Showing 1–10 of 10 results for author: Shenvi, A

Learning from data with structured missingness

Authors: Robin Mitra, Sarah F. McGough, Tapabrata Chakraborti, Chris Holmes, Ryan Copping, Niels Hagenbuch, Stefanie Biedermann, Jack Noonan, Brieuc Lehmann, Aditi Shenvi, Xuan Vinh Doan, David Leslie, Ginestra Bianconi, Ruben Sanchez-Garcia, Alisha Davies, Maxine Mackintosh, Eleni-Rosalina Andrinopoulou, Anahid Basiri, Chris Harbron, Ben D. MacArthur

Abstract: Missing data are an unavoidable complication in many machine learning tasks. When data are `missing at random' there exist a range of tools and techniques to deal with the issue. However, as machine learning studies become more ambitious, and seek to learn from ever-larger volumes of heterogeneous data, an increasingly encountered problem arises in which missing values exhibit an association or st… ▽ More Missing data are an unavoidable complication in many machine learning tasks. When data are `missing at random' there exist a range of tools and techniques to deal with the issue. However, as machine learning studies become more ambitious, and seek to learn from ever-larger volumes of heterogeneous data, an increasingly encountered problem arises in which missing values exhibit an association or structure, either explicitly or implicitly. Such `structured missingness' raises a range of challenges that have not yet been systematically addressed, and presents a fundamental hindrance to machine learning at scale. Here, we outline the current literature and propose a set of grand challenges in learning from data with structured missingness. △ Less

Submitted 3 April, 2023; originally announced April 2023.

Where the Bee Sucks -- A Dynamic Bayesian Network Approach to Decision Support for Pollinator Abundance Strategies

Authors: Martine J. Barons, Aditi Shenvi

Abstract: For policymakers wishing to make evidence-based decisions, one of the challenges is how to combine the relevant information and evidence in a coherent and defensible manner in order to formulate and evaluate candidate policies. Policymakers often need to rely on experts with disparate fields of expertise when making policy choices in complex, multi-faceted, dynamic environments such as those deali… ▽ More For policymakers wishing to make evidence-based decisions, one of the challenges is how to combine the relevant information and evidence in a coherent and defensible manner in order to formulate and evaluate candidate policies. Policymakers often need to rely on experts with disparate fields of expertise when making policy choices in complex, multi-faceted, dynamic environments such as those dealing with ecosystem services. The pressures affecting the survival and pollination capabilities of honey bees (Apis mellifera), wild bees and other pollinators is well-documented, but incomplete. In order to estimate the potential effectiveness of various candidate policies to support pollination services, there is an urgent need to quantify the effect of various combinations of variables on the pollination ecosystem service, utilising available information, models and expert judgement. In this paper, we present a new application of the integrating decision support system methodology for combining inputs from multiple panels of experts to evaluate policies to support an abundant pollinator population. △ Less

Submitted 5 December, 2022; originally announced December 2022.

cegpy: Modelling with Chain Event Graphs in Python

Authors: Gareth Walley, Aditi Shenvi, Peter Strong, Katarzyna Kobalczyk

Abstract: Chain event graphs (CEGs) are a recent family of probabilistic graphical models that generalise the popular Bayesian networks (BNs) family. Crucially, unlike BNs, a CEG is able to embed, within its graph and its statistical model, asymmetries exhibited by a process. These asymmetries might be in the conditional independence relationships or in the structure of the graph and its underlying event sp… ▽ More Chain event graphs (CEGs) are a recent family of probabilistic graphical models that generalise the popular Bayesian networks (BNs) family. Crucially, unlike BNs, a CEG is able to embed, within its graph and its statistical model, asymmetries exhibited by a process. These asymmetries might be in the conditional independence relationships or in the structure of the graph and its underlying event space. Structural asymmetries are common in many domains, and can occur naturally (e.g. a defendant vs prosecutor's version of events) or by design (e.g. a public health intervention). However, there currently exists no software that allows a user to leverage the theoretical developments of the CEG model family in modelling processes with structural asymmetries. This paper introduces cegpy, the first Python package for learning and analysing complex processes using CEGs. The key feature of cegpy is that it is the first CEG package in any programming language that can model processes with symmetric as well as asymmetric structures. cegpy contains an implementation of Bayesian model selection and probability propagation algorithms for CEGs. We illustrate the functionality of cegpy using a structurally asymmetric dataset. △ Less

Submitted 21 November, 2022; originally announced November 2022.

