Twenty20Insight
Lead Research Organisation:
Durham University
Department Name: Computer Science
Abstract
As digital technology permeates every area of modern life, we risk becoming over-dependent on complex systems that operate in an opaque way, creating a risk that they exhibit emergent properties that adversely affect their users or their wider environment. This is particularly true as developers increasingly rely on AI or ML techniques as a means to define system behaviour when the problem space is too complex or poorly understood for human developers to explicitly specify that behaviour. We are tackling incompletely understood problems by developing systems whose behaviour and wider impact are by necessity also incompletely understood. This trend, which is largely enabled by an abundance of data harvested from (e.g.) mobile devices, sensors and social media, is radically changing how systems are developed and how they are used. We need a new approach to software engineering that
(i) places greater emphasis on making explicit the risks of unintended behaviour for innovative new software products either through limitations on our understanding of the envisioned product's behaviour or through misuse, and
(ii) actively supports explainability of the exposed behaviour by the running system.
Twenty20Insight is an interdisciplinary project bringing together academic experts in Software Engineering (SE), RE, Design Thinking and ML to help system stakeholders and developers understand and reason about the impact of intelligent systems on the world in which they operate.
(i) places greater emphasis on making explicit the risks of unintended behaviour for innovative new software products either through limitations on our understanding of the envisioned product's behaviour or through misuse, and
(ii) actively supports explainability of the exposed behaviour by the running system.
Twenty20Insight is an interdisciplinary project bringing together academic experts in Software Engineering (SE), RE, Design Thinking and ML to help system stakeholders and developers understand and reason about the impact of intelligent systems on the world in which they operate.
Planned Impact
We expect the project to have a major impact on five classes of beneficiaries beyond the academic research community:
Industrial partners. Twenty20Insight will have an early impact on our industrial partners, with whom we will work to advance the state-of-the-art in stakeholder engagement for the co-design of intelligent systems, and techniques for reasoning about uncertainty and making ML systems' decision-making more transparent.
Early career researchers. The project will provide the attached PDRAs with advanced training in research methods, publication and career development, which will eventually develop the PDRAs into independent researchers by the end of the project.
Research community. Twenty20Insight will develop fundamental new knowledge about engineering intelligent systems both within and at the boundaries between the design, RE/SE, AI and ML research communities.
Industries revolutionised by AI. AI/ML can be applied to nearly every industrial sector. Designing the increasingly complex AI systems requires stakeholders to understand both the Horizon of Possibilities and Envelope of Acceptability when applying ML to their problems. This project has a great potential to tackle such a problem and thus will significantly impact all the industries revolutionised by AI.
Software development companies. The new tools and techniques that will be developed in this project will have the potential to enhance software development companies to meet the increasing demand for intelligent systems operating in an uncertain world.
Wider society. The principal beneficiary of initiatives that aim to generate transformative ideas in order to advance technology, such as the Twenty20Insight project, is ultimately the wider society. This is particularly true of Twenty20Insight since one our explicit aims is to discover techniques to better understand systems' societal impact and enable stakeholders to handle such knowledge in a systematic and transparent way.
Industrial partners. Twenty20Insight will have an early impact on our industrial partners, with whom we will work to advance the state-of-the-art in stakeholder engagement for the co-design of intelligent systems, and techniques for reasoning about uncertainty and making ML systems' decision-making more transparent.
Early career researchers. The project will provide the attached PDRAs with advanced training in research methods, publication and career development, which will eventually develop the PDRAs into independent researchers by the end of the project.
Research community. Twenty20Insight will develop fundamental new knowledge about engineering intelligent systems both within and at the boundaries between the design, RE/SE, AI and ML research communities.
Industries revolutionised by AI. AI/ML can be applied to nearly every industrial sector. Designing the increasingly complex AI systems requires stakeholders to understand both the Horizon of Possibilities and Envelope of Acceptability when applying ML to their problems. This project has a great potential to tackle such a problem and thus will significantly impact all the industries revolutionised by AI.
Software development companies. The new tools and techniques that will be developed in this project will have the potential to enhance software development companies to meet the increasing demand for intelligent systems operating in an uncertain world.
