ARCHIVE of the abaci Library from 2015

Free PDF Resources ordered by date - most recent first

This page was formerly provided by The abaci Partnership - which closed in 2015. This library has been retained here for Internet continuity.
It is a free resource, providing readers with practical guidance applicable to AI, climate change, decision-making in the face of the unprecedented and military topics. The work exploits insights derived from complexity science.

There are supporting articles on adaptation, and on how to effect organisational change in practice, on this Complexity Demystified blog.

• Title: [Effective] Employment of Tools and Models Appropriate to Complex Real-world Situations.
• Article PDF 6.6MB
• Date: Apr 2015. Pre-publication draft of an invited contribution to LSE's Handbook of Research Methods in Complexity Science and their Application, published in the UK later that year. Following on from the various workshops, seminars and presentations below, on applications of complexity science, this article is the 'swansong' - bringing together all the themes in one place. Its focus is mainly on the applicability, and otherwise, of tools such as agent-based modelling (ABM) and its insights are relevant to the current (2023) concerns about AI / ChatGPT / LLMs (large language models). Particularly in the way that their abilities are hyped or frankly (willfully) misunderstood.

• Title: Climate Change Adaptation (CCA), an African example - compared to a Zero-carbon future for Britain.
• PDF version of PowerPoint Presentation 2.3MB
• Date: Oct 2013. Given at CAT Members Conference 2013 at the Centre for Alternative Technology in Wales, UK. This presentation explained to the CAT Members that many African countries were ahead of the UK, and Europe, in making preparations for climate change. It also described some of the measures being introduced.

• Title: Information Superiority to Information Advantage.
• Doctrine Note (PDF File) 645KB
• Date: Jul 2013. Our Research Director was co-author - bringing a commonsense pragmatism to a controversial topic. As a result, in 2018 Joint Concept Note JCN 2/18 was issued under the new concept of 'Information Advantage'. This approach requiring active endeavours to gain advantage in the face of enemy action, not just passive collection of 'more stuff'.

• Title: Insights from Complexity Science applied to Climate Change Adaptation (CCA) challenges.
• PDF version of PowerPoint presentation 4.7MB
• Date: Jul 2013. For the Oxfam Complexity Group in Oxford, UK - Oxfam website. This presentation described the way in which insights from complexity science had been applied in Africa to help communities make preparations for climate change.

• Title: Narratives and Posters about different ways that people adapt to climate change in Africa.
• Leaflet (PDF File) 242KB and Posters (PDF) 2.5MB
• Date: Mar 2013. In support of the Africa Climate Change Resilience Alliance (ACCRA) . These narratives compare different types of approach to the climate emergency - from just reacting passively (unwise) to actively anticipating (much better). They were used in community learning activities in Ethiopia, Mozambique and Uganda.

• Title: Adapting to a Complex and Uncertain Future - lessons for Climate Change Development Aid Work.
• Paper (PDF File) 1.8MB
• Date: Feb 2013. With the Overseas Development Institute, UK [With Lindsey Jones, Eva Ludi, Patrick Beautement, Christine Broenner and Carina Bachofen]. This report detailed the progress made by the end of this phase of work on climate change mitigation in Africa (Uganda, Ethiopia and Mozambique). It explains the methods used and the results achieved, as well as indicating what communities learned from the activities.

• Title: The need to Sustain 'Peak Precision' for a Zero-carbon future for Britain.
• Article (PDF File) 135KB
• PDF version of PowerPoint presentation 3.9MB
• Date: Feb 2013. This presentation was given at the CAT Members Conference in 2012. The key point being that achieving a zero-carbon future is critically, yes absolutely reliant on the kind of science, engineering and industrial / chemical industries usually so hated by alternative communities / 'eco-people'. Particularly important is maintaining the ability to make tiny objects to fine tolerances reliably and repeatedly - and this ability has a long 'logistic tail' of related skills and industries. Hand tools, permaculture and 'back to basics' is just not going to work for 8 billion people ... The talk was also published as an article in the Centre for Alternative Technology's magazine 'Clean Slate' Number 87 later that year.

• Title: A Complexity Science-based Approach to Conflict Analysis and Influence.
• PDF version of PowerPoint presentation 6.2MB
• Date: Apr 2012. IMA Conference at RMA Sandhurst, UK. abaci's Research Directory gave an invited plenary presentation. The paper shown how techniques adapted from complexity science aided the understanding of both conflict itself, and of approaches to conflict resolution, and to influencing peaceful outcomes. A case study example from Syria (2008) was examined in detail and then developed in seminar sessions later.

