Japanese ver.

New application of AI:
Personalization of taste

Galaxy of Food from Sho Izumo on Vimeo.

Bold prediction about artificial chef in 2012

Modern dietary styles have been more inclined toward unhealthy eating patterns1). As such, there is an emerging consensus that public health efforts should be directed to developing an innovative technology to change dietary style to become more healthy.

While there is compelling research in nutrition sciences on what makes a recipe healthy2), this does not necessarily mean that such a recipe is matched to one’s unique food preferences. For example, the Japanese dietary style as measured in the 1970s is reported to have been healthy3). However, recipes in such a Japanese style may not be readily acceptable for those that prefer Southern American dietary style.

Furthermore, not limited to such a public health perspective, with growing diversity in personal food preference and dietary style, personalized information systems that can transform a recipe into any selected dietary style that a user might prefer would help food companies and professional chefs create new recipes.

Given that, in 2012, Prof. Varshney*, who led IBM’s Chef Watson project and is now at Univ. of Illinois Urbana-Champaign, made bold prediction as follows:

“In 5 years, computers will know what you like to eat better than you do4).”

Time flies and five years have passed already. Apparently, there are two significant challenges for realization of the prediction: 1) Let the computers understand our unique taste; and 2) Develop an algorithm that shifts our current taste into new frontier.

At 2017 SXSW, in collaboration with Prof. Varshney, we proposed a novel system which can not only understand our taste but suggest what we like to eat.

How does the computers understand our taste?

The first step for the computers to understand our taste is to identify where we are in the “galaxy of food5)”. To make the story simple, let such galaxy be two-dimension as illustrated in Fig.1.


Fig.1 Food Style Finder

This galaxy is created through vectorizing the variety of recipes around the world6). By doing so, the computers can understand our taste or the degree of dietary country/regional style mixture. For example, if you enter a week of recipe you have eaten, the system can identify where you are in the galaxy of food, as illustrated in Fig.2.


Fig.2 The computers can understand where you are in the galaxy of food

How does the computers propose a creative recipe which interests us?

After understanding our taste, next step for the computers to suggest what we like to eat is to calculate the degree of creativity of a recipe. As suggested by Varshney et al (2013), creativity can be theoretically and mathematically defined by “novelty” and “flavor pleasantness”7).

Note that there would be fundamental trade-off between novelty and flavor pleasantness. For example, Japanese Natto, fermented soybeans, might be novel but less flavorful to those from Austin, the U.S.


Fig.3 Fundamental trade-off between novelty and flavor pleasantness

Based on above, we developed an algorithm which can suggest what we like to eat considering our unique taste. In another word, the system can lead us to move in the galaxy of food from current style to new frontier. As an example, a traditional Japanese Sukiyaki was transformed into French style (see table for detail).

Table. Alternative ingredients suggested by our system and country probability of changing food ingredients in order from the top.


Future research for developing artificial chef

Let us make bold prediction that the old saying, you are what you eat, will be replaced as follows in 10 years:

“You are what AI cooks.”

It is extremely difficult for humanity to make a healthy decision making on food. However, if we let the machines make such decisions, we will be able to stay healthy without effort and enjoy novel and quality food every day.

We are more than welcome to collaborate with researchers, engineers, and entrepreneurs around the globe to realize above prediction together.

Further information


  1. Imamura, F., Micha, R., Khatibzadeh, S., Fahimi, S., Shi, P., Powles, J., et al.(2015). Dietary quality among men and women in countries in 1990 and 2010: a systematic assessment. The Lancet Global Health 3, e132–e142.
  2. Khatibzadeh, S., Micha, R., Afshin, A., Rao, M., Yakoob, M. Y., and Mozaffarian, D. (2012). Major dietary risk factors for chronic diseases: a systematic review of the current evidence for causal effects and effect sizes. Circulation 125, AP060–AP06.
  3. Tsuduki, T. (2014). Influence of Japanese food on senility and health maintenance. Yakugaku zasshi: Journal of the Pharmaceutical Society of Japan 135, 57–65215.
  4. IBM 5 in 5: Taste (2012)
  5. Interactive 3D version of the galaxy of food.
  6. M Kazama, M Sugimoto, C Hosokawa, K Matsushima, Lav R. Varshney, Y Ishikawa. A Neural Network System for Transformation of Regional Cuisine Style.( arxiv / frontiers )
  7. Lav R. Varshney, Florian Pinel, Kush R. Varshney, Debarun Bhattacharjya, Angela Schoergendorfer, Yi-Min Chee. (2013) A Big Data Approach to Computational Creativity. arXiv:1311.1213 [cs.CY]


Principal investigator : Yoshiki Ishikawa | Professor : Lav R. Varshney | Chef : Keisuke Matsushima
Data scientist : Masahiro Kazama | Nutrition scientist : Minami Sugimoto | Data visualization designer : Sho Izumo

Contact info: ishikawa@habitech.jp