JPEN
PKSHA Enterprise AI

PKSHA Security

Eliminate crime through the eyes of a machine.

Eliminate unnecessary crime with algorithms. With the increasing digitization of all activities, fraudulent activities are becoming more sophisticated every day.
PKSHA Technology provides solutions that effectively deter fraud by using AI algorithms to learn fraud patterns on a daily basis, such as detecting fraudulent activity in financial institutions and monitoring fraudulent postings on Internet services, thereby providing powerful support to companies fighting against fraud.

For such customers

  • Credit card company

    • Credit card companies that want to effectively detect and deter fraudulent card use

  • Insurance company

    • Insurance companies that want to effectively detect and deter fraudulent insurance claims

  • Internet Service Provider

    • Service providers who wish to monitor content posted by unauthorized users

Solutions tailored
to your needs

Credit card fraud detection solutions

Unique AI algorithms detect fraudulent card transactions and deter damage. The system uses a "trend-following algorithm" to automatically and continuously learn the latest fraudulent methods to deter fraudulent use with a high degree of accuracy while reducing man-hours required for operation.
The system has already been adopted by many major credit card companies.

  • Trend-following machine learning model

    • Automatic extraction of features representing fraud trends based on historical data of fraudulent use.
    • Repeated updates of the model using recent trends and tracking of daily changing fraud patterns maintain stable fraud detection accuracy over the long term.
    • Effective abuse detection and reduced man-hours required for operation

  • Already implemented at major financial institutions
    Loss reductions of several hundred million yen/year

    • We are the number one solution with an overwhelming track record of use by major credit card companies represented by Credit Saison, JACCS, etc., and we boast several tens of percent of the domestic market share on a transaction basis.

  • Track record of cooperation with various systems

    • Many cases of introduction of linkage with fraud detection systems such as ACEPlus provided by Intelligent Wave, Inc.
    • By introducing AI algorithm solutions in conjunction with existing systems, additional system development is kept to a minimum.

Insurance Fraud Claims Detection Solution

Unique algorithms detect fraudulent insurance claims.
In addition to calculating a fraud score, the system also suggests important items for fraud judgment, providing highly convincing results in the final human judgment.

  • Multiple insurance companies have introduced or are in the process of introducing the system.

    • High detection rate compared to manual detection.

  • Algorithm construction based on deep understanding of fraud patterns

    • We understand the latest fraud patterns through repeated discussions with our assessors.
    • The knowledge is reflected in the detection algorithm to ensure a high detection rate and accountability.

  • Realization of speedy and high quality implementation

    • Can be implemented in 1/3-2/3 of the time compared to similar solutions.
    • The shortened implementation time reduces the burden on the customer at the time of implementation and achieves early results.

Illegal Posting Monitoring Solution

This solution uses state-of-the-art deep learning models to detect non-compliant expressions in messages sent by people.
As verbal expressions are becoming more diverse than ever before, the quality of decisions made by human operators alone can vary, but AI algorithms can raise the bar.

  • Introduction results at matching service providers

    • Compared to manual monitoring of postings, collaboration between algorithms and humans reduces man-hours by about 80%.

  • State-of-the-art natural language deep learning algorithms

    • Inappropriate expressions such as obscenity, slander, and solicitation range from direct to euphemistic, and euphemisms in particular are increasingly difficult to detect.
    • By learning from a large amount of fraudulent posting data, various euphemisms can be detected with a high degree of accuracy, greatly reducing the number of unauthorized postings.

Why PKSHA ?

  • Utilization of PKSHA Modules

    You can leverage already existing algorithmic modules developed by PKSHA ReSearch for natural language processing, speech/image recognition and synthesis, prediction, inference, and optimization.

  • Flexible customization to your needs

    Unlike packaged products, we can customize our products to meet the business needs of each industry and company, creating added value to achieve business strategy goals.

  • Over 2,000 companies have installed our products

    In the 10 years since our founding, we have provided software algorithms to more than 2,000 partners.
    We focus on commercial use, not POC.

Introduction Steps

Introductory Example

  • Credit Card Fraud Detection Solution

    PKSHA Algorithm Provides Fraudulent Use Detection Solution to Reduce Damage from Card Crime to JACCS, Inc.

    JACCS Co.,Ltd.

    Content
    The fraud detection solution developed by PKSHA Technology using its proprietary algorithm was customized and provided by Intelligent Wave, Inc. for JACCS' fraudulent use detection system.

    This algorithmic module constantly learns the latest fraud tactics to quickly respond to changes on the part of attackers and deter fraudulent use with a high degree of accuracy.

    JACCS and PKSHA conducted a demonstration test and confirmed that the fraud suppression effect exceeded expectations, leading to the introduction of this system.

    PKSHA Technology's fraud detection solution has been deployed by several credit card companies, and including clients who plan to deploy it in the future, our algorithms will operate behind the scenes of the approximately 10 trillion yen domestic credit card payment distribution network, contributing to fraud prevention. In the future, we plan to develop even more advanced fraud detection methods by cooperating and sharing knowledge and information with other credit card companies, and also provide services to small and medium-sized financial institutions in cooperation with financial systems.

  • Credit Card Fraud Detection Solution

    PKSHA Algorithm Provides Fraudulent Use Detection Solution to Reduce Damage from Card Crime to Credit Saison Co.

    CREDIT SAISON CO., LTD.

    Content
    PKSHA Technology's fraud detection algorithm was customized and provided to Credit Saison for use with ACEPlus, a fraud detection system provided by Intelligent Wave Corporation, which had been used by Credit Saison for some time.

    By assigning a score to each transaction in real time, the system makes it easier to determine if a transaction is fraudulent.

    In July-September 2020, approximately two years after the introduction of the system, the amount of damage caused by card fraud was 49.8% of the previous year's level, which was considerably lower than the average for the industry as a whole.

  • Insurance Fraud Detection Solutions

    Insurance Fraud Detection Solution Using PKSHA Algorithm Introduced by e-Design Insurance for Automobile Insurance

    E.design Insurance Co.,Ltd.

    Content
    With the provision of learning data and industry knowledge from e-Design Insurance, we provided our fraud detection algorithm customized for the automobile insurance industry, and it has been implemented. This algorithmic module learns fraudulent insurance claim techniques and detects suspicious claims with a high degree of accuracy, thereby deterring damage from insurance fraud.

    Countermeasures against fraudulent insurance claims are an important issue for the entire insurance industry from the viewpoints of maintaining a sound and stable non-life insurance system, maintaining trust in the insurance business, and ensuring fairness. On the other hand, the complexity and sophistication of fraudulent claims in recent years have made it necessary to understand a wide range of fraud patterns and to have a high level of expertise in order to counter fraudulent claims. Our machine learning algorithms automatically detect suspected fraudulent claims quickly, efficiently, and accurately. This enables the construction of a workflow that allows the person in charge to concentrate on the final decision, thereby contributing to the advancement of underwriting operations and the deterrence of payment for fraudulent claims.