Wider society. The principal beneficiary of initiatives that aim to generate transformative ideas in order to advance technology, such as the Twenty20Insight project, is ultimately the wider society. This is particularly true of Twenty20Insight since one our explicit aims is to discover techniques to better understand systems' societal impact and enable stakeholders to handle such knowledge in a systematic and transparent way.
Publications
Combemale B
(2024)
Declarative Lifecycle Management in Digital Twins
Cámara J
(2022)
The uncertainty interaction problem in self-adaptive systems
in Software and Systems Modeling
Darby A
(2023)
Approaching Unanticipated Consequences
Garcia L
(2024)
Decision Making for Self-Adaptation Based on Partially Observable Satisfaction of Non-Functional Requirements
in ACM Transactions on Autonomous and Adaptive Systems
Kühn T
(2022)
History-aware explanations
Parra-Ullauri J
(2023)
A Framework for History-Aware Hyperparameter Optimisation in Reinforcement Learning
Parra-Ullauri J
(2021)
Event-driven temporal models for explanations - ETeMoX: explaining reinforcement learning
in Software and Systems Modeling
Related Projects
| Project Reference | Relationship | Related To | Start | End | Award Value |
|---|---|---|---|---|---|
| EP/T017627/1 | 31/08/2020 | 30/08/2021 | £586,520 | ||
| EP/T017627/2 | Transfer | EP/T017627/1 | 30/09/2021 | 13/07/2024 | £390,864 |
| Description | We have the following achievements and findings: Further techniques for quantifying the impact of values on requirements and software engineering (O1, WP2, WP3). For the above, we have added new publications: 2 Conference Publications: - "History-aware explanations: towards enabling human-in-the-loop in self-adaptive systems", Juan Parra-Ullauri, Antonio García-Domínguez, Nelly Bencomo, Luis Garcia-Paucar/ Book Proceedings of the 25th International Conference on Model Driven Engineering Languages and Systems: Companion Proceedings, MODELS 2022 - "The Implications of 'Soft'Requirements", Alistair Sutcliffe, Pete Sawyer, Nelly Bencomo Conference 2022 IEEE 30th International Requirements Engineering Conference (RE) 178-188, IEEE 1 Journal paper (in SoSym) to support runtime awareness about the levels of satisfaction of quality properties and measure degrees of uncertainty to support autonomous decision-making (O2, O3, WP3, WP4, WP5) - "The uncertainty interaction problem in self-adaptive systems", Javier Cámara, Javier Troya, Antonio Vallecillo, Nelly Bencomo, Radu Calinescu, Betty HC Cheng, David Garlan, Bradley Schmerl, 2022, Journal Software and Systems Modeling, 21,. 4 1277-1294, Springer Berlin Heidelberg Further steps towards incorporating speculative techniques in Requirements Engineering to better anticipate unwanted emergent behaviour in uncertain environments. (O1, WP1, WP2). One further workshop was delivered to the industry partner (BT) and students and colleagues at Durham University. Based on the results, we have 1 Conference paper under submission. |
| Exploitation Route | Two research proposals are under submission, and two are being prepared to follow the work of "towards incorporating speculative techniques in Requirements Engineering and Software Engineering" and decision-making under uncertainty for Digital Health. |
| Sectors | Digital/Communication/Information Technologies (including Software) Healthcare |
| URL | https://aihs.webspace.durham.ac.uk/Twenty20insight/ |
| Description | We continued to take the first steps towards incorporating speculative techniques in Requirements Engineering to better anticipate unwanted emergent behaviour in uncertain environments. The process has been used to deliver two new workshops on using Digital Twins with our Industry partner (BT) and with researchers at Durham University. We have submitted a research paper to a Conference and have applied for further funding taking our work to be applied on the domain of Digital Health. We have been granted a new research proposal called EPSRC IAA: WeDecide: Clinical Tool For Shared Decision-Making For Treatment Of Menopause Symptoms, which included the NHS as a partner. weDecide is based on outcomes of the EPSRC project Twenty20Insight[*]. We also got three small the EPSRC seedcorn projects ReqModAI, iDecide and weDecide (with the NHS as a partner), led by Dr Bencomo. These projects focussed on developing AI-based techniques for autonomous decision-making under uncertainty. Crucially, iDecide allowed us to make initial steps towards using AI for Personalised and Shared Decision-Making (PSDM) [NHS1] in Digital Health (i.e. the 12-month project EPSRC IAA: WeDecide: Clinical Tool For Shared Decision-Making For Treatment Of Menopause Symptoms https://aihs.webspace.durham.ac.uk/wedecide/). A new EPSRC bid is being prepared based on the above. I have also been granted the SRF ( strategic research funding) project JusTN0W - £5M- https://www.durham.ac.uk/research/institutes-and-centres/csdlp/justn0w/. The results of Twenty20Insight have informed the proposal of JusTN0W (Just Transitions to a Net Zero World). This project just started (Jan 2025) so next year's results will be updated accordingly. |
