AI perfume is coming. Will you buy it?

Want to make APP Same thing as WeChat , Can run small programs smoothly ？ | Experience will send you to Xinjiang 、 Huawei 、 Cherry keyboard ！>>>

writing | Huang Kangxi

source | Intelligent relativity （ID：aixdlun）

“A woman who doesn’t wear perfume has no future.”—from Coco Chanel

As a customized art with a long history , Perfume has been welcomed by millions of people for thousands of years. . Professional flavorists need decades of study and training , Only then can we acquire the valuable skill of developing perfume. . In recent years, , In the image 、 In the field of image and speech recognition, artificial intelligence has begun to permeate the perfume industry. . Interestingly ,AI Perfume does not depend on smell. , Its principle is similar to artificial intelligence composition and writing , Through a large number of analysis of perfume ingredients and sales information , Using big data operations such as probability statistics and permutation and combination , To help the flavorer to make the formula that meets the consumer's preference more efficiently .

AI Perfume = Existing spices + formula + Sales information

2018 year , World renowned flavor production company Symrise And IBM Research We have jointly developed a new product called Philyra Of AI Perfume debug system , Its name is inspired by the perfume goddess in Greek mythology. .Philyra By analyzing the existing aromatic formula , Mix its ingredients with the sales area 、 Customer age and other data are matched , Modulate the perfume that best suits the customers' preferences. .

at present , Perfume industry has approximately 1300 Spices , Contains synthetic flavors , Additional flowers 、 moss and lichen 、 Herbs and fruit extracts . Customer base coverage Estée Lauder、AVON、Coty and Donna Karan And other famous perfume companies. Symrise, towards IBM It provides a set of , common 170 Database of ten thousand recipes as Philyra Learning materials for .IBM Input this huge perfume formula information and sales performance information. Philyra, And Symrise Provide other customer data for comparison , For example, which flavors sell best in which places 、 Who are the main buyers 、 Which flavors do consumers of different ages prefer . By learning the existing perfume materials 、 Formula , And the analysis and comparison of the existing sales data ,Philyra These data can be used to predict human preferences , This creates new formulations for specific populations .

Philyra The algorithm mainly covers four main categories ： They are supplements or substitutes of raw materials in the formula 、 Flavor raw material consumption 、 Prediction of human response to fragrance and its novelty . That is, by comparing the newly developed perfume samples and the finished products on the market. , Discover its novelty . This technology can also be used for research and analysis of competitive products .

Successful experiment

2018 year ,Philyra and Symrise Senior vice president and senior Perfume Perfumes David Apel We did an experiment together . They made three totally different perfumes. : One of them is completely made by artificial intelligence ; The second perfume body is completed by AI. , The flavorer made some adjustments ; The third perfume is mainly made by the perfumes. , Artificial intelligence as an aid . The three perfumes were tested by consumers. , Most people chose 100% Perfume produced by AI .

2019 year , The second largest beauty shop in Brazil O Boticário and Philyra cooperation , Living in Brazil 95 And after 00 The last is the target group , I hope to develop a perfume that suits their taste. .

Philyra The flavor of the region was compared with the corresponding age range , Two formulations were created . After formulation , from Apel Adjust the product , Highlight a particular aroma , And prolong its duration on the skin . The two recipes were made to match the preferences of most girls , Mixed fruit aroma 、 Floral and vanilla Egeo ON Me; And mixed with nutmeg pods , carrot seed 、 Milk fragrance 、 cream , Rich in tone Egeo ON You. Forecast based on data ,Philyra The creation of perfume formula should be able to get good feedback from target groups. . These two fragrances are already available. 2019 year 6 month 12 Issued on Valentine's day in Brazil , in fact ,Egeo ON Me and Egeo ON You You bet In a small range of promotion to obtain excellent response , Even the other young people who love the perfume of Brazil are ranked among the best. .

It is different from the traditional flavoring method ,Philyra Do not rely on fragrance to identify which flavor should be added next. , But rather After comprehensive analysis of deep learning algorithm , Then decide the order of spices . Artificial intelligence fragrance system will not be biased by culture 、 Personal preferences 、 knowledge 、 Experience or the effect of comfort on a substance , To be able to Discover possibilities that have never been explored in the past .Symrise The flavorer Apel Express , When Philyra When cooking spices such as cardamom pods and fenugreek seeds are combined with milk and butter base materials , He was very surprised .“ It's a recipe I've never thought of before .”Apel say ,“ It's not in the range of ordinary materials that I would use .”AI The new blending method of the flavorer has overturned the thinking of the past , It provides a brand new idea for perfume research and development. .

Time saving and labor saving

Same year , Switzerland based well-known flavors and fragrances Givaudan Fragrances, Launched a model called Carto Artificial intelligence fragrance system . This system uses Givaudan The spice , To create a “ Odor value map ”(Odour Value Map).

Carto Pyramid as the underlying logic in the field of perfume development , Digital quantification of the sense of smell of consumers , Provide the fragrance maker with suggestions on composition combination through machine learning technology . The flavorer operates through the touch screen , You can easily invoke data from the brand's complex perfume recipes. , Combine different flavors together . Besides ,Cartoc Also contains “ Rapid sample preparation ” function , We can finish the proofing work of perfume products in a very short time. , Make these spices into perfume. , Let the flavorer test their new fragrance immediately .

“ This method saves the flavorer more time .”Givaudan Chief flavorer Calice Becker say . these years , The development of perfume is constantly developing. , A new perfume is needed. 6 Months to 4 Years of time . Artificial intelligence has greatly shortened the development cycle of perfume. , And greatly improve the production efficiency .