• Title: The Defence Enterprise is More Than Just a Supermarket Chain.
• Paper (PDF File) 330KB
• Date: Mar 2012. In RUSI Defence Systems Journal. The article discusses the increasing trend towards assuming that 'doing it the way business does' is the best approach. But this is absolutely not true for the fighting parts of military forces (nor is it suitable for the parts of the NHS which deal face-to-face with patients). The consequences of following the 'supermarket-chain' model inappropriately are dire - as the article clearly demonstrates. In the case of the military it means predictability on the battlefield, it means that you lose ... horribly.

• Title: Outcomes from the workshop 'Putting Complexity to Work - Supporting the Practitioners': Implications for Health Care.
• Paper (PDF File) 440KB
• Date: Dec 2011. In Journal of Evaluation in Clinical Practice. The insights from this workshop are that complexity science does provide practical solutions - but that implementing them is not straightforward. This is because practitioners face many organisational barriers (vested interests, 'not-invented-here' resistance and plain stubbornness to change). The workshop identified many good strategies for making progress despite these hurdles and these are listed in the article. [With Christine Broenner].

• Title: Complexity Demystified - A Guide for Practitioners - full text.
• Summary Paper (PDF File) 220KB
• Full text of the book - the pre-publication version (PDF File) 9.8MB
• Date: May 2011. The authors use a new approach to reveal why, instead of fixed bureaucratic processes and 'call-centre / supermarket-chain thinking', flexible, common sense human-centred practice and actions are required in real-world endeavours. Not only does the book point out how and why certain ways-of-working can go wrong, it also provides pragmatic, hands-on solutions for dealing appropriately and effectively with open-ended and often unpredictable situations in practice.
  The book contains case studies, showing these techniques in action, for: Fuel protests in the UK in 2000; Upgrading Slum Housing in Cameroon; dealing with Social Housing in Ludlow, UK; the challenges of Journalism in Lebanon; how the Deep Water Horizon (DWH) Oil Rig crisis was inevitable; the difficulties of community engagement in the case of 'Lost Town' at Dunwich - an East of England Development Agency project; and the book then critiques the effectiveness, and otherwise, of the Approach. [Co-author Christine Broenner].

• Title: Making Complexity Work in Practice.
• Paper (PDF File) 1.7MB
• Date: Feb 2011. Invited speaker at the Workshop 'Understanding and Influencing Causality of Change in Socio-Technical Systems' in Gold Coast, Australia. The workshop organiser invited Nobel Laureates, leading thinkers and practitioners involved in bringing about change in our complex world by putting that complexity 'to work'. The aim being to identify common themes for further investigation. A book called 'Understanding and Managing Causality of Change in Socio-Technical Systems' was published by IOS in 2012.

• Title: Outcomes from the Workshop: 'Putting Complexity to Work - Supporting the Practitioners'.
• Paper (PDF File) 740KB
• Date: Dec 2009. From ECCS'09 - European Conference on Complex Systems, 2009, Warwick, UK. This White Paper presents the insights and recommendations arising from a Workshop held on the 24th September 2009 at Warwick University, UK as part of the European Conference on Complex Systems (ECCS'09). The Workshop was called "Putting Complexity to Work - Supporting the Practitioners" and its aim was to understand how to harness more effectively the insights coming out of complexity science. Particularly to provide practical, relevant support to practitioners - the people tasked with make a difference in their day-to-day work (such as those in the medical profession, in social services, enterprise and business and, of course, politicians). [With Christine Broenner].

• Title: Putting Complexity and Causality to Work - A Practical Analysis.
• Presentation (PDF File) 5.7MB
• Date: 27 Nov 2009. At S4C - Causality in Complex Systems, Workshop, Paris. The aim of the workshop was to focus on the synthesis of ideas and experience to prepare the way for 'practice' - so that people can put causality 'to work'. First the participants had to acknowledge the practical realities within our synthesis - the givens and constraints of human activity in the real world. This presentation then explained how a systematic analytical approach had been developed and how it could be used so that practitioners can 'do things differently' and more effectively in the face of the uncertain and unexpected. For the complexity science community, a great deal has been found out about causality, but, what is its 'value added' contribution going to be? Finally, an assessment is made as to how near our insights are to being 'Real-world-Ready' - and what we should be doing to bridge the gap. [With Christine Broenner].

