#ArtificialIntelligence #68: Digital Twins - end to end flow

We announced the next version of our much awaited course Artificial Intelligence: Cloud and Edge Implementations. - the course always has a waiting list - so if you are interested, please apply

I have also been evolving my thinking on the flow of? Digital twins for our University of Oxford Digital Twins course???

The question is:?

Given a dataset - how can you manifest as a twin(s)??

Some notes:

• In our vision, simulation is the key (discrete event simulation, agent-based modelling, systems dynamics)
• The feedback loop is essential but could be? insights / manual / dashboard?
• We take a ?‘system of system view’ reflecting the state of the world - time t + x as per diagram
• Because Simulation is the main value add it is different from data science per se (typically Data science has only one target variable)
• Causal relationships are an input to simulation
• The diagram represents some if the libraries we are working on. There could be others

Some more outstanding questions:

• Is twins a graph problem or a time series problem and how the two translate . One example of a graph neural network and time series is? time series traffic forecasting where the graph models the dependencies of traffic segments on each other. Thus, ? ?we can define the traffic network as a graph and consider the traffic speed as a signal on this graph
• How should we incorporate process workflow???
• In addition to a time series and graph views, should we also consider a Business process view

Also Dr Francesco ciriello shared his view in a complimentary diagram as below - which I also like?

Welcome thoughts also

If you want to study with us, please have a look at

Artificial Intelligence: Cloud and Edge Implementations. and our University of Oxford Digital Twins course???