paper/aa.bib

@@ -2860,39 +2860,3 @@ booktitle = {American Institute of Physics Conference Series},
   year={2022},
   publisher={Nature Publishing Group UK London}
 }
-
-@article{lin2017selecting,
-  title={Selecting feature subsets based on SVM-RFE and the overlapping ratio with applications in bioinformatics},
-  author={Lin, Xiaohui and Li, Chao and Zhang, Yanhui and Su, Benzhe and Fan, Meng and Wei, Hai},
-  journal={Molecules},
-  volume={23},
-  number={1},
-  pages={52},
-  year={2017},
-  publisher={MDPI}
-}
-
-@inproceedings{khoshgoftaar2013survey,
-  title={A survey of stability analysis of feature subset selection techniques},
-  author={Khoshgoftaar, Taghi M and Fazelpour, Alireza and Wang, Huanjing and Wald, Randall},
-  booktitle={2013 IEEE 14th International Conference on Information Reuse \& Integration (IRI)},
-  pages={424--431},
-  year={2013},
-  organization={IEEE}
-}
-@article{al2005feature,
-  title={Feature subset selection using ant colony optimization},
-  author={Al-Ani, Ahmed},
-  journal={International journal of computational intelligence},
-  year={2005},
-  publisher={International Journal of Computational Intelligence (IJCI)}
-}
-
-@inproceedings{kumar2020problems,
-  title={Problems with Shapley-value-based explanations as feature importance measures},
-  author={Kumar, I Elizabeth and Venkatasubramanian, Suresh and Scheidegger, Carlos and Friedler, Sorelle},
-  booktitle={International Conference on Machine Learning},
-  pages={5491--5500},
-  year={2020},
-  organization={PMLR}
-}

paper/coverletter.tex

@@ -399,10 +399,7 @@ The study is potentially interesting but, in my opinion, the methodology, result
   The additional (thousands) runs are solely for the purpose to derive errors that indicate the effect of the measurement uncertainty. For this step a sensitivity analysis would be appropriate. However, we would still need to deal with dependencies between the individual features and would require a similarly involved/ comprehensive approach like, for example \verena{select something ..} It is also feasible to compute the error propagation through the k-means updates analytically. However, already with seven parameters that can all depend on each other this is a quite lengthy operation. 
 
   Review on supervised feature selection \citep{chandrashekar2014survey}Feature selection: filter (for example with correlation or mutual information), wrapper (machen wir,  \citep{javed2010feature,battiti1994using}) and embedded (integrates feature selection into the training process, also  criteria    often based in MI)
-  similar review in \citet{khoshgoftaar2013survey} addss search based feture selection
 
-  example with ACO: \citet{al2005feature}.
-  
 \citet{law2004simultaneous}  ``Both approaches, filters and wrappers, usually involve
 combinatorial searches through the space of possible
 feature subsets; for this task, different types of heuristics,
@@ -411,7 +408,6 @@ search, beam search, bidirectional search, and genetic
 search have been suggested [9], [33], [47], [63]''
   unsupervised feature selection
   
-\citet{kumar2020problems}: SHAP values for feature importance
   \citet{pudil1995feature}
 
   using unsupervised clustering as feature selection: \citet{mitra2002unsupervised}
@@ -423,9 +419,7 @@ search have been suggested [9], [33], [47], [63]''
 
   look into SHAP again
 
-  very different application but interesting: \citet{lee2022comparison}
+very different application but interesting: \citet{lee2022comparison}
-
-  feature selection which iteratively removes the least imporatant feature \citet{lin2017selecting}
 }
 
 Instead, the authors choose different comparison criteria, which are not very well introduced or explained, and end up giving contradictory results.