1. Home
  2. LEA
  3. All Papers 2019
  4. Features of deep study neural network
  5. Review of an article submitted to the conference

Review of an article submitted to the conference

Approval status: Accept after acknowledging of remarks
Date Published: 09.10.2019 - 14:15
Article’s title 
Features of deep study neural network
The article's title reflects the content and purpose of the article? 
Partially
Was the aim of the work clearly defined and successfully accomplished? 
No
Does the article embrace contemporary issues in the area? 
Partially
Does the article contain new and not published results? 
I do not know
Was the article clearly written and easily understood? 
Not clear
Conclusions illustrate the research results, recommendations and giving suggestions for future research 
Partially
The references are full and grounded? 
No
How adequate was the writing and used terminology? 
Rather adequate
Remarks and suggestions to the authors of the article 

According to the abstract of the article, it explores the peculiarities of the solution of the classification task using multi-layer neural networks
However, in the text of paper "In this work, studies were conducted to divide the two sets with different degrees of their overlap using a single-layer neural network"
for some reason is used single-layer neural network.
Therefore, the title of the paper does not match its content. In addition, scientists usually use the term "deep learning neural network" instead "deep study neural network"
The annotation states that "It is established that for division of two such sets it is necessary that the optimal number of elements (parameters) of sets is more than 20." However, it is not clear from the text of the paper how the authors has investigating it.

The authors stated, again in the annotation, that the experimental study was conducted using "a set of symptoms of diseases". However, no description of, or reference to, this sample of data is provided.

Regarding the text of the paper, it must be read again for fixing mistakes. In particular, the terms should be harmonized : "reverse error propagation", "Errors Back Propagation (EBP)".
Some of the wording is not clear: "paralysis of the network" , "for one teaching pair" - maybe "training pair" and so on...
The authors provide a code snippet based on an existing library. However, the topology of the neural network investigated in the paper is missing.

I confirm that there is no conflict of interests regarding reviewed article. 
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.