Feasibility plan for putting together the proposal
Proposal is just about getting ground funding. This is just a feasibility study
PI would need the following information for the proposal (overview project, computational resources and team member)
no restriction for funding
What are the stakeholders’ expectations? (managing expectations + what's the minimum viable product that they want / can expect?)
publication within 2 years summarising features related
present to relevant european governement 3 years
Drive changes 4 years
dashboard - not minimum viable product
What are their reservations about data science approaches?
PI has concerns with black box model. PI would want more interpretable model and understand what is driving them
What is in-scope and out-of-scope? (how long do we have to produce minimum viable)
How Europe compare to USA and South Africa - out of scope
Time trend - out of scope
How does the output of the project look like and how is it going to be used?
As above
What are the expectations from using deep learning specifically? (recommend using a simpler approach)
a) can we use ML over DL
b) will there be domain expertise to help us extract the features?
c) what compute facilities are needed
PI suggested Deep learning initially as research group had successful publications. But PI is open to suggestions as PI prefer a model that is understandable but would like it to make exact predictions ideally
mostly postdoc for domain expertise
no specific access to computer facilities, may have funding for cloud computing, PI has no access to high performance computing
The expectations
Feasibility plan for putting together the proposal
Proposal is just about getting ground funding. This is just a feasibility study PI would need the following information for the proposal (overview project, computational resources and team member) no restriction for funding
What are the stakeholders’ expectations? (managing expectations + what's the minimum viable product that they want / can expect?)
What are their reservations about data science approaches?
PI has concerns with black box model. PI would want more interpretable model and understand what is driving them
What is in-scope and out-of-scope? (how long do we have to produce minimum viable)
How Europe compare to USA and South Africa - out of scope Time trend - out of scope
How does the output of the project look like and how is it going to be used?
As above
What are the expectations from using deep learning specifically? (recommend using a simpler approach)
a) can we use ML over DL b) will there be domain expertise to help us extract the features? c) what compute facilities are needed
PI suggested Deep learning initially as research group had successful publications. But PI is open to suggestions as PI prefer a model that is understandable but would like it to make exact predictions ideally
mostly postdoc for domain expertise
no specific access to computer facilities, may have funding for cloud computing, PI has no access to high performance computing