Ben van Lier
Over the past few decades, the concept of fault-tolerant systems has become essential for the functioning of systems such as aeroplanes. According to Wensley and Lamport et al. [1], fault tolerance is achieved by making as much use of software programs as possible. Software programs enable distributed systems to reach a consensus and run decision-making procedures that let them execute information transactions independently. Running these consensus procedures requires what are known as voter routines, which enable the efficient and effective voting needed for systems to reach a consensus and make decisions. Once made, decisions are recorded in a distributed ledger based on a protocol. Decisions have to be recorded consistently and irreversibly in distributed ledgers as one block, so that the whole can permanently function as a virtual unit, while also offering a steady overview of previous decisions.
To run these voting routines quickly and adequately, there are three roles that Lamport deems essential: the proposer, the acceptor and the learner [2]. The interesting question is whether the combination of distributed ledgers, consensus and learning can also play a role in communications between separate and autonomous systems such as machines, factories or entire supply chains in a developing Industrial Internet of Things.
Distributed ledgers and the Industrial Internet of Things
Within a developing Industrial Internet of Things, a diversity of industrial systems and components of these systems will be interconnected in networks. Examples include wind turbines that are networked in a smart grid and communicate with other energy producers or consumers in their environment. Or locomotives that are able to independently communicate with other locomotives and components therein in their environment. In its reference architecture , the Industrial Internet Consortium (IIC) specifies that such industrial systems or components thereof “must be autonomous, and able to act independently based on the plan and information from other independently operating components nearby.” An essential requirement for communication is that the parties involved trust the communication and the information that is exchanged and shared. The IIC reference architecture therefore states that: “Trust is established before it is needed, and is necessarily hard to change with very formal procedures in place for transfer between commands.” Communication between such systems must, in principle, be fault-tolerant, which according to the IIC reference architecture refers to “the ability of the connectivity framework to ensure that redundant connectivity endpoints are properly managed, and appropriate failover mechanisms are in place when an endpoint or a connection is lost.”
As pointed out earlier, trust in intercommunications and in the information exchanged and shared between the parties hinges on factors such as consensus, decision-making and consistent and distributed storage of decisions and related information. Companies of the likes of General Electric are exploring the possibilities for such an industrial fault-tolerant communication system. In a blog post [3], GE already highlighted that: “the innovative shared-ledger technology offers transparent, immutable and mathematically verifiable record syncing across organisations with no need for trusted middlemen.”
Machine learning
To be able to reach a fault-tolerant decision on an information transaction within an Industrial Internet of Things, the decision-making procedure will have to be executed by four or more autonomous and distributed systems. As stated earlier, Lamport has identified three essential roles in reaching consensus on such a decision: the proposer, the acceptor and the learner. In Lamport’s view, it is up to the proposer to propose a decision-making procedure for an information transaction that is to be executed, while acceptors will have to indicate whether or not they can accept this proposal.
What then becomes particularly interesting in this context is the role of the learner, because, as Lamport explains, the learner: “can learn what value has been chosen.” According to Domingos , a learner is a learning algorithm that enables systems to learn from data and information. Amir [4] sees machine learning as a sub-area of artificial intelligence that focuses specifically on: “computerised automatic learning from data of patterns.” The purpose of machine learning is, in Amir’s view, “to use training data to detect patterns, and then to use these learned patterns to automatically answer questions and autonomously make and execute decisions.” In Domingos’ opinion, a learner’s capacity for learning is still limited within the framework of machine learning, leading him to state that: “learners can extract some things from data, but nothing you’d confuse with real knowledge.” In Domingos’ theory, the learner’s learning is only as good as the data available to the learner to learn from. He therefore states that: “He who controls the data controls the learner.” Domingos claims that, over the coming decade, the development of machine learning will be dominated by deep analogy, i.e.: “combining in one algorithm the efficiency of the nearest neighbour, the mathematical sophistication of support vector machines, and the power and flexibility of analogical reasoning.” Such deep analogy algorithms are currently primarily used for content or product recommendations that tie in with a specific profile, as used on websites such as Netflix, Amazon, or Bol.com. Such algorithms are also used for real-time monitoring of robotic arms in industrial settings.
