February 9th, 2018

101 DeBartolo Hall

1:00 PM - 2:00 PM

Decentralized Workflows Using Vector Symbolic Architectures


Decentralized analytics require a means of specifying distributed data and computing control dependencies amongst services without the centralized coordination of the workflow. The distributed nature of such workflows puts far more emphasis on discovery mechanisms to ensure that each node can operate completely autonomously and cooperate with each other when needed. Consequently, the workflow must be capable of discovering the node(s) it needs to interact with at any specific time and in the most resource efficient way possible. To this end, a scheme for distributing the coordination information is needed that can minimize communication overhead, whilst providing comprehensive information surrounding the workflow being executed. In this talk, we explore the use of structured associative memory models called vector symbolic architectures for representing and orchestrating complex decentralized workflows. Such an approach offers a number of desirable features: it can encode workflows containing multiple coordinated sub-workflows in a way that allows the workflow logic to be unbound on-the-fly and executed in a completely decentralized way; the workflow and associated complex metadata can be embedded into a single vector; this vector representation is extremely compact; and is completely self-contained and can be passed around using standard group transport protocols. We describe the approach applied to multi-level complex workflows.


Dr. Graham Bent was formally a Senior Technical Staff Member in the IBM Emerging Technology Services (ETS) group at IBM Hursley and is an IBM Master Inventor. He retired from IBM in January 2016. He now works for IBM Research as a contractor and is director of his own company called Neurosynapse Limited. Over the past 10 years Graham has been involved in a UK MoD/US DoD research program - the International Technology Alliance in Network and Information Sciences (NIS ITA) undertaking research on large scale distributed databases; new encryption techniques for distributed secure computing using fully homomorphic encryption; and neuromorphic processing using the SyNapse technology (TrueNorth). He is currently involved in a new International Technology Alliance program on Distributed Analytics and Information Science (DAIS ITA). His current research is in the development of intelligent agents for distributed analytics using brain inspired neuromorphic computation.


Mr. Chris Simpkin graduated with honours in Electronics in 1983 and MSc in Computer Science in 1990. He worked as a design engineer on high speed control systems including design of analogue control loops, dedicated digital computer control systems and programming in both assembler and high-level languages. Chris worked for IBM for 10 years in various areas including stress testing of IBM’s flagship S/390 G5 Parallel main frames, IBM CICS and IBM Message Queuing products. In 1998 Mr Simpkin qualified as an Optometrist, gaining his third degree, and spending 10 years managing an Optometry business, before returning to computing, working on artificial intelligence, genetic computing, neural networks. Chris is currently doing a PHD at Cardiff University, UK, focusing in the use of machine learning algorithms for distributed data analytics applications.


Presentation Slides: