Hello everyone, My name is Kenshin Fujiwara and I am the CEO and founder of HACARUS Inc.
Through this series of blogs, I have discussed a wide range of topics of AI, from the history of AI to practical tips for the successful application of AI projects. I hope my blog posts will help you better understand AI and solve your business issues.
In the previous blog, we discussed the advantages of sparse modeling in the human-in-loop approach. As our last blog of this series, today, I will discuss the competitiveness of sparse modeling in the business context.
Sparse Modeling, the Best Option for Analysis with Small Amounts of Data
As a company, HACARUS first started in 2014 as a healthcare app developer. In the beginning, we provided an AI system that analyzed meal photos provided by the user and gave nutritional guidance. However, we ran into a problem where we weren’t able to gather enough data. This was because the task of taking a photo of each meal was tedious and users often forgot to do so.
Back then, our mission statement was, “to extend the life of everyone on Earth to 120 years.” Because of this, we were consumed by the idea of building a system that would allow HACARUS to perform analysis using even small amounts of data. We knew that deep learning had an issue with data volume, and this became a stepping stone for us to transition toward using sparse modeling methods.
Back in 2016, I met Dr. Masayuki Ozeki who works in the Department of Basic Information Science as part of the Graduate School of Information Science and Technology at Tohoku University. Professor Ozeki is well versed in algorithms using sparse modeling and joined HACARUS as our Chief Scientific Advisor.
While meeting with Dr. Ozeki, we formed a bond right from the beginning. In our meetings, I listened to him talk passionately about his research and was convinced that sparse modeling was the best option for HACARUS.
AI implementations for Business Purposes
Since this initial meeting, sparse modeling AI has gained more recognition and has been introduced into the pharmaceutical and manufacturing fields. Other universities are also taking an interest in sparse modeling and are currently conducting research for future implementations.
After conducting our own research, it would appear that HACARUS is the only company to introduce sparse modeling AI into actual business environments. We are currently working day and night to develop our business model. We also believe that sparse modeling had not yet gained the recognition that it deserves.
It would also seem that there are many people in North America, where deep learning development is particularly strong, who are skeptical of our claims.
This claim is ironic since sparse modeling is said to have originated in the U.S. While there are several theories, many believe that the regression analysis method proposed back in 1996 by Professor Robert Tibshirani of Stanford University was the catalyst for the sparse modeling model.
Combination of Sparse Modeling with Conventional Machine Learning Methods
Although AI designed for business purposes has become relatively well-known in the field of data science, it still has a long way to go. To recap, in this blog, I have talked about how sparse modeling directly competes with deep learning. However, in reality, they can be used in combination.
Since the main strength of sparse modeling is its ability to perform with small amounts of data, it can be used as an engine for other machine learning methods including deep learning.
Following a rise in consumption taxes back in 1997, Japan plunged into two years of deflation. Japan was said to be the only country in the world that had such long-term deflation.
While developed and developing countries are digitally arming themselves at the national and private levels to achieve super-fast economic growth, Japan is lagging behind when it comes to IT, including corporate DX and government digitalization.
As we begin to enter into Society 5.0, Japan needs to push forward to become an AI-advanced nation. We also need to innovate so that Japanese companies can regain their former levels of international competitiveness. I believe that AI technologies, like sparse modeling, that operate using small amounts of data, are highly interpretable, and have low levels of power consumption are optimal for Japan.
To everyone who has read this blog, I hope that it will be of some help to you on your own AI journey. I also hope that all of you will use AI to solve problems in your own professional life. For your updated knowledge and insight about AI technology, subscribe to our newsletter or visit HACARUS website https://hacarus.com.