Almanach-HF
/

flowshop_transformer

Model card Files Files and versions

younadi commited on Feb 21

Commit

f6cee7f

·

1 Parent(s): 3aae83b

readme.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -24,7 +24,7 @@ Different launch sequences result in different completion times of all the jobs.
 # Proposed approach
 Given an PFSP instance, i.e. a Jobs x Machines matrix, the idea we would like to investigate is to recover an optimal/pseudo-optimal schedule using a two learning phases process:
-<img src="presentation/schemas/global_architecture.png" width="600">
 **Phase 1:** In this phase, the goal is to train a neural model (which we will call the objective surrogate or simply surrogate) to learn latent job embeddings and to predict/estimate the MakeSpan associated with job schedules that are represented as sequences of those job embeddings, specifically:

 # Proposed approach
 Given an PFSP instance, i.e. a Jobs x Machines matrix, the idea we would like to investigate is to recover an optimal/pseudo-optimal schedule using a two learning phases process:
+<img src="presentation/schemas/global_architecture.png" width="1200">
 **Phase 1:** In this phase, the goal is to train a neural model (which we will call the objective surrogate or simply surrogate) to learn latent job embeddings and to predict/estimate the MakeSpan associated with job schedules that are represented as sequences of those job embeddings, specifically: