Version 2.1 by Maximilian Kasperowski on 2022/08/26 08:10
Show last authors
1 In top-down layout a strategy needs to be used to set node sizes without knowledge of the hierarchical contents of the node as that has not been processed/laid out at that point. Current strategies are:
2
3 * using a default base size
4 * counting the number of children and taking the square root as a multiplication factor for the default base size
5 * computing the layout of only the children (look-ahead layout)
6
7 The main challenge is to get an approximation that gives a suitable aspect ratio (close to what will actually be required).
8
9 Because graphs are complex feature vectors and the solution space is very large without necessarily one correct and optimal answer a ML-based approach may help find good solutions.
10
11 This topic will be supervised in cooperation with the [[Intelligent Systems>>url:https://www.ins.informatik.uni-kiel.de/en||shape="rect"]] group.
12
13 = Goals =
14
15 * use kicodia benchmarking tool to extract feature vectors from existing models
16 * train and evaluate an ML model on the extracted data sets
17 * integrate the model as a new node size approximator into top-down layout
18
19 == Example Top-down Layout of an SCChart ==
20
21 [[image:attach:Controller_topdown_v3.png]]
22
23 = Related Work/Literature =
24
25 [WIP] Top-down layout paper
26
27 [[http:~~/~~/neuralnetworksanddeeplearning.com/index.html>>url:http://neuralnetworksanddeeplearning.com/index.html||shape="rect"]]
28
29 [[https:~~/~~/www.deeplearningbook.org/>>url:https://www.deeplearningbook.org/||shape="rect"]]