News

Examination for AI Related Invention in Japan

by Hirohito Katsunuma (Katsunuma International Patent Office – Japan)

The number of AI-related patent applications reached 3,065 applications in 2017 in Japan, which present a 65% increase compared with the previous year. Responding to this increase, the Japan Patent Office (JPO) further clarified the description requirement and the inventive step of AI-related inventions.

The JPO provided additional examination case examples of AI-related inventions on January 30, 2019. These additional examples include five examples for the description requirements and five examples for the inventive step. One example each for the description requirement and the inventive step are described hereunder.

1. Description Requirements for AI-Related Inventions

  1. A Certain Correlation Between the Cause Data and the Result Data
    When an invention uses an AI system to learn the training data including the cause data and the result data, a certain correlation is required to exist between the cause data and the result data. The existence of the correlation should be presumable by taking common technical knowledge at the time of filing into consideration.
  2. Allowable Claim and Non-Allowable Claim of Case Example No. 49
    The JPO provides a case example (No. 49) of which claim 1 is rejected and claim 2 is allowable. The invention relates to a “body weight estimation system” which can be used to estimate the body weight of a person based on his face image and body height. Claim 1 claims a system which learns data of “feature values representing face shapes” and “body heights” of people as the cause data and data of “body weights” of the same people as result data. Claim 2 claims that the “feature values representing face shape” is a “face-outline angle”. In addition, statistical data of the correlation between “BMI” and “face-outline angle” is disclosed in the description.
  3. The Explanation for the Case Example No. 49 Given by the JPO
    The JPO explains that claim 1 should be rejected as a correlation between “feature values representing face shapes” and “body weights” cannot be presumed, while claim 2 should be allowable as statistical data of the correlation between “face-outline angle” and “BMI” is shown so that the correlation between “face-outline angle” and “BMI” can be presumed.
  4. Points to Be Noted
    The applicant may overcome the rejection issued to claim 1 by showing the correlation between the “feature values representing face shapes” and the “body weight” was known to a person skilled in the art at the time of filing by filing an Argument. It should be noted that statistical data or experimental data at the time of filing can fulfill the description requirement.

2. Inventive Step of AI-Related Inventions

  1. A Use of New Training Data or Preprocessed Training DataWhen the AI-related invention uses new training data or preprocessed data by which the invention can produce a result to significantly fit the facts, the invention will be deemed to have an inventive step.
  2. Allowable Claim and Non-Allowable Claim of Case Example No. 34The JPO provides a case example No. 34 of which claim 1 is rejected and claim 2 is allowable. The invention relates to an estimation system of a hydroelectric power generating capacity of a dam. Claim 1 claims a system which uses data of precipitation amounts, flow rates of the upstream river and water inflow rates into the dam as the training data (input data as well). Claim 2 claims a system which uses temperature in addition to the training data of claim 1.
  3. The Explanation for the Case Example No. 34 Given by the JPOThe JPO explains that claim 1 should be rejected as training data/input data used by the system of claim 1 are commonly used in estimating a hydroelectric power generating capacity of a dam. However, claim 2 should be allowed as the use of temperature is new and leads to a significantly precise estimation of the hydroelectric power generating capacity of a dam.

For further information, please visit the following JPO websites;

https://www.jpo.go.jp/e/system/laws/rule/guideline/patent/document/ai_jirei_e/jirei_e.pdf

https://www.jpo.go.jp/e/system/laws/rule/guideline/patent/document/ai_jirei_e/jirei_tsuika_e.pdf