Bayesian Optimization Algorithm for High-Cost Evaluations. Paper available here.
This code is an addition to the existing Bayesian Optimization Package, developed here. We change the bayesian_optimization.py file as well as the util.py file to include Expected Value Over Time, a new acquisition function that takes into account both function values and the length required to evaluate them. Our work is found in Main.ipynb, which utilizes an image recognition model defined in imagerecognitionmodel.py. We then analyze the values in Analysis.ipynb. Our code is all open-source and is completely re-creatable after downloading the cifar-10 dataset online. Alternatively, our results are in data, which can be analyzed freely.