HACARUS-X

AI that's understandable by human

HACARUS-X is a non-deep learning based AI service that can work with
very small data set and yields a result understandable by human.

Our strength

  • Problems with
    too few data sets.

    Our sparse modeling
    algorithm only requires
    small data set for analysis.

  • Need results in a
    short period of time.

    Able to identify
    the characteristics that affect
    result during a trial period
    for 1 month or so
    .

  • Want to keep initial investment as low as possible.

    You will not need
    the large scale computers
    or software licenses
    that deep learning requires.

Sparse Modeling

Sparse modeling is a data analysis methodology that extracts and simplifies only the essential elements of the calculation formulas used in complex analysis. It makes it possible to derive meaningful analysis results from small amounts of data, and is used to automate analysis and identification tasks when the large amounts of learning data required for deep learning cannot be obtained. Another feature of sparse modeling technology is that it can clarify the causal relationships between different pieces of data. We can use it to derive "the correct data itself," which is not possible with deep learning.

Chief Science Advisor
Masayuki Ohzeki

Masayuki OHZEKI, associate professor at the Graduate School of Information Science at Tohoku University. His research interests are broad, including machine learning and its potential from a perspective of theoretical physics and itself. He is promoting techniques of the “sparse modeling”. He graduated with a Ph.D. in physics from Tokyo Institute of Technology in 2008, and subsequently spent one and a half years as a postdoctoral fellow. He worked as an assistant professor in the Kyoto University and also researcher of Rome University. Since 2016, he has been a present position. He was awarded the 6th Young Scientists’ Award of the Physical Society of Japan, and the Young Scientists’ Prize by The Commendation for Science and Technology by the Minister of Education, Culture, Sports, Science and Technology in 2016.

Target Area

  • Image analysis
    - Medical image data
    - Drone aerial image data
    - Architectural image data
  • Chronological data analysis
    - Healthcare information
    - Access log data
    - Machinery operation data from factories, etc
  • Other
    - Conversation log data
    - Analysis of natural language in texts

Case

Detecting areas for repair on a building wall with drone aerial images (iRobotics Inc.)
In this case, our task was to detect cracks, loose bits, peeling, and dirt from aerial images of a building taken by a drone in order to identify areas for repair. Using sparse modeling to extract characteristics from small amounts of data, we were able to detect the areas that needed repair with just a few images.

Flow

Contract

  • - Interview
  • - Non-disclosure agreement
  • - Sample data

Pilot $30K〜

1-3 months

In this phase, we look at the data
you have to verify what sort of analysis
we can perform with Hacarus technology.

  • - Analysis of actual data
  • - Sparse modeling
  • - Application of other methods

Consultation $30K〜

3-6 months

We help you examine and collect the data according to the task.
We decide what direction to take with the model.

  • - Defining necessary data
  • - Defining system coordination requirements
  • - Planning implementation

Model construction and API development ASK

2-month

We build the model based on our implementation plan.
At the same time, we also design and develop a system
to coordinate with outside applications.

  • - Model construction
  • - API service design and development

Application/ maintenance ASK

We help you maintain the model for continued data analysis

  • - Model maintenance
  • - API service application

FAQ

  • Q.
    The contract can be divided into stages, so you can check the report for the trial period and decide if you want to proceed to the next phase.
  • Q.
    With the sparse modeling method, we can analyze data starting from just a few dozens of data.
  • Q.
    Depending on the results of your consultation, we may propose developing the entire system or working with an existing development contractor.
  • Q.
    Our company is focused on the healthcare field, but Hacarus-X can accommodate data from all kinds of industries.