In DepthCriminal Justice

Are algorithms good judges?

See allHide authors and affiliations

Science  19 Jan 2018:
Vol. 359, Issue 6373, pp. 263
DOI: 10.1126/science.359.6373.263

eLetters is an online forum for ongoing peer review. Submission of eLetters are open to all. eLetters are not edited, proofread, or indexed.  Please read our Terms of Service before submitting your own eLetter.

Compose eLetter

Plain text

  • Plain text
    No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g. higgs-boson@gmail.com
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Statement of Competing Interests
CAPTCHA

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.

Image CAPTCHA
Enter the characters shown in the image.

Vertical Tabs

  • RE: Good judges come from good models, good controls and good information quality
    • Cedric Fan, Professor, MIT Information Quality Program- Data Quality & Info Security Lab, Nanjing Tech University

    Catherine Matacic wrote an IN DEPTH entitled "Are algorithms good judges?"(Science, 19 Jan 2018: Vol. 359, Issue 6373, pp. 263) (1). Our answer is also "yes", however, it depends on whether the whole system is “good” or not.

    First of all, models should be “good”, which basically means appropriate models and correct parameters. “Good” models can also have the ability to adopt technologies such as reinforcement learning to solve those fuzzy and time-varying problems for judges. Secondly, controls should be “good” to have enough system supports and correct system controls. Finally, the information should be “good” to have both enough information quantity and satisfied information quality, which can have the largest influence on the result.

    For more flexibility, some human judges can also be included in the whole system to give out the good judges.

    REFERENCES
    1 Catherine Matacic, Science, 359, 6373 (2018)
    2 National Natural Science Foundation of China (71671089)

    Competing Interests: None declared.
  • Reinforcement learning is better than human judges

    Catherine Matacic wrote an article entitled "Are algorithms good judges?" (1). Answer is "yes". Reinforcement learning without human knowledge which is one of machine learning algorithms is better than any human champions in the game of Go (2). In other words, reinforcement learning system can learning by itself using data where better algorithms are automatically created and competitions among them can further improve their qualities. Since our lifetime is limited, we have a limited time for exercises in order to improve our skills. Reinforcement learning can be trained within a short time with massive computer resources which may be equivalent to more than 1000 years' human training. Reinforcement learning "elmo" is a champion algorithm in the game of Shogi in 2017 (3). Machine learning algorithms have been evolving from human-teaching with supervised learning to training-by-itself with reinforcement learning without human knowledge.

    References:
    1. Catherine Matacic, "Are algorithms good judges?" Science, 19 Jan 2018: Vol. 359, Issue 6373, pp. 263
    2. David Silver et al., " Mastering the game of Go without human knowledge," Nature 550, 354–359 (19 October 2017)
    3. http://www2.computer-shogi.org/wcsc27/

    Competing Interests: None declared.