Because samples can be created in real time , The fragrance technician can almost always debug perfume with customers. , Obviously improved the efficiency and quality of perfume customization service. , Perfume for personalized perfume and experience as a selling point. （Parfum Bar） Marketing helps a lot . at present ,Givaudan It has been set up in fragrance creation centers in various regions of the world Carto System . At the same time of serving the flavorer , The system is also involved in every step of creating perfume. .

according to Euromonitor The prediction of ,2022 year , The value of the global perfume market will reach 700 Billion dollars . China's perfume industry market grew year by year 29.88%, achieve 80 RMB 100 million , among 70% The above shares are occupied by international brands .2019 In the first half of , China's perfume market has increased. 43%, achieve 22.4 One hundred million yuan . among , The high-end perfume market is not afraid of external economic factors. , Achieved the highest growth rate in three years ,2022 Domestic perfume market is expected to break through 400 One hundred million yuan .

Opportunity or challenge ?

Perfume industry has broad prospects for development. , Future period .Philyra and Carto The emergence of artificial intelligence flavoring equipment , No doubt the gospel of perfume company. . rely on AI Great computing power , Can quickly learn and analyze a large number of recipes 、 Raw materials and sales data , Provide technical support for human flavoring . For big companies , It can open up new paths , It can save time and labor , It's a very useful technology . But for small perfume producers, , It may not work .

First ,AI A flavorer must learn 、 Only by analyzing a large amount of data of the target customer group can it work . Only like Symrise and Givaudan Fragrances This kind of system has been established for a long time 、 The scale of the enterprise is large 、 Wide range of sales A large company , It has a list of perfume materials that can support AI training requirements. 、 Formula and quantity of customer information . Smaller perfume companies may not have enough ingredients and consumption statistics to help. AI Study .

secondly , The operation principle of the artificial intelligence flavoring system is to use big data analysis 、 Probability statistics and other algorithms , Look for spices 、 Relationship between formula and sales results , Use powerful computing power again , Calculate the most popular flavor formula for target customers . This algorithm has been used in other fields for a long time , Relatively mature . For Perfume Companies , And IBM And other companies with mature algorithms and systems , Compared with our own research and development AI More time and effort ; And AI After the popularity of flavorers , Can greatly reduce the manpower of perfume research and development 、 Time cost , It is a beneficial and necessary investment . Small perfume companies may not have their own research and development. AI Fragrance system or purchase AI Capacity and budget of flavoring equipment .

because AI Flavoring technology has not yet been popularized , at present , The influence of artificial intelligence system on human perfumery and perfume consumers is not great. .AI The flavorer can not only automatically create recipes that meet the preferences of the target customers , It can also produce samples in real time , Greatly shorten the development and manufacturing time of perfume . Like all automated production equipment , Application of artificial intelligence in perfume industry , It seems only a matter of time . With AI Perfuming technology and the gradual maturity of automatic production of perfume , Perfume companies have lower R & D costs. 、 More efficient 、 More accurate direction , Maybe the Matthew effect of perfume industry. , It may also help minority perfume manufacturers select their target consumer groups. .

Conclusion

But that doesn't mean the human flavorer is about to be AI Fragrance system replacement , Just as composing robots cannot completely replace human musicians . It's like Philyra Members of the R & D team Richard Goodwin Yes ：“ For hundreds of years , People are always exploring how to combine ’ art ’ and ’ science ’ Method of designing perfumes perfumes .” And now with AI The assistance of , Human flavorists just need to focus on what they're good at “ art ” part ,“ science ” We only need to give the part of the algorithm system , The combination of the two greatly improves the efficiency of deployment .

( The above pictures are from the Internet )

Reference material :

1.Chavie Lieber《Is AI the future of perfume?IBM is betting on it. 》

2.Andria Cheng《Now,Even Your Perfume May Be The Result Of Artificial Intelligence》

3. Research in the network 《 use AI Perfume ? World's largest fragrance and fragrance company AI Tool creation “ Odor value map ”》

4. Core reading 《 Today's core sound | Great changes in perfume industry ： Craftsmanship is no longer great , All hands “ Others ”》

5. Luxuriant tzu 《AI Perfume system Philyra Artificial intelligent deployment of the first premium perfume will be available. 》

6. Competitive Technology 《 Perfume non-toxic , There is only a lack of it “Ai”》

7. Tencent network 《2022 year Global perfume market will reach 700 Billion dollars Perfume bar becomes the focus of domestic perfume market 》

Dig deep into the well of intelligence , Tonghao add vx：zhinengxiaoyan

This content is 【 Intelligent relativity 】 original ,

To represent only one's point of view , unaccredited , No one can use... In any way , Including reprint 、 Excerpt 、 Copy or mirror .

Some pictures are from the Internet , And the copyright is not verified , Not for commercial use , In case of infringement , Please contact us .

Intelligent relativity （ WeChat ID:aixdlun）：

•AI Industrial new media ;

• Winner of Qingyun plan TOP10;

• Surging news technology list monthly top5;

• The article is long-term “ Occupy ” Top articles in titanium media TOP10;

• The author of 《 Artificial intelligence 100000 why 》

•【 Focus areas 】 Smart appliances （ Including white electricity 、 Black electricity 、 A smart phone 、 UAV, etc. AIoT equipment ）、 Intelligent driving 、AI+ Medical care 、 robot 、 The Internet of things 、AI+ Finance 、AI+ education 、AR/VR、 Cloud computing 、 Developers and chips behind them 、 Algorithm etc. .