Conclusion
Interconnectedness of distributed systems and the availability of increasingly intelligent algorithms will lead to systems acquiring an ever greater level of autonomy in independently making decisions for the execution of information transactions. To be able to adequately perform these information transactions, the distributed systems involved will have to become more intelligent through fast and efficient learning from available data and information. Domingos therefore correctly finds that: “the role of data and ownership of the models learned from it is what many of the twenty-first century’s battles will be about – between governments, corporations, unions and individuals.” The importance of reliable and fault-tolerant data and information exchange and sharing between a wide range of different systems can therefore be considered a fundamental precondition in the development of the Industrial Internet of Things.
1 Wensley, J. H., Lamport, L., Goldberg, J., et al. (1978) SIFT: Design and analysis of a fault tolerant computer for aircraft control. Proceedings of the IEEE, Vol. 66, no 10, October 1978.
2 Industrial Internet Consortium (2015) Industrial Internet Reference Architecture. Version 1.7 June 2015
3 Domingos, P. (2015) The master algorithm. How the quest for the ultimate learning machine will remake the world. New York, Basic books ISBN 9780465065707
4 Amir, E. (2014) Reasoning and decision making. in: The Cambridge handbook of Artificial Intelligence. Eds. Frankish, K. and Ramsey, W. M. Cambridge UK, Cambridge University Press ISBN 978521691918 (pp. 191-212)
Ben van Lier works at Centric as an account director and, in that function, is involved in research and analysis of developments in the areas of overlap between organisation and technology within the various market segments.
Original link:
http://www.centric.eu/NL/Default/Themas/Blogs/2016/04/08/Blockchain-distributed-ledgers-and-learning-machines-
To run these voting routines quickly and adequately, there are three roles that Lamport deems essential: the proposer, the acceptor and the learner [2]. The interesting question is whether the combination of distributed ledgers, consensus and learning can also play a role in communications between separate and autonomous systems such as machines, factories or entire supply chains in a developing Industrial Internet of Things.
Distributed ledgers and the Industrial Internet of Things
Within a developing Industrial Internet of Things, a diversity of industrial systems and components of these systems will be interconnected in networks. Examples include wind turbines that are networked in a smart grid and communicate with other energy producers or consumers in their environment. Or locomotives that are able to independently communicate with other locomotives and components therein in their environment. In its reference architecture , the Industrial Internet Consortium (IIC) specifies that such industrial systems or components thereof “must be autonomous, and able to act independently based on the plan and information from other independently operating components nearby.” An essential requirement for communication is that the parties involved trust the communication and the information that is exchanged and shared. The IIC reference architecture therefore states that: “Trust is established before it is needed, and is necessarily hard to change with very formal procedures in place for transfer between commands.” Communication between such systems must, in principle, be fault-tolerant, which according to the IIC reference architecture refers to “the ability of the connectivity framework to ensure that redundant connectivity endpoints are properly managed, and appropriate failover mechanisms are in place when an endpoint or a connection is lost.”
As pointed out earlier, trust in intercommunications and in the information exchanged and shared between the parties hinges on factors such as consensus, decision-making and consistent and distributed storage of decisions and related information. Companies of the likes of General Electric are exploring the possibilities for such an industrial fault-tolerant communication system. In a blog post [3], GE already highlighted that: “the innovative shared-ledger technology offers transparent, immutable and mathematically verifiable record syncing across organisations with no need for trusted middlemen.”
Machine learning
To be able to reach a fault-tolerant decision on an information transaction within an Industrial Internet of Things, the decision-making procedure will have to be executed by four or more autonomous and distributed systems. As stated earlier, Lamport has identified three essential roles in reaching consensus on such a decision: the proposer, the acceptor and the learner. In Lamport’s view, it is up to the proposer to propose a decision-making procedure for an information transaction that is to be executed, while acceptors will have to indicate whether or not they can accept this proposal.