Beyond Conjugacy for Chain Event Graph Model Selection

Authors: Aditi Shenvi, Silvia Liverani

Abstract: Chain event graphs are a family of probabilistic graphical models that generalise Bayesian networks and have been successfully applied to a wide range of domains. Unlike Bayesian networks, these models can encode context-specific conditional independencies as well as asymmetric developments within the evolution of a process. More recently, new model classes belonging to the chain event graph famil… ▽ More Chain event graphs are a family of probabilistic graphical models that generalise Bayesian networks and have been successfully applied to a wide range of domains. Unlike Bayesian networks, these models can encode context-specific conditional independencies as well as asymmetric developments within the evolution of a process. More recently, new model classes belonging to the chain event graph family have been developed for modelling time-to-event data to study the temporal dynamics of a process. However, existing model selection algorithms for chain event graphs and its variants rely on all parameters having conjugate priors. This is unrealistic for many real-world applications. In this paper, we propose a mixture modelling approach to model selection in chain event graphs that does not rely on conjugacy. Moreover, we also show that this methodology is more amenable to being robustly scaled than the existing model selection algorithms used for this family. We demonstrate our techniques on simulated datasets. △ Less

Submitted 7 November, 2022; originally announced November 2022.

Score Equivalence for Staged Trees

Authors: Conor Hughes, Peter Strong, Aditi Shenvi

Abstract: Staged trees are a recently-developed, powerful family of probabilistic graphical models. An equivalence class of staged trees has now been characterised, and two fundamental statistical operators have been defined to traverse the equivalence class of a given staged tree. Here, two staged trees are said to be statistically equivalent when they represent the same set of distributions. Probabilistic… ▽ More Staged trees are a recently-developed, powerful family of probabilistic graphical models. An equivalence class of staged trees has now been characterised, and two fundamental statistical operators have been defined to traverse the equivalence class of a given staged tree. Here, two staged trees are said to be statistically equivalent when they represent the same set of distributions. Probabilistic graphical models such as staged trees are increasingly being used for causal analyses. Staged trees which are within the same equivalence class can encode very different causal hypotheses but data alone cannot help us distinguish between these. Therefore, in using score-based methods to learn the model structure and distributions from data for causal analyses, we should expect that a suitable scoring function is one which assigns the same score to statistically equivalent models. No scoring function has yet been proven to have this desirable property for staged trees. In this paper, we present a novel Bayesian Dirichlet scoring function based on path uniformity and mass conversation, and prove that this new scoring function is score-equivalent for staged trees. △ Less

Submitted 17 January, 2023; v1 submitted 30 June, 2022; originally announced June 2022.

Building A Bayesian Decision Support System for Evaluating COVID-19 Countermeasure Strategies

Authors: Peter Strong, Aditi Shenvi, Xuewen Yu, K. Nadia Papamichail, Henry P Wynn, Jim Q Smith

Abstract: Decision making in the face of a disaster requires the consideration of several complex factors. In such cases, Bayesian multi-criteria decision analysis provides a framework for decision making. In this paper, we present how to construct a multi-attribute decision support system for choosing between countermeasure strategies, such as lockdowns, designed to mitigate the effects of COVID-19. Such a… ▽ More Decision making in the face of a disaster requires the consideration of several complex factors. In such cases, Bayesian multi-criteria decision analysis provides a framework for decision making. In this paper, we present how to construct a multi-attribute decision support system for choosing between countermeasure strategies, such as lockdowns, designed to mitigate the effects of COVID-19. Such an analysis can evaluate both the short term and long term efficacy of various candidate countermeasures. The expected utility scores of a countermeasure strategy capture the expected impact of the policies on health outcomes and other measures of population well-being. The broad methodologies we use here have been established for some time. However, this application has many novel elements to it: the pervasive uncertainty of the science; the necessary dynamic shifts between regimes within each candidate suite of countermeasures; and the fast moving stochastic development of the underlying threat all present new challenges to this domain. Our methodology is illustrated by demonstrating in a simplified example how the efficacy of various strategies can be formally compared through balancing impacts of countermeasures, not only on the short term (e.g. COVID-19 deaths) but the medium to long term effects on the population (e.g increased poverty). △ Less

Submitted 18 December, 2021; v1 submitted 12 January, 2021; originally announced January 2021.

Comments: 32 pages including supplementary material

A Bayesian decision support system for counteracting activities of terrorist groups

Authors: Aditi Shenvi, F. Oliver Bunnin, Jim Q. Smith

Abstract: Activities of terrorist groups present a serious threat to the security and well-being of the general public. Counterterrorism authorities aim to identify and frustrate the plans of terrorist groups before they are put into action. Whilst the activities of terrorist groups are likely to be hidden and disguised, the members of such groups need to communicate and coordinate to organise their activit… ▽ More Activities of terrorist groups present a serious threat to the security and well-being of the general public. Counterterrorism authorities aim to identify and frustrate the plans of terrorist groups before they are put into action. Whilst the activities of terrorist groups are likely to be hidden and disguised, the members of such groups need to communicate and coordinate to organise their activities. Such observable behaviour and communications data can be utilised by the authorities to estimate the threat posed by a terrorist group. However, to be credible, any such statistical model needs to fold in the level of threat posed by each member of the group. Unlike in other benign forms of social networks, considering the members of terrorist groups as exchangeable gives an incomplete picture of the combined capacity of the group to do harm. Here we develop a Bayesian integrating decision support system that can bring together information relating to each of the members of a terrorist group as well as the combined activities of the group. △ Less

Submitted 16 December, 2021; v1 submitted 8 July, 2020; originally announced July 2020.

