Dynamic Weight and Mapping Mutation Operation-Based Salp Swarm Algorithm for Global Optimization

Round 1

Reviewer 1 Report

1. How restrictive is considering X(N,D) a 2-dimensional distribution?

 

2. Which type is the randomness considered in X(N,D)? (Uniform, gaussian...).

 

3. Justify the type of randomness considered in (2). 

 

4. How did you determine the part *(ub-lb)+lb in eq.2. 

 

5. Where does it come from c1?

 

6. How do you define G(B) in eq. (8)?

 

7. What is the probability distribution corresponding to fig. 2?

 

8. Fig. 7 is very good to show concave functions. Does there exist a way

to guarantee that such shape always exist?

 

9. About (8), can you guarantee that the minimum (a) actually exist and (b) can actually be attained?

 

10. Can you determine the algorithmic complexity of your whole algorithm? How better is this

result with respect to the compared algorithm complexities?

 

 

 

 

 

 

 

None.

Author Response

Word

Author Response File: Author Response.docx

Reviewer 2 Report

The submitted version of the paper is clearly missing a few sections of text, namely the section between lines 179 (on page 4) and 184 (on page 6) and lines 232 (on page 7) and 243 (on page 9), as well as the figure captions on page 5 and page 8. Mistakes can always happen, but I think authors should be more careful before submitting a manuscript.

Having said that, I have reviewed many papers on evolutionary global optimization methods, and they all look very similar, proposing new algorithms or finely-tuned new operators that are supposed to overcome the limitations of the previous algorithm or set of operators. All of these new methods work better on a fixed (and often cherry-picked) set of test problems, but inevitably fail when faced with new real-world problems. And this leads to the emergence of a new algorithm or of a new operator.

This paper follows the same path, but has the merit of being well written (although the English often needs careful improvement) and clearly presenting the possible benefits of adding dynamic weighting to the simple SSA algorithm. For this reason, it deserves to be published in Applied Sciences, although I would have preferred the authors to choose a couple of actual problems to show the benefits of their approach instead of the usual long (and admittedly boring) list of bolded improvements.

However, before final acceptance of their submission, authors should address the following issues:

- line 67: I guess POA should read PSO;

- Figure 2: the form of the plot is inappropriate; it would be better to plot the true statistical distributions of the random numbers or a scatter plot showing the R_n+1 vs. the R_n random point;

- Table 2: a computer is not an "experimental" environment, it is a computing environment;

- Figure 5: the yellow plot is very hard to see; in any case there is no reason to use a scatter plot for unrelated data as those shown here, it would be much better to use min/max bars or some other more appropriate representation;

- as noted above, the text sections between lines 179 and 184 and lines 232 and 243, as well as the figure captions on pages 5 and 8 are missing and should be added.

 

English is acceptable overall but should be carefully revised as many sentences are incorrect, e.g. "in light of [the free lunch theorem]", "same as other", "the surplus of this", "four celebrated", "recorded on 30 times", "experiment is repeated 30 iterations", "are not reflected", "by implementing... which provide", just to name a few.

Author Response

Word

Author Response File: Author Response.docx