| First Year Of Impact | 2024 |
| Sector | Digital/Communication/Information Technologies (including Software),Healthcare |
| Impact Types | Societal Policy & public services |
| Description | Requirements Models for Artificial Intelligence (ReqModAI): Framework and Case study |
| Amount | £4,991 (GBP) |
| Funding ID | RF090115 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 05/2022 |
| End | 06/2022 |
| Title | RDM case study Logistic Regression to support Digital Twins Fidelity |
| Description | RDM case study Logistic Regression to support Digital Twins Fidelity |
| Type Of Material | Database/Collection of data |
| Year Produced | 2022 |
| Provided To Others? | Yes |
| Impact | The state of a self-adaptive system (SAS) can be described using the quality properties that it needs to satisfy (i.e. its quality objectives). However, the states of these quality properties are often not directly observable due to uncertain environmental contexts. Instead, a belief over these hidden states can be maintained. Partially Observable Markov Decision Processes (POMDPs) provide a principled approach to decision-making for SASs as they seek to satisfy their, often competing, quality objectives. Here, beliefs in POMDPs can represent the probabilities of the SAS's quality objectives being met. However, how accurately do these beliefs reflect the real but hidden states of the quality objectives? In this paper, we focus on answering this question, as poor accuracy would inevitably lead to poor decision-making. We evaluate the beliefs for both single- objective (with single reward) and multi-objective (with multiple rewards) POMDPs to underpin the decision-making in two case studies, in the networking and Internet of Things (IoT) domains. Logistic Regression is used to estimate the fidelity (i.e. accuracy) of these beliefs when reflecting the hidden state of the SASs in terms of the satisfaction of the quality objectives. Our results provide strong evidence that beliefs in POMDPs offer a good representation of the state of the quality objectives in SASs. |
| URL | https://gitlab.com/re_research/rdmcasestudylogisticregression/ |
| Description | EPSRC IAA: WeDecide: Clinical Tool For Shared Decision-Making For Treatment Of Menopause Symptoms |
| Organisation | NHS England |
| Country | United Kingdom |
| Sector | Public |
| PI Contribution | 10 per cent of women leave the workforce due to menopause, while one in four considers leaving. The weDecide project aims to evaluate a novel technique to support personalised and shared decision-making (PSDM) for menopause treatment that considers not just the criteria of a menopause expert but also that of the woman experiencing the peri-menopausal symptoms. The technique builds on the decision-making AI-based tool developed in the EPSRC project Twenty20Insight, which considers the priorities/preferences of humans. weDecide is based on outcomes of the EPSRC project Twenty20Insight[*], and the EPSRC seedcorn projects ReqModAI[**] and iDecide[***], led by Dr Bencomo. These projects focussed on developing AI-based techniques for autonomous decision-making under uncertainty. Crucially, iDecide allowed us to make initial steps towards using AI for Personalised and Shared Decision-Making (PSDM) [NHS1] in Digital Health. |
| Collaborator Contribution | weDecide focuses on assessing the clinical relevance and acceptability of the decisions prompted by the weDecide Tool to inform an EPSRC bid. The NHS GP offers us her expertise in Menopause. |
| Impact | No outputs have been delivered yet as the collaboration has just started. |
| Start Year | 2024 |
| Description | SRF: Just Transitions To A Net Zero World (Strategic Research Funding Project funded by Durham) |