What then becomes particularly interesting in this context is the role of the learner, because, as Lamport explains, the learner: “can learn what value has been chosen.” According to Domingos , a learner is a learning algorithm that enables systems to learn from data and information. Amir [4] sees machine learning as a sub-area of artificial intelligence that focuses specifically on: “computerised automatic learning from data of patterns.” The purpose of machine learning is, in Amir’s view, “to use training data to detect patterns, and then to use these learned patterns to automatically answer questions and autonomously make and execute decisions.” In Domingos’ opinion, a learner’s capacity for learning is still limited within the framework of machine learning, leading him to state that: “learners can extract some things from data, but nothing you’d confuse with real knowledge.” In Domingos’ theory, the learner’s learning is only as good as the data available to the learner to learn from. He therefore states that: “He who controls the data controls the learner.” Domingos claims that, over the coming decade, the development of machine learning will be dominated by deep analogy, i.e.: “combining in one algorithm the efficiency of the nearest neighbour, the mathematical sophistication of support vector machines, and the power and flexibility of analogical reasoning.” Such deep analogy algorithms are currently primarily used for content or product recommendations that tie in with a specific profile, as used on websites such as Netflix, Amazon, or Bol.com. Such algorithms are also used for real-time monitoring of robotic arms in industrial settings.
Conclusion
Interconnectedness of distributed systems and the availability of increasingly intelligent algorithms will lead to systems acquiring an ever greater level of autonomy in independently making decisions for the execution of information transactions. To be able to adequately perform these information transactions, the distributed systems involved will have to become more intelligent through fast and efficient learning from available data and information. Domingos therefore correctly finds that: “the role of data and ownership of the models learned from it is what many of the twenty-first century’s battles will be about – between governments, corporations, unions and individuals.” The importance of reliable and fault-tolerant data and information exchange and sharing between a wide range of different systems can therefore be considered a fundamental precondition in the development of the Industrial Internet of Things.
1 Wensley, J. H., Lamport, L., Goldberg, J., et al. (1978) SIFT: Design and analysis of a fault tolerant computer for aircraft control. Proceedings of the IEEE, Vol. 66, no 10, October 1978.
2 Industrial Internet Consortium (2015) Industrial Internet Reference Architecture. Version 1.7 June 2015
3 Domingos, P. (2015) The master algorithm. How the quest for the ultimate learning machine will remake the world. New York, Basic books ISBN 9780465065707
4 Amir, E. (2014) Reasoning and decision making. in: The Cambridge handbook of Artificial Intelligence. Eds. Frankish, K. and Ramsey, W. M. Cambridge UK, Cambridge University Press ISBN 978521691918 (pp. 191-212)
Ben van Lier works at Centric as an account director and, in that function, is involved in research and analysis of developments in the areas of overlap between organisation and technology within the various market segments.
Original link:
http://www.centric.eu/NL/Default/Themas/Blogs/2016/04/08/Blockchain-distributed-ledgers-and-learning-machines-
Pour lire tous les articles Finyear dédiés Blockchain rendez-vous sur www.finyear.com/search/Blockchain/
Participez aux prochaines conférences Blockchain éditées par Finyear :
Blockchain Vision #4 + Blockchain Pitch Day #1 (28 juin 2016)
Blockchain Hackathon #1 (octobre 2016)
Participez aux prochaines conférences Blockchain éditées par Finyear :
Blockchain Vision #4 + Blockchain Pitch Day #1 (28 juin 2016)
Blockchain Hackathon #1 (octobre 2016)
Les médias du groupe Finyear
Lisez gratuitement :
Le quotidien Finyear :
- Finyear Quotidien
La newsletter quotidienne :
- Finyear Newsletter
Recevez chaque matin par mail la newsletter Finyear, une sélection quotidienne des meilleures infos et expertises en Finance innovation, Blockchain révolution & Digital transformation.
Les 6 lettres mensuelles digitales :
- Le Directeur Financier
- Le Trésorier
- Le Credit Manager
- The Chief FinTech Officer
- The Chief Blockchain Officer
- The Chief Digital Officer
Le magazine trimestriel digital :
- Finyear Magazine
Un seul formulaire d'abonnement pour recevoir un avis de publication pour une ou plusieurs lettres
Le quotidien Finyear :
- Finyear Quotidien
La newsletter quotidienne :
- Finyear Newsletter
Recevez chaque matin par mail la newsletter Finyear, une sélection quotidienne des meilleures infos et expertises en Finance innovation, Blockchain révolution & Digital transformation.
Les 6 lettres mensuelles digitales :
- Le Directeur Financier
- Le Trésorier
- Le Credit Manager
- The Chief FinTech Officer
- The Chief Blockchain Officer
- The Chief Digital Officer
Le magazine trimestriel digital :
- Finyear Magazine
Un seul formulaire d'abonnement pour recevoir un avis de publication pour une ou plusieurs lettres