Propagation for Dynamic Continuous Time Chain Event Graphs

Authors: Aditi Shenvi, Jim Q. Smith

Abstract: Chain Event Graphs (CEGs) are a family of event-based graphical models that represent context-specific conditional independences typically exhibited by asymmetric state space problems. The class of continuous time dynamic CEGs (CT-DCEGs) provides a factored representation of longitudinally evolving trajectories of a process in continuous time. Temporal evidence in a CT-DCEG introduces dependence b… ▽ More Chain Event Graphs (CEGs) are a family of event-based graphical models that represent context-specific conditional independences typically exhibited by asymmetric state space problems. The class of continuous time dynamic CEGs (CT-DCEGs) provides a factored representation of longitudinally evolving trajectories of a process in continuous time. Temporal evidence in a CT-DCEG introduces dependence between its transition and holding time distributions. We present a tractable exact inferential scheme analogous to the scheme in Kjærulff (1992) for discrete Dynamic Bayesian Networks (DBNs) which employs standard junction tree inference by "unrolling" the DBN. To enable this scheme, we present an extension of the standard CEG propagation algorithm (Thwaites et al., 2008). Interestingly, the CT-DCEG benefits from simplification of its graph on observing compatible evidence while preserving the still relevant symmetries within the asymmetric network. Our results indicate that the CT-DCEG is preferred to DBNs and continuous time BNs under contexts involving significant asymmetry and a natural total ordering of the process evolution. △ Less

Submitted 29 June, 2020; originally announced June 2020.

Constructing a Chain Event Graph from a Staged Tree

Authors: Aditi Shenvi, Jim Q. Smith

Abstract: Chain Event Graphs (CEGs) are a recent family of probabilistic graphical models - a generalisation of Bayesian Networks - providing an explicit representation of structural zeros, structural missing values and context-specific conditional independences within their graph topology. A CEG is constructed from an event tree through a sequence of transformations beginning with the colouring of the vert… ▽ More Chain Event Graphs (CEGs) are a recent family of probabilistic graphical models - a generalisation of Bayesian Networks - providing an explicit representation of structural zeros, structural missing values and context-specific conditional independences within their graph topology. A CEG is constructed from an event tree through a sequence of transformations beginning with the colouring of the vertices of the event tree to identify one-step transition symmetries. This coloured event tree, also known as a staged tree, is the output of the learning algorithms used for this family. Surprisingly, no general algorithm has yet been devised that automatically transforms any staged tree into a CEG representation. In this paper we provide a simple iterative backward algorithm for this transformation. Additionally, we show that no information is lost from transforming a staged tree into a CEG. Finally, we demonstrate that with an optimal stopping criterion, our algorithm is more efficient than the generalisation of a special case presented in Silander and Leong (2013). We also provide Python code using this algorithm to obtain a CEG from any staged tree along with the functionality to add edges with sampling zeros. △ Less

Submitted 16 December, 2021; v1 submitted 29 June, 2020; originally announced June 2020.

Journal ref: Proceedings of Machine Learning Research 138:437-448 (PGM 2020)

A Bayesian Dynamic Graphical Model for Recurrent Events in Public Health

Authors: Aditi Shenvi, Jim Q. Smith

Abstract: To analyze the impacts of certain types of public health interventions we need to estimate the treatment effects and outcomes as these apply to heterogeneous open populations. Dynamically modifying populations containing risk groups that can react very differently to changes in covariates are inferentially challenging. Here we propose a novel Bayesian graphical model called the Reduced Dynamic Cha… ▽ More To analyze the impacts of certain types of public health interventions we need to estimate the treatment effects and outcomes as these apply to heterogeneous open populations. Dynamically modifying populations containing risk groups that can react very differently to changes in covariates are inferentially challenging. Here we propose a novel Bayesian graphical model called the Reduced Dynamic Chain Event Graph (RDCEG) customized to such populations. These models generalize the tree-based Chain Event Graphs to a particular class of graphically supported semi-Markov processes. They provide an interface between natural language explanations about what might be happening to individuals and a formal statistical analysis. Here we show how the RDCEG is able to express the different possible progressions of each vulnerable individual as well as hypotheses about probabilistic symmetries within these progressions across different individuals within that population. We demonstrate how well-developed Bayesian Network technologies can be transferred almost seamlessly to this class. Our work is motivated by the challenge of modeling non-pharmacological interventions for recurrent event processes. We illustrate our methodology in two settings: an intervention to reduce falls among the elderly and a trial to examine effects of deferred treatment among individuals presenting with early epilepsy. △ Less

Submitted 11 October, 2019; v1 submitted 21 November, 2018; originally announced November 2018.

Comments: 50 pages including supplementary materials