| Organisation | Durham University |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | https://www.durham.ac.uk/research/institutes-and-centres/csdlp/justn0w/ As part of this collaboration, my contributions are primarily centered on shaping the research agenda, attracting key talent, and fostering interdisciplinary integration between Computer Science and Law (plus Bioscience and Economics). My previous work in EPSRC Twenty20Insight is now informing the research direction and methodologies applied in JusTN0W. 1. JusTN0W - Just Transitions to a Net Zero World My role has been instrumental in defining the AI-driven decision-making methodologies that will support sustainable transitions. I have been actively involved in recruiting and building a team of Computer Science researchers who will apply their expertise in AI, optimization, and uncertainty modeling to real-world legal and environmental challenges. The project aims to bridge the gap between computational methods and legal frameworks, ensuring that policy and regulatory decisions benefit from data-driven insights. The methodologies developed in Twenty20Insight-particularly in decision-making under uncertainty and AI-assisted analytics-are now being adapted and expanded within JusTN0W to address sustainability policy and regulation. 2. EPSRC Twenty20Insight Influence on JusTN0W Twenty20Insight, which concluded last year, provided foundational research on AI-driven decision-making under uncertainty, particularly in risk-aware autonomous systems. The insights from this project are now being applied to sustainability challenges, where legal and policy decisions require robust, data-driven support. The computational tools and uncertainty quantification techniques developed in Twenty20Insight are being refined to align with the needs of legal frameworks, contributing to JusTN0W's efforts in just transitions and Net Zero strategies. Through this collaboration, we are hiring and integrating Computer Science researchers into legal and environmental applications, facilitating a cross-disciplinary knowledge exchange that will drive JusTN0W forward. The focus ahead is on building computational models, refining decision-support tools, and ensuring AI applications align with legal and policy considerations for sustainable decision-making. |
| Collaborator Contribution | 2. EPSRC Twenty20Insight Influence on JusTN0W Twenty20Insight, which concluded last year, provided foundational research on AI-driven decision-making under uncertainty, particularly in risk-aware autonomous systems. The insights from this project are now being applied to sustainability challenges, where legal and policy decisions require robust, data-driven support. The computational tools and uncertainty quantification techniques developed in Twenty20Insight are being refined to align with the needs of legal frameworks, contributing to JusTN0W's efforts in just transitions and Net Zero strategies. Through this collaboration, we are hiring and integrating Computer Science researchers into legal and environmental applications, facilitating a cross-disciplinary knowledge exchange that will drive JusTN0W forward. The focus ahead is on building computational models, refining decision-support tools, and ensuring AI applications align with legal and policy considerations for sustainable decision-making. |
| Impact | Not yet. |
| Start Year | 2023 |
| Description | 3rd Winter Modeling Meeting San Vigilio di Marebbe, Italy - February 4-11 2023 |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Other audiences |
| Results and Impact | The scope of the meeting was to bring together world-class experts in modeling, model-based engineering, model-driven engineering, and language engineering (briefly M-ALL) to examine the impediments that prevent M-ALL from being more prominently considered by students, researchers, and practitioners. M-ALL has been successfully applied in many software and system engineering areas and now addresses even broader and novel scenarios, e.g., digital twins and analytics, in an increasing number of application domains, including natural and multi-physics systems. Such new perspectives challenge both the seminal, yet limited, foundations, e.g., conformity and instantiation, on which most of the tools and methods are based, as well as the way of teaching M-ALL to possibly heterogeneous communities (from software engineers to domain experts and citizen developers). During the meeting, we discussed, defined and formalized the new M-ALL foundations required to address the aforementioned upcoming scenarios. I presented the work done with my researchers in Twenty20Insight about the Fidelity of DTs and Models@run.time. A paper is under developement and will be submitted to a Journal in Summer. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://wmm23.notion.site/ |
| Description | Human-Machine Teaming for Shared Decision-Making under Uncertainty |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | A talk presented in the Turing Workshop on Human-Centric Artificial and Computational Intelligence organised by Prod Federica Sarro (UCL) and Dr Ke Li (Exeter). A EPSRC bid is being led by me with Dr K Li (Exeter) and Prof Yulan He (KCL) and it was triggered by that presentation. |
| Year(s) Of Engagement Activity | 2023 |
| URL | https://colalab.ai/turing_workshop/ |