• Title: Is Agent-based Modelling 'Real-world-ready'? - A Systematic Analysis.
• Presentation (PDF File) 4.4MB
• Date: 25 Nov 2009. At UCL Global System Dynamics (GSD) Workshop. The aim of this discussion is to examine the assertion that Agent-based Modelling (ABM) is not yet 'real-world ready' and identify why not. As a result, the attendees considered what may have to be 'done differently' if the assertion proves to be justified. The context for the presentation was the UK's critical National Infrastructure for the 21st Century. It was taken as a given that those attending all knew that 'All models are wrong but that some are useful', and that no one disputes the vital relevance of Gödel's 'Incompleteness Theorem'.

• Title: Putting Complexity to Work - achieving effective Human-machine Teaming.
• Paper (PDF File) 227k
• Date: Sep 2009. At ECCS'09 - the European Conference on Complex Systems, 2009, Warwick, UK. When people work with a computer, or any digital device, they are in effect inviting the machine to become part of their team - so-called 'Human-machine Teaming'. But just how good are machines at being 'team players' in human endeavours? The answer is, awful! Even devices like 'Alexa', 'Siri' and others are poor because they have no conception of the social world of which they have become part. Most importantly, they have no way of even sensing, let alone understanding, the real-world consequences of the advice they offer. The article makes some suggestions as to how this can be improved - Step One being to recognise how bad the devices are at being team members and that they can even cause damage!

• Title: Complex Multi-modal Multi-level Influence Networks - Affordable Housing Case Study.
• Paper (PDF File) 128k
• Date: Feb 2009. Presented at the "Complex 09" conference in Shanghai, China. Most influence network models are depicted as nodes and links operating in the manner of a feed-forward neural network where both nodes and links appear to be homogenous in their nature (as in the Large Language Models, AKA 'AI', much in the news in 2023). Experience has shown that not only do these networks fail to deal adequately with reality, but also that programmers and practitioners struggle to understand why. This paper addresses this challenge by examining the rich, multi-level and multi-modal nature of influence networks and proposes an approach drawing inspiration from complexity science. Importantly, multiperspective techniques are needed which enable influence networks to be used more effectively to capture, visualise and understand complex situations - so providing insights to support effective decision-making. The paper gives evidence (from a case study looking at the provision of affordable housing in the UK) which illustrates how the techniques have been employed and what benefits were then realised.

• Title: Complex Phenomena in Orchestras - Metaphors for Leadership and Enterprise.
• Paper (PDF File) 58k
• Date: Feb 2009. Presented at the "Complex 09" conference in Shanghai, China by Patrick Beautement and Christine Broenner. This paper recognises the comparisons that can be made between the role of the conductor of an orchestra and the leaders of large enterprises. It also notes that little has been done to show how the complex dynamics of orchestras can provide metaphors for transformational and / or evolutionary behaviour in complex enterprises (such as the government, the military and in social settings. The paper identifies some of the dynamic musical patterns and phenomena that exist in orchestras and shows how these can provide insights applicable to other domains - especially situations where similar complex federated structures emerge 'on-the-fly'. To do this, the paper provides a complexity-inspired analytical framework.

• Title: Impact of Cyberspace on the Nature of Command - C2 and Cyberwar.
• PDF version of Powerpoint presentation 2.2MB
• Date: Jun 2008. Invited presentation - given to the Portuguese Military Academy. Relevant to the Dec 2016 stand-off between Obama's USA and Putin's Russia. Did Russia hack the USA's voting machines resulting in Donald J Trump being President Elect? If so, what is the scope of possible cyberwar responses? This presentation examines to what extent any world power can have 'Command and Control (C2) of Cyberspace'? Of course, they can't. So what does that mean for the future of conflict - especially when it is a state-sponsored one facing non-state actors?

• Title: Complex Systems Engineering with the DART Framework.
• Presentation (PDF) 1.2MB
• Date: Jul 2007. At Complex 07, Brisbane, Australia. The paper discusses a new way of designing, building and employing systems such that they are, by intention, adaptable dynamically 'on-the-fly'. It shows how most military systems are designed against a fixed specification which does not include, at design time, the idea that users may adapt the system in the field. Yet this is exactly what personnel do, by hacking the systems, in the face of the uncertainties of conflict. The DART Framework (Design, Assemble, run-Time) is an approach to enabling this behaviour from project conception. [With Anthony Alston and Lorraine Dodd].

