Implement all suggested revisions for JuliaCon proceedings paper provided here.
[x] In section 2.2 you mention different desiderata for counterfactual explanations. It would be a good idea to provide functions to evaluate the different generated counterfactuals on some standard metrics for these properties.
We now actually have a whole new module for evaluating and benchmarking CE. This probably deserves its own subsection, but I'm already pushing the page limit. Is there some flexibility for this?
[x] Figure 1 is difficult to read. I recommend increasing the font size of the text in the boxes.
Good point, done.
[x] It might be a good idea to give a description of counterfactual search outside gradient descent so that people unfamiliar with the topic understand what approaches there can be besides gradient descent.
I have a group of students currently working full-time on the package. Among other things they have added support for a gradient-free approach. Will add a sentence or two on this.
[x] In the paper you refer to the code in various locations as Listings. However, the code blocks have no numbering. I recommend provide such numbering to the blocks to make them easy to find.
Not sure how to address this, because of this issue I opened a while ago. Found a fix.
[x] Section 7.1 has a missing reference for the PROBE method.
Will fix, thanks.
[x] Section 2.2: realistic plausible -> realistic and plausible
Implement all suggested revisions for JuliaCon proceedings paper provided here.
We now actually have a whole new module for evaluating and benchmarking CE. This probably deserves its own subsection, but I'm already pushing the page limit. Is there some flexibility for this?
Good point, done.
I have a group of students currently working full-time on the package. Among other things they have added support for a gradient-free approach. Will add a sentence or two on this.
Not sure how to address this, because of this issue I opened a while ago.Found a fix.Will fix, thanks.
Fixed.