| Description | Invitation to talk at Dagstuhl Seminar 22362 Model-Driven Engineering of Digital Twins ( Sep 04 - Sep 09, 2022 ), Germany |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Other audiences |
| Results and Impact | The intended primary goal of the seminar was to create a community and establish a research roadmap; we discussed the following topics: - Challenges faced in the real-world development of Digital Twins (DT). - Opportunities offered by MDE. - Active exploration of collaboration opportunities. A key outcome of the seminar was a research roadmap for the new software engineering discipline for Digital Twins (DT). I presented the work done with my researcher Huma Samin in Twenty20Insight about the Fidelity of DT. A paper has been developed and will be submitted to a Journal soon. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.dagstuhl.de/en/seminars/seminar-calendar/seminar-details/22362 |
| Description | MDENet's Annual Symposium 2022. https://mde-network.com/annual-symposium-dec-22-recordings/ |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Dr Huma Samin and I presented the work done in Twenty20Insight about ReqModAI Requirements Models for Artificial Intelligence: Framework and Casestudy (WP3). |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://mde-network.com/annual-symposium-dec-22-recordings/ |
| Description | Modelling Human-Machine Teaming for Shared Decision-Making under Uncertainty |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Keynote given in the Modellierung 2024 Cinference in Germany. https://scholar.google.co.uk/citations?view_op=view_citation&hl=en&user=86H7HmkAAAAJ&sortby=pubdate&citation_for_view=86H7HmkAAAAJ:1lhNe0rCu4AC |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://bpt.hpi.uni-potsdam.de/modellierung2024/keynotes/ |
| Description | Participation in an activity, workshop or similar - Seminar Model Hybridization for Digital Twins - https://www.bellairs2025.ece.mcgill.ca/schedule.htm - Feb 2025 |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | The 1st 1-Week Workshop on Model Hybridization for Digital Twins at the Bellairs Research Institute of McGill University in Holetown, Barbados, is taking place from Friday, February 21st, 2025 to Friday, February 28th, 2025 with a special emphasis on the use of different forms of models (e.g., analytical models and machine learning models) in the context of digital twins. Digital twins promise tremendous potential for gaining insights, optimize operations, and improve decision-making for cyber-physical systems across various industries, including manufacturing, healthcare, transportation, and more. Recent developments show that MDE can play a central role in systematically leveraging the potential of digital twins, and many researchers from the MDE community have applied MDE technology to build digital twins in recent years. In the future, the predictive capabilities of digital twins will be leveraged for dynamically evolving eco-systems to address challenges such as sustainability at a much more complex scale. To this aim, techniques for the coordinated use of heterogeneous descriptive and prescriptive models need to be elaborated, and the propagation of uncertainty across digital twin boundaries investigated, leading towards the definition of interfaces for DT composition and a unifying theory for inductive and deductive reasoning. The modelling-related Bellairs workshop series have traditionally had a strong focus on software composition and modularity (aspect-oriented modelling (2010, 2011, 2012, 2013), concern-oriented modelling (2015, 2017), interfaces for software reuse (2018), polyglotism (2023), digital twins composability (2024)), but also on the interplay of data and models (data and models (2019, 2020), feedback-driven decision making (2022)). We therefore want to dedicate the upcoming Bellairs workshop on investigating the hybridization of different forms of models (e.g., analytical models, architecture models, machine learning models) for digital twins. The workshop was highly interactive and open-minded, with the main objective to reflect on the combination of different forms of models in complex scenarios for digital twins. At least a publication is under way. |
| Year(s) Of Engagement Activity | 2025 |
| URL | https://www.bellairs2025.ece.mcgill.ca/schedule.htm |