• Title: Value-driven Reasoning Software Agents - Challenges and Issues.
• Presentation (PDF File) 550KB
• Date: Mar 2007. To a Staff Seminar at the Institute of Human and Machine Cognition (IHMC, University of West Florida UWF), USA. The key issue in using machines / digital devices, 'AI' as partners in human enterprises is the extent to which they can appreciate and understand the kind of value measures that people use. This is more than just considering ethical and legal issues - devices need to be able to reason about what different human beings think is 'good' or 'bad' or 'better' or 'worse' - and why the people have those value judgements. This discussion looks at where common ground can, and cannot, be developed between humans and the devices that work with them - especially where there is ambiguity and 'grey areas'.

• Title: Run-time Science as a Route to Exploiting Emergent Phenomena.
• Presentation (PDF) 3.15MB
• Date: Jun 2006. At 1st International Workshop on Engineering Emergence for Autonomic Systems (IWEEAS), Dublin, Ireland. So-called 'emergent phenomena' are things which happen which seem to have no obvious mechanism. Think of the swirling flight of flocks of starling or shoals of fish. They look almost alive, like a single entity - yet no one creature seems to be in charge. In the human situation we have crowds and mobs, 'going viral' and other phenomena that the come out of the blue. What if we could make these things happen purposefully to achieve particular outcomes? This presentation explores that question. [With Anthony Alston and Lorraine Dodd]

• Title: Agile and Adaptive Coalition Operations - Leveraging the Power of Complex Environments.
• Article (PDF) 116k
• Date: Apr 2006. At CCRTS, San Diego, USA. All life is a competition for the control and use of resources and the world of military endeavour is no different. That competition takes place an environment - our world - which is complex and challenging. This paper offers strategies for not only making those challenges tractable but also for turning them to advantage. Throughout the worlds of the military, commerce and government and in everyday life we are increasingly connected through complex networks of relationships and interactions, which can be represented at different levels of abstraction. For coalition organisations, the key word is agility - being able to adapt to the uncertainties of the world without dislocation. This as true for the cyberspace that supports people as it is to the real world that they inhabit. They must be capable of engaging in continuous and innovative adjustment. This paper considers some of the mechanisms which can be exploited to achieve decisive advantage in coalitions.

• Title: Controlling Edge Organisations - Exploiting Emergence.
• Paper (PDF) 328k
• Presentation (PDF) 1.49MB
• Date: May 2005. At 10th ICCRTS, Virginia, USA. During 2003, CCRP published "Power to the Edge" which described a new kind of organization, an 'Edge Organization' (EO), which would display exceptional agility. The paper looks at the challenges in exploiting the advantages of such an organisation in terms of the dynamic interactions between all the entities and the environment. Critically, it considers the 'run-time' properties of the artefacts, actors and interactions, as well as the dynamic adaptive mechanisms, as being the key focus of attention - as opposed to the static, design-time engineering of their parts. This paper considers how to ensure the collective, adaptive and secure behaviour of EOs and offers ideas to the research community for discussion

• Title: The Coalition Agents Experiment: A Prototype for Network-Enabled Coalition Capabilities.
• Paper (PDF) 158k
• Date: Jul 2004. Published by RUSI, in "Defence Systems Journal". When coalition partners are familiar with each other then doctrine, systems and procedures are aligned in advance - they have 'interoperability of the mind'. But in reality, as this paper explores, there are always uncertainties about exactly which capabilities will be provided by whom and about how the forces will be configured. Hence, coalition operations trigger the need for rapid, on-the-fly responses and cannot be predicated on using pre-existing co-ordinated systems - hence the need for flexible approaches, discussed in this paper, which allow capabilities to be assembled at 'run time'.

• Title: Software Agents for the Warfighter. ITAC with Bradshaw et al.
• Booklet (PDF) 7.2MB
• Date: October 2002. In the future software agents (AI bots, grids, personal assistants, robots and others) will be working with military forces in a supporting role. Infrastructure for these software agents will consist of services and tools used by the agents as well as by their developers and users. The infrastructure needs to address: security, scalability, robustness / fault tolerance, effectiveness, accessibility, and interoperability. While somewhat similar to distributed systems, software agents have a number of characteristics that are unique - such as long-lived identity and state and an autonomous execution model with the ability to adapt and be proactive. Software agent infrastructure needs to address agent specific requirements although it can build upon existing research and implementations of distributed system infrastructure. Services that should be provided by the infrastructure include identity generation, authentication, registration and lookup, message transport, resource management, persistence, logging, and encryption. In additional tools to support administration and management, debugging, visualization, and deployment are needed. A number of efforts are currently underway to develop software agent infrastructure. The CoABS Grid is an AI integration framework for multiple agent systems and infrastructures being developed by DARPA and the Pentagon.

