At this time of the year, every year, people all across the world start brainstorming what unusual costume they would like to wear. From cats, nurses, to what is trending in pop culture, millions flock to the stores to buy the next best Halloween costume.
Beyond consumers, retailers can look to artificial intelligence and machine learning to create mockups of ideas that would be great for production. Not only could a retailer leverage these AI-generated costumes as unique products to sell, but brands can also take advantage of this to truly stand out during a time of the year where everyone wishes to flare.
Designing novel ideas for Halloween costumes
AI researcher, Janelle Shane, is one of several data scientists who have been using AI in fun, entertaining ways that might be quite useful for the retail industry. With the help of her computer, Shane built a neural network with the ability to create Halloween costume ideas.
Upon feeding her computer data on the 4,500 Halloween costume names scraped from the Internet, the AI algorithm figured out how to riff through those names and toss around new costume ideas. While the algorithm did spit out several items that were far from perfect, there were several Shane found to be quite intriguing, from the sad pumpkin king to the dragon of liberty.
Examples like this give a glimpse into the many weird applications AI can do for fashion designers and retailers at large. While the capabilities of AI are far from ever replacing human creativity, it can definitely serve as a good tool to come up with unique, novel ideas for a time of the year where people desire to dress up as something other than themselves.
Terrifying mask ideas for trick-or-treat
Much like the example above, another exciting way that retailers can consider using AI is mocking up design ideas for masks. One creative technologist named Matt Reed did precisely this, by training a neural network through feeding 5,000 images of popular Halloween masks.
What would have otherwise taken days or even weeks to conceive, the AI algorithms were able to design in no more than 24 hours, delivering a collection of creepy, yet unique mask ideas which retailers and designers could have a whole lot of fun.
To get the results seen above, Reed noted using general adversarial network, a type of AI algorithm. He explains the process as putting two networks against each other, which then allows them to improve over time continuously.
Creating the right Halloween mood
Another exciting way to heighten the Halloween mood is through music. Researchers Pinar Yanardag and a team worked on an AI that did exactly this during her time as a postdoctoral associate at the MIT Media Lab and Motherboard. They attempted to create scary songs.
By feeding the AI algorithms roughly 5 to 10 seconds of horror movie soundtracks, they generated some unique melodies. These efforts would not stop there, as the team would go on to collect user-generated songs to teach the AI to auto-generate much more authentic scary songs, befitting the Halloween season.
While there have been several cases of AI generating its own music, the primary goal of this project was to understand what exactly made music sound scary.
From mask designs to names for costumes, there is no end to how retailers can use artificial intelligence for thinking of more creative ways to promote their products during Halloween. There is a vast amount of opportunity to be made from integrating technology into the business.
Besides the examples listed above, AI-driven predictive analytics is another way for retailers to leverage big data to optimize their business. Brands can look at what items in their inventory are performing best and spend their money more accurately. This insight can also help fashion designers to learn which type of products are in demand, thereby delivering the exact items that customers want.