| Description | Seminar 4th Winter Modeling Meeting - https://wmm25.notion.site/4th-Winter-Modeling-Meeting-120f3eb8c00f815f9a8cc3ef502b358f. Feb 2025 |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | I attended this event to contribute insights from my research in Twenty20Insight, particularly on decision-making under uncertainty and the adaptability of software systems. Modern software-based systems must continuously adapt to dynamic environments, blurring the boundary between engineering-time and execution-time. My work in Twenty20Insight explored how AI-driven decision-making can enhance temporal adaptability, allowing systems to evolve in response to changing conditions, much like human cognition. A key challenge is managing abstractions, which enable both cognitive flexibility for users and scalability for systems. Software engineering must now focus on Abstraction Engineering-developing dynamic, context-aware abstractions that balance complexity reduction with real-time adaptability. My contribution to the event highlighted how AI-driven models, uncertainty quantification, and adaptive decision-making from Twenty20Insight can inform new approaches to software adaptability, ensuring that abstractions evolve alongside changing requirements. (For a more detailed discussion, see: https://arxiv.org/pdf/2408.14074). |
| Year(s) Of Engagement Activity | 2025 |
| URL | https://wmm25.notion.site/4th-Winter-Modeling-Meeting-120f3eb8c00f815f9a8cc3ef502b358f |
| Description | keynote at Modellierung 2024 in Germany |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Models for Human-Machine Teaming for Shared Decision-Making under Uncertainty There is growing uncertainty about the runtime environment of software systems. Therefore, how the system should behave under different contexts cannot be fully predicted at design time. It is considerations such as these that have led to the development of self-adaptive systems (SAS), which can dynamically and autonomously reconfigure their behaviour to respond to changing external conditions. The use of Machine Learning (ML) and AI has exacerbated the issues by adding more uncertainty sources. The scope of the talk is in the areas of Model-driven Engineering (MDE), Requirements Engineering (RE), software engineering (SE), and the development of techniques to quantify uncertainty to improve decision-making. The explicit treatment of uncertainty by the running system improves its judgment to make decisions supported by evaluating evidence found during runtime, possibly including the human-in-the-loop. The speaker will discuss how quantification of uncertainty can improve requirement elicitation (using simulations, for example). The talk will also cover different approaches to quantifying uncertainty, models@run.time and their role in Human-Machine Teaming. |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://bpt.hpi.uni-potsdam.de/modellierung2024/keynotes/ |
| Description | keynote at the 15th International Conference on Pattern Recognition Applications and Methods (ICPRAM 2025) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Title: Bayesian Theory of Surprise to Quantify Degrees of Uncertainty Location: ICPRAM 2025 Date: 25/02/2025 In software engineering for self-adaptive systems (SASs), we are witnessing a growing intersection between Artificial Intelligence (AI) and Software Engineering (SE)-and we are just getting started! This talk is based on work in progress, which I have been developing in deep collaboration with Huma Samin and Amel BELAGGOUN and, more recently, Dylan Walton. I am incredibly grateful for their contributions and insights-this work would not be possible without them! In this talk, I will introduce a novel Bayesian definition of Surprise as a quantitative tool to measure uncertainty and deviations from expected behavior. Surprise is a powerful mechanism that helps systems determine when observed data challenges prior assumptions-enabling them to adapt accordingly or flag anomalies. We will explore applications of the Bayesian Theory of Surprise in self-adaptive systems, leveraging Bayesian Inference and Partially Observable Markov Decision Processes (POMDPs). Additionally, I will discuss different Surprise-based approaches to quantifying uncertainty, including: ? Bayesian Surprise ? Shannon Surprise ? Bayes Factor Surprise ? Digital Twins and their role in adaptive decision-making |
| Year(s) Of Engagement Activity | 2025 |
| URL | https://icpram.scitevents.org/KeynoteSpeakers.aspx?y=2025 |
| Description | webinar 'Navigating the Future with Digital Twins' in collaboration with ACM India -October 2024 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Presentation about the role of abstraction and AI in Digital Twins. in collaboration with ACM India and Tata Consultancy Services https://www.linkedin.com/posts/tcs-research_fireside-tcsresearch-inventingforimpact-activity-7241403752765648896-j3zF?utm_source=social_share_send&utm_medium=member_desktop_web&rcm=ACoAAAHyi6YB20LaL1A7YoWaLgJeeFy9u7gSLUI |
| Year(s) Of Engagement Activity | 2024 |
| URL | https://www.linkedin.com/posts/tcs-research_fireside-tcsresearch-inventingforimpact-activity-7241403... |