FRONTEO Reports Results of Experiment Conducted in the Financial Services Agency’s “FinTech Experiment Hub”.

2018-08-02 16:34

FRONTEO Reports Results of Experiment Conducted in the Financial Services Agency’s “FinTech Experiment Hub”.

Time reduced by 42%, correct detections doubled, and also effective in the standardization and enhancement of capability

August 1, 2018

Confirms the Financial Services Agency’s positive approach to implementation of checking operations using AI

TOKYO, August 1, 2018 -- FRONTEO, Inc. (“FRONTEO”) (NASDAQ: FTEO) (TSE: 2158), a leading provider of artificial intelligence (“AI”) based big data analysis services, announced the results of an experiment comparing the operational productivity of “manual only checking” and “checking using KIBIT artificial intelligence” in the operations of financial institutions selected for the FinTech Experiment Hub* by the Financial Services Agency.
The experiment was conducted from May 2018 with the cooperation of MUFG Bank, Resona Bank, the Bank of Yokohama, and SMBC Nikko Securities as participating financial institutions.

Overview of the Experiment
Work Covered:
[Banking] Checking of sales discussion records when selling financial products such as investment trusts
[Securities] Checking of customer opinions and comments based on recorded call records
Details of the Experiment:
How many “Correct answers = records that should be found in checking” can be found during the time determined based on meeting records and voice call records?

Until now, financial institutions have verified a large number of meeting records from day-to-day selling of financial instruments by salespersons and an enormous volume of opinions and requests from customers via telephone through manual checks by people alone. This experiment involved quantitative comparison of detection precision, productivity, and the rate of standardization of work when records with random correct answers are only manually checked and when records are scored by KIBIT (assigning points) and those assigned as priority are manually checked.

One result of this experiment was to show that issues such as enhancement of operations and improvement of efficiency in financial institutions can be resolved by building models and verifying precision from a variety of perspectives using artificial intelligence through the practical application of KIBIT.

Main Results of the Experiment
- Compared to manual only checks, the time required for completing checks using KIBIT was 38% shorter for banks and 55% shorter for securities companies.
- The correct detection rate (number of correct detections ÷ number of correct answers available) when using KIBIT is equal to or greater than when performed by people alone, and it was proven that KIBIT is adequately able to learn the tacit knowledge of people.
- Compared to manual checks by people alone, it was proven that it is possible to check the content of all records more quickly and comprehensively, and that correct detections are possible while significantly reducing the number of hours of work.
- When checking samples in a predetermined amount of time (unit of time), the number of correct detections made when utilizing KIBIT was double that when using people alone.
- It was found that there was little variation in the number of correct detections per unit of time, and that there was a standardization effect, with a reduction in quality differences even when checks were performed by staff with different work experience and skills.

Furthermore, even with regard to the enhancement of detection capability, it was confirmed that precision in matters such as the ratio of correct answers included in the confirmed records (conformance rate) and the extent to which supplied correct answers (reproduction rate) were improved.

Based on these results, financial institutions that participated in the experiment indicated that they would like to further utilize AI for future checking operations because the results were of a high level. Concerns mentioned by financial institutions when utilizing artificial intelligence in checking operations, and opinions of the Financial Services Agency in supervision guidelines are as follows.

<Concerns mentioned by financial institutions>
・ In checking operations, are there any problems with operations when primary checks are performed by AI and secondary checks are performed by people?
・ Are there any problems with each financial institution setting their own criteria for decisions using AI and conducting operations accordingly?
・ Are there any problems with each financial institution setting their own frequency of checking the reliability of AI learned models?

<Views of the Financial Services Agency>
・ There is not thought to be any particular problems with operations using AI for primary verification in the verification operations of financial institutions under laws and supervisory guidelines if operations are conducted appropriately, such as by verifying the reliability of decision criteria and learned models using AI with reasonable methods and intervals.

Based on the results of this experiment and the views of the Financial Services Agency, FRONTEO and the participating financial institutions intend to achieve both improved productivity and fiduciary duty through the use of KIBIT in future checking operations.

About the FinTech Experiment Hub
The FinTech Experiment Hub was established in the Financial Services Agency on September 21, 2017 from the perspective of accelerating challenges aimed at innovation using FinTech to eliminate the hesitation and concern that FinTech firms and financial institutions are inclined to have in conducting unprecedented tests. The requirements for implementation of demonstration tests include (i) whether the test details and points for discussion are clear (transparency), (ii) the inclusion of user benefits and improvements in business productivity in Japan through the realization of the service (social significance), and (iii) recognition of the innovative nature of the service that they are trying to provide (innovation).
The Financial Services Agency will provide ongoing support regarding the points of discussion for compliance and supervision that were raised through the test, and the points of discussion for administrative work related to the interpretation of laws and ordinances that could arise when the service is provided to general users.

KIBIT is an AI engine developed by FRONTEO. KIBIT is a word coined by combining "kibi," a Japanese word meaning "subtlety," and "bit," the smallest unit of digital information, in order to indicate an AI capable of understanding the subtle elements of human behavior and personality.

About FRONTEO, Inc.
FRONTEO, Inc. (“FRONTEO”) (NASDAQ: FTEO) (TSE: 2158) supports the analysis of big data based on behavior informatics by utilizing its technology, "KIBIT". FRONTEO's KIBIT technology is driven by FRONTEO artificial intelligence based on knowledge acquired through its litigation support services. KIBIT incorporates experts' tacit knowledge, including their experiences and intuitions, and utilizes that knowledge for big data analysis. FRONTEO continues to expand its business operations by applying KIBIT to new fields such as healthcare and marketing. FRONTEO was founded in 2003 as a provider of e-discovery and international litigation support services. These services include the preservation, investigation and analysis of evidence materials contained in electronic data, and computer forensic investigation. FRONTEO provides e-discovery and litigation support by making full use of its data analysis platform, "Lit i View", and its Predictive Coding technology adapted to Asian languages. The company name was changed from UBIC, Inc. to FRONTEO, Inc. as of July 1, 2016.
For more information about FRONTEO, contact or visit