• Title: Making Agents Acceptable to People. Bradshaw et al.
• Paper (PDF) 2.4MB
• Date: May 2004. Jeff Bradshaw with members of IHMC and Patrick Beautement.

And related to this is:
• Title: Terraforming Cyberspace. Bradshaw et al, IHMC..
• Paper (PDF) 665k
• Date: 2002. In these papers Jeff and his colleagues take the view that it is not for the human world to be forced to become like cyberspace. Instead, cyberspace should be 'terraformed', to make it human-friendly. The current (2023) debates about AI, ChatGPT and other inventions makes Jeff's insights even more relevant. Instead of being absorbed into the world of cyberspace (as in William Gibson's iconic cyberpunk book 'Neuromancer', which inspired the film 'The Matrix') we should take human sovereignty seriously and subordinate the machines to our will. Otherwise we may end up like the Outlanders in E. M. Forster's story "The Machine Stops" and be made 'Homeless' against our will ...

• Title: Network-Centric Security - Approaches to Collective Run-Time Adaptation.
• Paper (PDF) 204k
• Presentation (PDF) 4.5MB
• Date: 23 Nov 2003. At the Adaptive and Resilient Security workshop held at the Santa Fe Institute, SFI, New Mexico, USA. Making any kind of computer or digital system secure is a headache for programmers, system managers and techies generally. But it is not just a technical issue - it a social, cultural and organisational one too. The challenge gets even worse on the Internet / in cyberspace because no single person or organisation has absolute control. This presentation, inspired by nature / biomimetics, and the discussions which followed, wrestled with these issues and identified a number of useful strategies to follow.

• Title: Distributed Cognition: Toward a New Foundation for Human-Computer Interaction Research. James Hollan.
• Paper (Acrobat PDF) 174k
• Date: June 2001. For human-computer interaction to advance in the new millennium we need to better understand the emerging dynamic of interaction. One in which the focus task is no longer confined to the desktop but which extends into a complex networked world of information and computer-mediated interactions. Distributed cognition provides a radical reorientation of how to think about designing and supporting human-computer interaction. Holland shows how the theory of distributed cognition has a special role to play in understanding interactions between people and technologies - for its focus has always been on whole environments: what we really do in them and how we coordinate our activity when working as human-machine teams.

• Title: Software Agents as Facilitators of Coalition Operations.
• Paper (PDF) 540k
• Presentation (PDF) 4.3MB
• Date: 19 June 2001. Software agents can be viewed as semi-autonomous entities which help people cope with the complexities of working collaboratively in a distributed information environment. This paper describes the research that DERA carried out into AI Software Agents for use in Command Systems - and at the collaborative work done with 16 partners of an international Coalition Agents Experiment called 'COAX'. Specifically, the paper aims to show that using software agent-based C2 [command and control] frameworks is a useful way of dealing with the complexity of real-world problems. For example, such as supporting agile and robust 'come as you are' Coalition operations, and enabling interoperability between legacy or previously incompatible systems. In addition, Agent-enabled 'grids' can be used to rapidly integrate a wide variety of agents and infrastructures, with domain management services structuring agent relationships, limiting their behaviours and enforcing Coalition policies.

• Title: Exploiting the Phenomena of Emergence.
• Paper (PDF) 825k
• Date: 8 May 2001. Phenomena are said to be 'emergent' when they arise from interactions among the parts of a complex 'system' in a manner that is not always apparent (when the inactive state of their components are examined). Emergence seems to create powerful positive and negative effects (often appearing as tangible phenomena which persist over time) almost out of 'nothing'. In reality, emergent phenomena are a collective property of interacting 'systems' which are manifested as higher levels of organisation, abstraction and apparently sophisticated behaviour. This paper shows that further investigation of these behaviours could provide humans with a way of exploiting the phenomena in a purposeful manner. This would enable us to influence emergence to provide useful 'tools' for citizens, governments, commerce, the military and humanitarian and other organisations.

