Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Research for practice: troubling trends in machine-learning scholarship
Lipton Z., Steinhardt J. Communications of the ACM62 (6):45-53,2019.Type:Article
Date Reviewed: Aug 20 2019

The article reviews the attention paid to an important issue in machine learning research, but the problem raised can be generalized to other fields within computer science and informatics.

Machine learning and the research area known as artificial intelligence (AI) is a fashionable, fast-growing research field that is prone to business interests. The authors discuss the problems related to the scientific content of papers, as well as how the presented ideas and measurements can be evaluated by reviewers and the scientific community. The authors emphasize that scientific rigor has developed essentially over the past decade; however, some phenomena should be dealt with. A summary of the raised issues include: the source of demonstrated empirical advances should be clearly and experimentally underpinned; requirements for mathematical formalism are excessively demanding, sometimes blurring the obscure idea that leads to “mathiness”; and the use of language that misleads readers by creating connotations without profound grounding.

The article highlights that other scientific disciplines must deal with similar problems, and is a recurring phenomenon even in AI. It concludes that self-correction through the scientific community can happen via recurring debates. Keeping the problem on agenda--“what constitutes reasonable standards for scholarship”--is an important task. My personal experiences with reviewing scientific papers from diverse areas in computing buttress the opinions expressed in this article. We can hope that the evolution of the scientific discipline might define the appropriate standards that are to be applied.

Reviewer:  Bálint Molnár Review #: CR146662 (1911-0400)
Bookmark and Share
  Reviewer Selected
Editor Recommended
Featured Reviewer
 
 
Learning (I.2.6 )
 
 
Miscellaneous (K.3.m )
 
 
Miscellaneous (A.m )
 
Would you recommend this review?
yes
no
Other reviews under "Learning": Date
Learning in parallel networks: simulating learning in a probabilistic system
Hinton G. (ed) BYTE 10(4): 265-273, 1985. Type: Article
Nov 1 1985
Macro-operators: a weak method for learning
Korf R. Artificial Intelligence 26(1): 35-77, 1985. Type: Article
Feb 1 1986
Inferring (mal) rules from pupils’ protocols
Sleeman D.  Progress in artificial intelligence (, Orsay, France,391985. Type: Proceedings
Dec 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy