Can Amazon Use AI To Change The Fashion Industry?

According to the world clothing shoes and hats net,

It's not high-end.

Synonyms, but companies are preparing to use artificial intelligence algorithms instead of stylists.

。

Researchers at the e-commerce company are developing machine learning systems to discover the latest fashion trends and react more effectively, and even fashion fashions.

These studies will enable Amazon to track the trend of social media in real time, thereby improving its retail business.

In particular, it can help companies expand their advantages in the apparel industry and even dominate the field.

Kavita Bala, a professor at Cornell University, took part in the seminar on machine learning and fashion held by Amazon last week. He said: "companies like Amazon have been trying to understand the trend of fashion around the world.

It is changing the whole industry. "

Many forward-looking retailers have been using social networks like Instagram and Pinterest to track the latest trends and react quickly.

Startups such as subscription service Stitch Fix have personalised product recommendations based on user preferences and social media activities.

At the same time, Amazon is committed to developing its own clothing business, developing its own clothing brand, improving the quality of product display pictures, and launching Prime Wardrobe service, allowing users to try on clothes before deciding to buy them.

Amazon's Echo Look apps will even give you the right dress matching.

And that is far from enough. Amazon is still committed to developing more powerful functions.

For example, Amazon researchers in Israel have developed a machine learning algorithm that can analyze the labels attached to the images to infer whether the specific collocation can be regarded as fashion.

The software can be adjusted according to the fashion feedback or suggestions provided.

Previously, computers usually need a large number of tags to learn from visual information, but in many real-world situations, such as posting to Instagram images, there may be only one tag, and Amazon's machine learning algorithm has its own advantages.

Lab126, an Amazon Research Center in San Francisco, has developed an algorithm for learning specific fashion styles from images, and then can produce new products of similar styles. This algorithm is a simple AI fashion designer.

At present, this algorithm is not practical and implies a possibility.

Related research uses a frontier tool called GAN.

It consists of two deep neural networks, which can be effectively learned through raw data.

GAN determines the attributes of a specific style by analyzing multiple samples, and then applies this style to existing clothing products.

GAN, developed by Google Brain team researchers, is also a hot topic in machine learning today.

The two projects were unveiled at the seminar organized by Amazon.

This activity is mainly for academic researchers who explore fashion trends through machine learning.

The company declined to comment on these items.

The seminar also showed how to develop technologies to track fashion trends, so as to gain a deeper understanding of human behavior.

Bala and her colleagues are using the information collected from Instagram as the basis for anthropological research.

"We are trying to understand people's daily life," she said.

In human history, this is really unprecedented. We have so many visual records.

Others are exploring ideas that can directly explore people's wardrobe.

A research team from the Bana champagne campus of University of Illinois has demonstrated an algorithm for identifying fashionable social networking accounts.

A team from Myntra, a clothing website in India, has developed its own program to guess the correct size of a particular garment from the history of user purchase.

Tim Oates, a professor at Maryland Baltimore county university (Tim Oates), introduced the details of pferring a specific style from one garment to another.

He believes that this method can be used to design new clothing products.

Oates said, "you can train an algorithm through the existing clothes, then you can design a jacket or a pair of trousers for the machine, and I want to adapt to my style."

But fashion designers may not have to worry about it.

He pointed out that machine learning still has a long way to go in the field of fashion.

He said frankly, "people are also innovating in music, fashion and film.

But what we haven't seen is a music or fashion style really generated by machines that resonates with people.

More interesting reports, please pay attention to the world clothing shoes and hats net.