• Title: Battlespace Digitisation - Coping with Uncertainty in the Command Process.
• Paper (PDF) 394k
• Presentation (PDF) 480k
• Date: 1 July 1999. Future military Command Processes will be different from those in use today because of increased uncertainty about the types of conflict to be faced and how they will be experienced. The design of future applications and infrastructures (to support a digitised environment) must not constrain commanders and must enable them to adapt their capabilities (both before deployment and during operation) and to evolve them in the longer term. The paper deals with a method of capturing and representing Command Processes and of mapping the Command Processes to appropriate solutions - such that neither flexibility nor the emergence of novelty are constrained. The approach has three inter-linked elements: a model of the command process (ie a model of domain problem area); a model of the application and infrastructure architectures (ie a model of potential solutions); and a methodology for mapping between the two. Overall, the paper will explain how the methodology works and show how it might be used. [With Anthony Alston].

• Title: Cost-effective Simulation of Enemy Forces through 'Effects-based' Wargaming. *Award-winning Paper at I/ITSEC 98*.
• Paper (PDF) 295k
• Presentation (PDF) 1.14M
• Date: December 1st, 1998. Winner of I/ITSEC 'Best Paper' Award. One of the biggest costs in setting up and running exercises is the provision of staff to run the 'White / Red Forces' response cells. This paper looks at a novel, but already proven, approach called 'Effects-based' Wargaming. When exercises (such as BLUE FLAG or UNION FLASH) are set up, the first task is to identify the 'training audience'. Once this has been done, the size of the white and red response cells and the amount of computer support required can be estimated. The assumption made is that for the exercise to be valid the white and red response cells and the computer simulation must simulate everything - which leads to large expensive response cells. This results in simulations which are cumbersome and inflexible to operate. The paper questions this assumption and maintains that for many exercises an 'Effects-based' wargame would be more than adequate. In 'Effects-based' wargaming all that is then required is to provide an environment around the training audience which causes the required thinking to take place. This environment is created by using STIMULATION systems which make heavy use of real-world message formats to simplify the decision-maker to simulation interface.

• Title: General Tzu's Army: OPFOR of the Future - The West's Inevitable Vulnerabilities.
• Article (PDF File) 312k
• Date: Spring 1997. By Michael R Lwin. Published in Joint Force Quarterly, original PDF here, USA. The article is a critique of the vulnerabilities of the West's command and control systems - especially when faced with so-called 'low-tech' adversaries such as ISIS / ISIL. In this fascinating and inspiring article Michael Lwin shows how Western forces, because of the way they are trained and their systems are designed and built, are inevitably and predictably vulnerable. Especially, in the way that the forces will fall prey to their own self-delusions - and therefore destruction. Opponents will therefore be able to turn the West's weaknesses against itself [as Russia's Putin is starting to do in Ukraine in 2023 by exploit divisions in NATO]. As Sun Tzu said 'Know your enemy and know that it is yourself'.

• Title: Effective Selection and Use of Conflict Simulations (wargames) for Operational Training or Campaign Analysis.
• Paper (PDF) 85k
• Presentation (PDF) 140k
• Date: November 30th, 1994. Wargames (including models/simulations) have, in general terms, been used mainly for analysis and training. The games and simulations available fall into many categories depending on the level of interaction (political to tactical) the style of 'play' and the type of execution (eg manual or automated in some way) required. This paper considers which wargame choices are available and what selection criteria should be used. Above all, it should be clearly understood why a wargame needs to be used at all - and the game selected should fulfill that need. Overall, the paper will inform readers about wargaming issues and provides methodologies for their effective selection and use.

• Title: Neural Network Theories, their Simulation and Applications (MSc Thesis).
• Paper (PPT) 243k
• Date: August 1984. The suggestions by F.H.Crick and G.Mitchison on "The functions of dream sleep" have created a lot of interest in autoassociative recall (networks which recall the whole of a stored pattern from a fragment, like a hologram), and in asynchronous, parallel (and self-organising) networks in general. In this project the associative network of the type described by Kohonen and Hopfield, the subtractive network of the type described by Dobson, and the randomly connected network implemented in the hardware WISARD system (of Professor Igor Aleksander) were implemented and studied. It was soon clear from testing the performances of the networks, that they had various complementary strengths and weaknesses. Combining the networks in various ways so that they can 'collaborate' holds out the possibility of ground-breaking emergent capabilities well beyond that of individual networks (discussed in Section 9). The ability to recognise patterns beyond human abilities is a possibility with a myriad of practical applications - yet to be realised.

Return to:

Resources / Library Page

06 Aug 2023