survival in prostate cancer following surgical intervention(2022-03) O'Donnell, Autumn; Moghaddam, Shirin; Cronin, Michael; Wolsztynski, Eric
- ItemMachine learning methodologies for modelling of time-to-event endpoints in prostate cancer(2022-08) O'Donnell, Autumn; Moghaddam, Shirin; Cronin, Michael; Wolsztynski, Eric; Science Foundation Ireland