Training Deep Neural Networks Using Conjugate Gradient-like Methods
Round 1
Reviewer 1 Report
The article is a very pertinent and, despite being a rather well-known topic, the Authors have developed a very thorough theoretical part, full of interesting ideas.
Despite this, some integration and clarification are needed: for these reasons it is proposed a "major revision".
Some comments:
- Lines 32-51 the general description is too concise: a slightly broader description could be useful for the reader
- Lines 234-236: please explain better the results (too concise)
- Section 6: the conclusions are presented in an unclear way: please explain better
Minor comments:
- Figure 1 is too little and not clear: please improve dimension and resolution
- Figure 2 is too little and not clear: please improve dimension and resolution
- Figure 3 is too little and not clear: please improve dimension and resolution
- Figure 4 is too little and not clear: please improve dimension and resolution
- Figure 5 is too little and not clear: please improve dimension and resolution
- Figure 6 is too little and not clear: please improve dimension and resolution
- Figure 7 is too little and not clear: please improve dimension and resolution
- Figure 8 is too little and not clear: please improve dimension and resolution
- Figure 9 is too little and not clear: please improve dimension and resolution
- Figure 10 is too little and not clear: please improve dimension and resolution
- Figure 11 is too little and not clear: please improve dimension and resolution
- Figure 12 is too little and not clear: please improve dimension and resolution
- The expression between lines 92/93 has no identification number
- The expressions between lines 141/142 have no identification numbers
- The expression top page 7 has no identification number
- Some expressions of “Lemma A1” have no identification numbers
- Some expressions of “Lemma A2” have no identification numbers
- Some expressions of “Theorem A1” have no identification numbers
- Expression (9): please arrange the two members on different lines and assign each one an identification number
- Expression (10): please arrange the 3 members on different lines and assign each one an identification number
- It is recommended an extensive reading to correct some sentences and typo errors.
I also suggest integrating the References with at least these three papers:
- Caciotta, M., Giarnetti, S., Leccese, F.”Hybrid neural network system for electric load forecasting of telecomunication station” (2009) 19th IMEKO World Congress 2009, 1, pp. 586-590.
- Caciotta, M., Giarnetti, S., Leccese, F., Orioni, B., Oreggia, M., Pucci, C., Rametta, S. “Flavors mapping by Kohonen network classification of Panel Tests of Extra Virgin Olive Oil” (2016) Measurement: Journal of the International Measurement Confederation, 78, pp. 366-372. DOI: 10.1016/j.measurement.2015.09.051
- Proietti, A., Liparulo, L., Leccese, F., Panella, M. “Shapes classification of dust deposition using fuzzy kernel-based approaches” (2016) Measurement: Journal of the International Measurement Confederation, 77, pp. 344-350. DOI: 10.1016/j.measurement.2015.09.025
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
A statistical test should be used for the comparison of the examined methods.
The authors should better explain why the proposed methodology seems to work well and present some information about the time efficiency of their method.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Accepted in present form
Reviewer 2 Report
The paper could be accepted in the current form