I have an educational assignment to make an floor-planning tool.

Can I use machine learning in some part of the algorithm? For example, I was reading the book Algorithms for VLSI Physical Design Automation by Naveed Sherwani, and here's a quote from there:

Initial estimate on the set of feasible alternatives for a block can be made by statistical means, i.e., by estimating the expected area requirement of the block.

Can I use ML for that? Or there's no sense in doing so? If no, is there some part where I can use ML?

Thanks in advance.

  • \$\begingroup\$ This question is probably a better fit on the Computer Science stackexchange than here. \$\endgroup\$ – The Photon Nov 10 '18 at 19:21

I don't work on floorplanning. But nowadays, many machine learning algorithm is base on a dataset and labels, now if you show all of these IC and layout to a network, the network can provide some sort of recommendation.

For example: https://github.com/timzhang642/3D-Machine-Learning#scene_synthesis section: Scene Synthesis/Reconstruction

for example this one: Make It Home: Automatic Optimization of Furniture Arrangement (2011, SIGGRAPH)

The main problem in machine learning is dataset, you haven't 1000 or 10000 IC layout, you don't have the tags of chips and their's failers or problem rate in the factory. sometimes such data is secret for a factory. If you have a big dataset, with previous best layouts or problematic ones you can show your layout and the network estimate future.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.